And so that's the main goal of this post: to tell the story of AI and energy on a global scale, to cast the net further out than any other analysis has done before. AI systems like Gemini and ChatGPT consume vast amounts of electricity, but building out the AI universe implicates energy networks and supply chains throughout the world. The rise of AI is certainly an electricity story, but it's also a story about energy, water, metals, materials, and thermodynamics. Steel and cement have to be produced to construct data centers. Fiber optic cables need to be laid down to relay data and information from data centers to businesses and consumers. Computer chips need to be manufactured with some of the most advanced machines human beings have ever designed. And these machines are so delicate and complex that they have to be disassembled when shipped and reassembled very carefully on-site. The Dutch semiconductor giant ASML ships hundreds of advanced lithography machines every single year. For a single Extreme Ultraviolet (EUV) machine, it takes ASML on average 3 cargo planes, 40 freight containers, and 20 trucks to ship the disassembled components to Taiwan Semiconductor Manufacturing Company (TSMC) in Taiwan, which uses these devices to produce the world’s most advanced chips.3 Taiwan’s electrical grid is dominated by coal and natural gas, and since TSMC alone consumes roughly 10% of the island’s electricity, the semiconductor industry in Taiwan is very carbon-intensive.4

These issues led us to begin asking about specifics and to conduct the necessary research to answer some major questions. How much energy is humanity using from the widespread adoption and diffusion of AI systems in the global economy? What share of global energy consumption in 2026 should be attributed to AI? What counts as an AI system in the first place? What amount of global greenhouse gas (GHG) emissions are produced from AI adoption and diffusion? How is the rise of AI affecting global energy networks and industrial supply chains? Where is all of this energy expansion going, and what effects could it have on the planetary biosphere over the following decade?

This is the first quantitative analysis we are aware of that attempts to unpack the global energy and emissions footprint of AI diffusion for a single year (2026), as well as to provide the evidence and context necessary for a comprehensive lifecycle analysis. Research studies on AI and energy are legion, but as we'll shortly see, they are usually restricted to various pieces of the larger puzzle without putting it all together. This piece will emphasize the lifecycle impacts of AI diffusion on a global scale. When researchers calculate the water footprint of beef, they don’t stop at the water consumed at the local McDonald’s. They look at the water used to raise the cow, the water used to grow the crops that feed the cow, and other parts of the supply chain, from which they conclude that 98% of the water footprint of meat products can be attributed to the water used for growing animal feed.5 They perform similar analyses if they want to calculate the emissions footprint of coal-fired plants or renewable energy technologies. The AI industry should be no different, yet too often public and academic discourse on the subject can feel like a haphazard caricature of what's actually happening.

That’s because AI is often seen as a product instead of an industry by many academic researchers and the general public. Products are spatially localized and isolated. The burger is right here in your hands. ChatGPT sends you a personal answer straight to your device. When all you see is isolated and disconnected units, the resulting consequences also look small and isolated almost by definition. But when you shift perspective and see AI as a global industry, you start to realize that it has interdependent supply chains, economies of scale, network effects, technological integration, and industrial agglomeration. Now it’s no longer an isolated thing. It’s a global megamachine devouring everything in its path. Precisely because AI is actually an industry with a broad range of products and services, we can subject it to the same kind of lifecycle analysis used for other industries.

The AI industry has an enormous environmental footprint that can be described through two basic components: operational and embodied. Operational emissions are the GHG emissions that result directly from the electricity consumed by AI systems for training, inferencing, or any other regular operations. By contrast, embodied emissions are all the lifecycle emissions produced from the manufacturing, transportation, construction, installation, recycling, and decommissioning of the physical hardware and infrastructure that support the broader cloud ecosystem, including activities like smelting steel, mixing concrete, cleaning silicon wafers, and so much more. Likewise, energy consumption can also be described in operational and embodied terms.

Tech companies and tech-friendly researchers have traditionally focused on operational carbon and energy consumption because they’re easy to measure and even easier to obscure through carbon offsets, credits, and other greenwashing mechanisms that do practically nothing to address the fundamental ecological issues resulting from the global AI rollout, including enormous water consumption and radical land use changes with far-reaching implications.6 As this analysis will show, the operational carbon and energy of AI electricity use are merely one significant piece of a much larger puzzle. The real impact is both upstream and downstream from on-site operations, from all the global supply chains that are furiously churning out the computing hardware and physical infrastructure that form the foundations of AI deployment to the rebound effects and downstream consequences on households, corporations, and governments that are using AI systems at unprecedented rates.

We use the term “shockwave” to emphasize the enormous levels of investment that are fueling the expansion and diffusion of AI technologies. We make no claims here as to whether the technologies themselves are impressive or revolutionary. Technologies can spread for all kinds of reasons, sometimes because they truly are revolutionary and other times because they are merely imposed on society by the ruling classes. There are features of AI models which are no doubt impressive and will change the nature of work and human society more broadly, but at the same time, these models also have fundamental weaknesses or potential issues that may limit their diffusion and universality going forward, including hallucinations, data walls, abstraction barriers, and many more.7 In any case, this piece doesn’t focus on the pros and cons of AI adoption. Its main focus is strictly on energy and emissions accounting, to give the public and researchers a better sense of the scale and scope of AI diffusion and the strains that it’s putting on the planet’s resources. The authors of this piece both believe that the financial speculation around the AI industry has become a bubble, and that, like all prior bubbles, this one too will end, and the scale of global investments will fall precipitously as a result. But when that might happen and what impact it could have on this assessment are also beyond the scope of this piece.

In 2026, our analysis finds that AI diffusion and deployment is responsible for as much as 1.1% of all global energy demand, substantially higher than what most naïve conventional estimates suggest. For example, a simplistic analysis from Our World in Data claims that AI diffusion accounts for only 0.2% of global primary energy demand.8 As we argue comprehensively in this piece, these kinds of assessments are generally less useful because they’re focused purely on electricity consumption and ignore the broader lifecycle energy dynamics powering AI diffusion globally. In reality, we show that AI-driven global energy use is likely already higher than the energy consumption of the entire United Kingdom. We also estimate that AI will be responsible for roughly 1.2% of all global GHG emissions for this year. These are stunning figures that undercut the typical narratives about the relative insignificance of AI on global energy consumption.

We do not claim that the results presented here are the final or definitive answers to the questions we have posed. In fact, we want to openly acknowledge right from the start that our methodology and calculations, presented in detail below, contain various flaws, omissions, and simplifying assumptions, almost all of which are necessary given the scope of this analysis and the corresponding data limitations. These issues will be discussed in detail, so we aim to spare our critics a great deal of wasted ink. The results in this analysis should be treated as a first-order approximation and as a gateway to further research and investigation, so that detailed second-order and third-order assessments can arrive at more precise answers in the future.

The amount of work that remains to be done in this area is quite vast. Our main objectives for this piece are as follows.

• We want to spark a shift in the way regular people, academic researchers, large corporations, and governments talk about AI and its impact on energy consumption, GHG emissions, and the planetary biosphere. That means first and foremost explaining how current narratives are woefully inadequate and subsequently offering better alternatives for how to think about the issues we’re facing.

• We hope to spur continued research in this area so that other groups may arrive at better answers than what we’ve presented here.

• We hope this piece can serve as a reference frame for future political action to assert greater public and democratic control over the scope, scale, and speed of the AI diffusion shockwave now gripping much of the world.

AI and Energy: An Introduction

The term “Artificial Intelligence” evokes many opinions and interpretations. The first starting point of this analysis is to determine what we mean by AI. In this context, we define AI as any generative, agentic, software, or robotic system partly or entirely based on classical symbolic logic (like TAMP robotic frameworks where symbolic logic planners intersect with computer vision, geometric pattern matching, etc), deep learning with neural networks, or some combination of the two. What’s excluded under this definition are models, systems, and processes based on things like logistic regression, random forests, or gradient boosting. What’s included in this definition are Large Language Models (LLMs) like ChatGPT, Claude, Copilot, and Gemini, but also AI agents, self-driving vehicles, industrial robots, autonomous robots, enterprise decisioning software, like clinical decision support systems in hospitals, and so much more. In plain English: if you depend on symbolic logic or neural networks in any way, you are classified as AI for the purposes of this analysis. Philosophical discussions about what constitutes intelligence and whether LLMs or AI agents are actually conscious or intellectual beings are beyond the scope of this analysis.

Most users interface with AI systems and usually ignore all the background conditions necessary for their creation. But before any AI model or system can be deployed, it has to be trained and developed. That requires storing vast quantities of data on the cloud. It requires deploying specialized equipment like graphics processing units (GPUs), accelerators, memory chips, server racks, network switches, optical transceivers, fiber optic cables, chillers, cooling towers, and so much more. To facilitate the demands of high-performance AI computing, thousands of GPUs are linked together to function as a de facto supercomputer, often supported by High Bandwidth Memory (HBM) chips that are vertically stacked right next to the GPUs. The GPUs receive inputs and instructions from the outside world and then use them to perform rapid calculations, in the form of matrix multiplications that implement the neural network architectures at the heart of modern AI systems (see Figure 2). All of these things have to be manufactured, transported, installed, repaired, and continuously operationalized to ensure they can properly support AI diffusion and deployment. And once AI models are trained and deployed, users send queries and prompts as inputs and receive various outputs, from written text and diagrams to images and videos.

Emma Strubell and her collaborators published a major paper in 2019 that brought to the world’s attention the energy-ravenous AI models under development at the time.9 For example, the paper concluded that training an early AI model with a neural architecture search methodology would emit the equivalent of 600,000 pounds of carbon dioxide, roughly what 30,000 cars in the US release in a single day under typical driving conditions. Although the paper was groundbreaking, it began the unfortunate trend of analyzing AI energy demand largely through the prism of operational electricity consumption, a trend that has persisted with many of the latest research studies, from which we get trivia-level facts like “training OpenAI’s GPT-4 took over $100 million and consumed 50 gigawatt-hours of energy, enough to power San Francisco for three days.”10 The push to analyze operational demands and capacity meant that most of the academic literature on the subject has largely ignored lifecycle dynamics, such as the upstream and downstream processes affected by AI deployment. Everyone has heard by now that a typical ChatGPT query can consume up to ten times more electricity than a regular Google search.11 But electricity consumption is only one angle of this story, and not even the most important one.

Udit Gupta and his collaborators challenged the focus on operational consumption in a groundbreaking 2021 paper which demonstrated that embodied emissions vastly exceed operational emissions for modern computing systems.12 There have been several studies examining the lifecycle links between AI models and energy and emissions. Sasha Luccioni and her team estimated the total lifecycle emissions of training the AI model BLOOM in a 2022 paper.13 Confirming the results from Gupta and others, the researchers showed that the embodied carbon of the computing hardware swelled beyond the operational emissions from training the model itself.

In a famous 2023 paper, Alex de Vries showed that the inferencing requirements of AI would soon dwarf the electricity consumption used for training.14 The paper modeled several future scenarios of electricity consumption, with a worst-case scenario in the near term showing Google’s electricity consumption ballooning to the same level as that of Ireland. However, the analysis only focused on electricity consumption and ignored the network effects and lifecycle dynamics of AI deployment. Other researchers have shown that AI systems can have radically different emissions footprints depending on where, when, and how they’re executed, especially when they lack sufficient battery storage and have variabilities from wind and solar power.15 For example, certain AI workloads could be transmitted to data centers powered by renewables, thus lowering their emissions.

Data centers and other AI infrastructure are fundamentally built on critical minerals and raw materials extracted from the Earth. Copper is used for wiring, lithium is used for batteries, and rare earth elements go into industrial magnets that are used for cooling fans and hard drives. All these critical minerals and elements are sourced from a variety of entangled and extractive global supply chains.16 The mining operations required to obtain these raw materials have a significant emissions footprint, as they end up destroying local carbon sinks and landscapes, not to mention the operational emissions from running gigantic diesel-powered vehicles and machines, like rotary blast drills. Processing a single ton of rare earth metals produces roughly 2,000 tons of toxic waste.17 In Baotou, China, a massive lake of toxic sludge emerged over the years from all the rare earths processing, ruining local water quality and devastating local communities.

The semiconductor industry is the beating heart of the data center buildout. Semiconductors are made in fabrication plants, or fabs. Many companies that are considered semiconductor companies don’t actually build their own chips or devices. Instead, they outsource production to third-party manufacturers. The most dominant semiconductor manufacturer in the world is Taiwan Semiconductor Manufacturing Company (TSMC). TSMC operates on a foundry business model, meaning they don’t design anything; they only build the physical devices and components. Other companies, like NVIDIA or AMD, are software companies that do the design work and then send the designs to TSMC so it can produce the corresponding chips. In a literal sense, TSMC is a chipmaker and a company like NVIDIA is a chip designer.

And these companies are barely scratching the surface of the semiconductor industry. Microchips are made through a very complex process involving steps like etching, photolithography, deposition, testing, and assembly, all of which require some of the most advanced manufacturing machines and devices human beings have ever made. The Dutch company ASML manufactures the world’s most advanced lithography machines, which use advanced sensors and lasers to etch complex patterns on silicon wafers. Applied Materials produces some of the world’s most advanced deposition systems. These AI chips are further supported by a physical ecosystem of server racks, network switches, optical transceiver, cooling systems, and fiber optic cables (see Figure 3). The server racks themselves require large amounts of steel and aluminum, both of which are built from energy-intensive industries that emit vast quantities of carbon dioxide.

In 2023, US data centers comprised about 4.4% of national electricity consumption, or 180 TWh, roughly as much as all of Thailand and about four times as much as New York City.18 Back in 2000, they only consumed 0.1% of total electricity consumption, so their share of national electricity consumption has risen by 4000% in under 25 years.19 In other countries, the data center share of national electricity consumption is even higher; they were responsible for over 10% of Ireland’s total power output in 2021, and are now likely hovering around 20% of total Irish power consumption.20

Let’s take a moment to explain our approach to units in this analysis. The most common unit we use is the terawatt-hour (TWh), a unit of energy. For perspective, 1 TWh of energy can power roughly 100,000 average American households for a year. We’ll also make reference to other units depending on what’s used in the original sources we cite, such as megajoules (MJ) or exajoules (EJ), which are both units of energy. We’ll also use units like the gigawatt (GW), a unit of power. Power is just energy per unit of time. For context, 1 GW of power capacity can cover the consumption of roughly 750,000 US households, though it can go up to almost 1 million in certain circumstances.21 Because we use so many different kinds of units, we’ll be converting back and forth many times among these units. The goal of this analysis is to keep the math and the equations to a minimum, so we won’t really show all the conversion steps involved.

Data center efficiency has increased substantially over time. The Power Usage Effectiveness (PUE) tracks the total energy consumed by a data center divided by the energy consumed by its computing equipment. It’s always going to be above 1 by definition, and the closer it is to 1, the more efficient the data center is considered to be. About 15 years ago, data center PUE stood at an average of around 2.22 Today, leading hyperscale sites often achieve 1.1-1.2, while broader fleet averages are closer to 1.4-1.6, thus representing a major gain in average efficiency.23 This efficiency boost was part of a broader trend in computing captured by Koomey’s Law, the observation that the number of computations per joule of energy dissipated has doubled roughly every two years.

Despite this incredible rise in efficiency, total electricity consumption from data centers has skyrocketed, as we explained above. It’s a canonical example of the Jevons Paradox at play. Named after British economist William Stanley Jevons, who observed in the 19th century that aggregate coal consumption kept rising in Britain even as steam engines were getting more efficient, the paradox states that the more efficient industrial and technological systems become under capitalism, the more energy they consume as production costs decline, products become cheaper or cost-effective, and demand rises as a result. The empirical evidence for it is overwhelming.24 Beyond just data centers, one can find numerous examples of it in modern history. Steel production ballooned after the Bessemer process made it more efficient to make steel.25 Fertilizer production skyrocketed after the Haber-Bosch process made it vastly more efficient to produce ammonia.26 Water consumption in global agriculture went through the roof in the late 20th century with the introduction of more efficient irrigation methods.27 The list goes on and on. As explained in The Physics of Capitalism, and in this post, technological change under capitalism is often channeled towards expanding exergy, the maximum useful work that can be extracted from a system as it reaches equilibrium with its surroundings. In other words, capitalists are routinely looking to expand the scale of the energy system by building more energy-intensive devices, then they exploit any resulting efficiency gains to drive more production and consumption, as the whole system is structurally oriented towards biophysical growth.

It’s easy to see this phenomenon at the device level as well. A typical Hopper AI chip from NVIDIA consumes roughly 700W.28 The next-generation Blackwell chip is far more efficient, in the sense that it can perform more calculations for every unit of energy consumed, but it also operates at about 1200W.29 Newer chips like Rubin will be even more efficient, but that will still likely lead to greater adoption and diffusion, and therefore energy consumption. AI chips might be getting more efficient, but they’re also getting more energy-intensive at the unit level.30 This is what happened with steam engines during the Industrial Revolution as well (see the post above and also this post for more details). Early steam engines could generate far less mechanical energy than later versions, so not only did they get more efficient, they also got bigger, more complex, and more energy-intensive.

Complex systems can expand in scale because they become more efficient, but it can also be the case that exogenous shocks to their energy scale can force or incentivize system components into becoming more efficient. These are some of the same dynamics driving the AI infrastructure buildout worldwide: the push for better is coinciding with a push for bigger, including bigger chips, bigger robots, hyperscale data centers, and so on. This infrastructure frenzy is also partly driven by the so-called AI scaling laws, which have convinced much of the tech world that throwing more data and more computing resources at neural networks will significantly improve their performance over time.31 One of the great unknowns for the future of the AI infrastructure buildout is in what direction the tech industry will travel if these scaling laws start hitting insurmountable barriers and limits. One of the most constrained biophysical inputs for the AI industry is water, but it’s also one of the most misunderstood topics in the broader AI energy story, so we need to develop a clear understanding of how it fits with everything else.

AI and the Heat-Water Nexus

This section is about water and heat and how they’re related to energy production for AI. But although water and electricity get the bulk of the attention, heat is perhaps the most important aspect of the AI energy story. Many data centers get their electricity from power plants burning fossil fuels, like coal and natural gas. It turns out, however, that the electricity produced by coal and gas plants is less than 40% of the total energy unleashed from burning fossil fuels. The rest of the energy is waste heat dumped into the external environment. The laws of thermodynamics imply that no real-world heat engine can convert heat to mechanical work with 100% efficiency, so some waste heat must always be rejected to a cooler external environment.

When fossil fuels are burned in a power plant, they convert water into hot and pressurized steam, which is then used to drive a turbine that produces electricity. The steam then needs to be cooled down and converted back to a liquid, so it can be recycled and reused for the next cycle starting in the boiler. But that heat carried by the steam has to be removed from the plant somehow, and the cooling loop performing that task is responsible for the vast majority of water consumption at power plants. Most power plants in the United States now deploy water-intensive recirculating systems that evaporate much of the water they withdraw from local watersheds straight into the atmosphere.32 The United States Geological Survey estimated in 2019 that roughly 57% of water withdrawals from recirculating systems are evaporated and consumed.33

Computing equipment in data centers can get really hot because of Joule heating (or resistive heating), which is the heat that’s generated when an electric current passes through a conductor. Since transistors are basically just electrical on-off switches, they collectively generate vast quantities of heat and run the risk of causing major malfunctions. As a result, this heat needs to be removed from the servers inside the data center, and many methods have been developed to do precisely that. For most of their history, data centers relied on air-powered cooling systems, either giant fans blowing air across the server hall or industrial-grade AC units blowing in cool air through large vents. But with the rise of cloud computing and generative AI, more and more data centers have shifted towards liquid-powered cooling systems, which are more efficient at removing heat than their air-powered counterparts.34

The most popular methods of heat rejection among liquid-powered systems are evaporative cooling and adiabatic cooling. They both exploit a crucial fact about the phase change of water: when water transitions from a liquid state to a gas state, it absorbs energy from the surroundings. It’s the same basic principle behind sweating on a hot summer day: as the sweat evaporates from your skin, it absorbs heat and helps keep your body cool. In evaporative cooling, the water collects the heat from the servers inside the data center and is then sent to the cooling towers outside the data center. Some of this hot water is then sprayed and exposed to the air, thus evaporating and absorbing heat from its immediate surroundings. The cooler water is then relayed back into the data center to start another cooling cycle. On particularly hot days with excess moisture, many data centers will resort to mechanical refrigeration with chillers instead of evaporative “free cooling.”

Evaporative cooling requires large quantities of operational water consumption, and so many data centers rely on adiabatic cooling to conserve water for their on-site operations. In adiabatic cooling systems, air is the primary method of heat rejection, not water. But water may be used to cool down the air on very hot summer days, when the incoming air is obviously insufficient to actually cool down the cooling fluid. What happens at a mechanical level is that the air is pushed through water pads that spray the hot air. As the sprayed water evaporates, it absorbs heat and cools down the air, which is finally blown across the sealed coils containing the cooling fluid. The water and the air never touch the sealed cooling fluid in this method. In thermodynamics, an adiabatic process is one where there is no net heat exchange between a system and its surroundings, and so this cooling process is called adiabatic because the total enthalpy (heat content) of the air-water mixture remains constant. The heat is simply shifted from one source to another.

Adiabatic cooling methods are becoming increasingly popular because their closed-loop setup allows hyperscalers to claim that they’re taking water conservation seriously, but the problem is that the fans doing much of the cooling need electricity to keep running, and research has shown that the indirect water footprint associated with that electricity generation can be far bigger than the on-site water consumption itself.35 Many adiabatic and dry cycle cooling methods are therefore unlikely to resolve the fundamental water consumption and conservations issues related to the AI infrastructure boom.

Training AI models requires large quantities of water. Researchers have estimated that training OpenAI’s GPT-3 in Microsoft data centers directly consumed roughly 185,000 gallons of water, and that figure balloons to 1.4 million gallons of water once the total water footprint is included, equivalent to the daily water consumption of roughly 14,000 Americans.36 But as AI scales globally across different industries, it’s inferencing for regular tasks, queries, and prompts that will comprise the greatest share of its global water footprint.

The direct operational water footprint of a typical data center is nothing astonishing on most days, but it can rise dramatically during hot summer days, straining local water resources severely. A major problem with evaporative cooling methods is that they simply expel and evaporate vast quantities of water, permanently removing a precious resource from local watersheds.37 Plenty of other industries directly consume more water than data centers, but many of these industries, like agriculture, return much of the water they initially consumed back to local watersheds. Data centers are a major reason why water utility bills are rising across much of the United States.38 With looming political and resource problems down the road, some hyperscalers have started building water reclamation facilities to reuse wastewater for cooling data centers instead of discharging it to local rivers and other habitats. In Douglas County, Georgia, Google’s data center campus gets over 95% of its cooling water from non-potable water treated at water reclamation facilities that were financed by Google itself.39 But for all the real problems with their direct operational water footprint, by far the biggest problems lie upstream, with the indirect and lifecycle demands that are necessary to keep data centers running in the first place.

In 2021, researchers estimated that roughly 75% of a data center’s water footprint was indirect, meaning that most of the water consumption attributed to data centers went towards generating the electricity required to power the data center.40 The vast majority of the electricity came from coal, gas, and nuclear plants. However, once chip and device manufacturing are taken into account, that share rises to well over 90%. In other words, the vast majority of the water consumption required to create and maintain a data center is either indirect or embodied. It’s absolutely fair to say that the water used to cool down a typical data center in the first place is mostly trivial in the grand scheme of things. But once the lifecycle consumption analysis comes into the picture, the total water footprint of a typical data center becomes enormous.

For perspective, the total water footprint of data centers in the United States had already surpassed the entire global water footprint of Coca-Cola in the mid-2010s.41 As of this publication, in the summer of 2026, it’s almost certainly the case that US data centers have a vastly bigger water footprint than all US golf courses, which used to outpace them earlier in the decade (see more details below). Given the phenomenal growth rates projected over the next few years, data centers are poised to become one of the dominant water-guzzling industries on the entire planet, potentially even rivaling the entire global agricultural system within a decade unless the AI bubble pops and infrastructure development screeches to a halt.

A few months ago, Anthropic CEO Dario Amodei published an essay called “The Adolescence of Technology.”42 The core theme of his essay—that AI diffusion carries many great risks but that we can keep those risks under control—is not the main focus of this piece. In the essay, Amodei indirectly dismisses ecological and energy-related concerns about AI by suggesting that data center water usage isn’t an actual problem at all; while he doesn’t overtly argue this position himself, he does link to a quantitative analysis by the tech research firm SemiAnalysis.43 In that analysis, the research firm compared the water consumption of Elon Musk’s Colossus 2 in Memphis to the lifecycle water consumption of an In-N-Out Burger joint, concluding that Colossus 2 only consumes as much water as about 2.5 In-N-Out Burger places.

The first facility, Colossus 1, initially consumed about 1 million gallons of water a day.44 The planned capacity might take it up to 5 million gallons of water a day, equivalent to the daily water usage of roughly 10,000 people in the US, or about 3,500 households.45 The facility uses a Supermicro closed loop liquid cooling system for internal servers, but then passes off the heat to an exterior loop that uses evaporative cooling, so much of the water it’s drawing from local Memphis sources doesn’t actually go back to local watersheds.46 However, Colossus 2 is largely powered by aeroderivative gas turbines, which typically don’t consume any water while they’re running.

This comparison from SemiAnalysis is severely flawed on multiple levels. First, the vast majority of data centers in the United States get their power from the regular utility grid, not from aeroderivative gas turbines or diesel generators.47 That means their water footprint is largely indirect. Using Colossus 2 as an example of water efficiency in data centers is therefore a complete red herring, since it’s not actually representative of how most data centers operate. Second, SemiAnalysis was happy to conduct a lifecycle water consumption assessment on the beef used for In-n-Out burgers, but they strangely ignored their own methodology on the other side of the ledger, preferring to look at the operational water consumption for the gas turbines, which conveniently for them happens to be zero. But it obviously takes vast amounts of water to build the gas turbines in the first place. It takes water to extract the raw materials and lots of water is also used in the manufacturing process, among other points in the supply chain.

A typical natural gas generator at Colossus weighs over 200,000 pounds, or roughly 91 metric tons, and there are 46 of them currently operational.48 The dominant structural components by mass are steel and iron, though many other metals, like copper and nickel, are also used in most natural gas generators. Based on manufacturing water inventory data from 2005, we’ll assume a conservative 6 cubic meters of water consumption per metric ton.49 A cubic meter is about 264 gallons, so that translates to 1,585 gallons of consumption per metric ton. The embodied water footprint of a natural gas generator at Colossus is therefore about 144,000 gallons, and the total for all 46 generators comes out to roughly 6.6 million gallons of water. The typical American household consumes 300 gallons of water a day, which on a per capita basis comes out to about 100 gallons a day.50 That means the embodied water footprint of the Colossus gas turbines alone is equivalent to the daily water consumption of almost 70,000 Americans. This analysis, of course, completely leaves out the embodied water footprint of the steel and concrete used in the data center itself, not to mention the enormous embodied water footprint of the computing hardware. Once all of that is factored in, the SemiAnalysis comparisons would look downright ridiculous.

The misconceptions around the true magnitude of data center water consumption appear constantly in popular news sources as well. A recent analysis from CBS News purported to show that many common American economic sectors, like golf courses, consumed more water than American data centers, about 230 billion gallons for the latter compared to 530 billion gallons for the former.51 There were no original insights in this analysis. It was largely a summary of prior research with good-looking charts. The CBS analysis relied on numbers from 2023, which are understandably outdated today. For an industry where the emissions and water footprints are practically doubling every two years, what happened three years ago is ancient history.

The analysis also assumed that the only component of the indirect water footprint is water consumed by power plants to produce the electricity for data centers. In reality, the indirect water footprint is much larger, as it includes water consumed by the semiconductor industry to produce chips and other computing hardware, water consumed by the steel industry to produce the steel for server racks and other infrastructure, water consumed to produce aluminum and other critical metals, and so on. Given some of the projections starting in 2023 contained in the CBS analysis itself, it’s highly likely that the total data center water footprint in the United States will vastly exceed that of golf courses in 2026. Of course, agriculture and some other sectors still consume significantly more water than data centers, as we mentioned earlier. But these sectors are not approaching anything near the astonishing growth rate of data centers, and therefore they don’t have the same level of marginal impact on global ecology. Furthermore, sectors like agriculture and golf courses reclaim large quantities of water. According to the data contained in the CBS report, the 530 billion gallons of annual water consumption attributed to golf courses drops closer to 400 billion gallons once the reclaimed water is included in the analysis. At that level, there is absolutely no contest in 2026 between golf courses and data centers; the latter win easily once their full embodied footprint is included.

This vast water demand puts enormous pressure on local public water utilities, especially since data centers are often geographically concentrated and the local utilities were not built to meet the insane water demands of modern computing networks. For a specific example, consider Loudoun County in Northern Virginia, the global leader in the data center buildout. Loudoun Water is the public utility that serves over 300,000 residents in the county. Researchers have estimated that Loudoun Water would be unable to meet total customer demand on some of the hottest summer days, when data centers need to use more water to keep their systems cool and local facilities may therefore be tempted to turn on evaporative cooling.52

Meeting this growing water demand requires the construction of new pipelines, plants, and infrastructure, which are costs that generally fall on local communities, often through higher water bills. It doesn’t help that many data center companies and hyperscalers often require local governments to sign non-disclosure agreements regarding future energy demands and requirements.53 Many critical details about on-site water usage are never reported, making it difficult for public water utilities to understand just how much water they need to provide and provision. As a result of all this uncertainty, many towns and cities are now fiercely resisting the expansion of data centers into their local communities. Data centers have recently been rejected everywhere from Minnesota to Arizona, with an estimated $130 billion worth of data centers blocked by local communities in Q1 2026.54

One of the world’s major water guzzlers is the semiconductor industry, which consumes vast amounts of water through its fabrication plants. That’s because the silicon wafers that form the basis of all microchips need to be kept extremely clean at all times. Virtually any impurities, even marginal amounts, would sabotage their desired electrical properties. As a result, they’re routinely washed by water at various stages of production. The ultrapure water (UPW) used in modern fabs is produced from extensive filtration systems in municipal water plants. A large fab can gobble up millions of gallons of water in a single day, and almost 4 billions gallons a year.55 When Taiwan experienced a historic drought in 2021, the government had to save the semiconductor industry by practically denying water to nearly all other economic sectors.56 The global semiconductor industry consumes over 260 billion gallons of water a year, at least as of 2022.57 Recent numbers are likely much higher. That’s equivalent to the total water consumption of Hong Kong, an alpha city of 8 million people.58 The indirect water footprint of the AI infrastructure boom is thus entangled across multiple global industries and supply chains.

Methodological Assumptions

The main goal in the rest of the piece is to calculate global energy consumption and GHG emissions specifically caused by AI diffusion and adoption, but first we need to articulate the major assumptions and theoretical frameworks guiding our analysis. There are no magical standards for measuring the energy and emissions footprint of a country or industry, and different organizations have used different standards over the years. Should we adopt partial substitution or the physical content method when measuring primary energy? How should we distribute consumption for geographically dispersed industries? If an airplane takes off in Paris and lands in New York, should the emissions from the airplane be assigned to France, the United States, both, or neither? What are the best ways of measuring emissions from land use changes? If you clear high-powered carbon sinks like peatlands and wetlands to build cities, palm plantations, or data centers, how should we measure the resulting emissions? What are the starting and final points of a lifecycle analysis? What stages in between should be included in the analysis? Where exactly do you draw the boundaries? You will get different answers depending on how you answer that question. These are just some among thousands of other important questions one could ask.

In this section, we are going to explain the major theoretical assumptions and practical choices guiding our calculations and how we arrived at our final figures, all of which are shown explicitly in the following section. Our decisions can certainly be questioned, precisely because in some cases there are different valid approaches and because in other cases we have broken new methodological ground in how we assess issues that people have been debating for decades. The first major assumption is that, with noted exceptions, we’re measuring the primary energy associated with AI, even though it’s definitely not a perfect metric. Nevertheless, we’re making this choice to be consistent with the way energy accounting is done by major international organizations, such as the International Energy Agency (IEA) and The Energy Institute. That means our AI-specific figures can be expressed as a share of the global total, thus allowing for additional context and comparisons.

Primary energy represents the direct use of energy sources without any prior conversions or transformations.59 Primary consumption includes activities like burning coal at a power plant and distilling crude oil at a refinery. Primary forms of energy are not useful on their own, so they are converted and transformed into secondary forms of energy. For example, we burn coal so we can turn the resulting steam energy into electricity, and we distill crude oil so we can produce gasoline. Coal and crude oil are primary forms of energy while electricity and gasoline are considered secondary forms. The secondary sources can also be converted into other tasks and end uses, collectively known as tertiary sources.

One of the chief theoretical struggles in primary energy accounting has been how to measure the energy consumption of renewable energy systems, like solar panels and wind turbines. There are two common methods of measuring primary energy: the partial-substitution method and the physical-energy content method.60 Let’s analyze them through some examples. When a power plant burns through coal, the primary energy simply equals the energy of the coal that goes up in flames. In the case of fossil fuels, then, things are pretty easy: record the amount of stuff we burn and call that primary energy. But the situation is more complicated for renewable energy sources, such as wind, solar, and hydropower, because nothing was burning while these energy sources generated electricity. Enter the two methods above.

In the physical-energy content method, we simply count the electrical energy produced by these sources as primary energy, even though electricity obviously qualifies as a converted form of energy. This is the method used by the International Energy Agency to measure energy consumption for renewables. In the partial-substitution method, we pretend that the produced electricity came from a hypothetical thermal power plant, and then we assume some efficiency rating for this plant. The idea behind the method is that the renewable energy sources are replacing a coal or gas plant that would’ve produced that same amount of electricity. For example, if the plant has a hypothetical efficiency of 20%, then we would multiply the electricity generated by the renewables with a factor of five. In this case, the primary energy required to produce that electricity is five times larger. The company British Petroleum historically adopted this partial-substitution method in its popular global energy annual reports, but after it handed over control over generating these reports to The Energy Institute, a British organization, the latter made the decision to switch over to physical content method as well, aligning with the UN and the IEA.61 The main reason why these differences matter is because they can lead to diverging estimates of energy consumption, especially for nations that rely heavily on renewables. For example, Norway’s energy consumption is roughly 60% higher when measured through the partial-substitution method, given its significant dependence on hydropower.62

Although we’re aiming to measure primary energy to be consistent with other organizations, there are major conceptual issues with this method of accounting, as explained by Erald Kolasi in The Physics of Capitalism.63 In that work, and in this Substack post, Kolasi explains how economic systems function like conversional networks (coronets for short), meaning they’re continuously converting energy into different forms through highly organized and stratified networks. This network view fundamentally destabilizes any rigid starting points and boundaries for energy analysis. It’s hard to say that burning the coal at the power plant is what’s really primary and fundamental when that coal had to be first mined, extracted, partially refined, and transported on-site before it could be used for anything. And that whole process of extraction and transportation depended on a whole array of other energy conversions and labor inputs, which were also embedded in a series of other conversions, and so on. Viewing economic systems as coronets highlights the interdependence of global supply chains and the importance of butterfly effects and non-linear dynamics in understanding their underlying energy systems. For that reason, we may adjust certain calculations over and above primary energy, and for certain sectors we may focus on downstream energy consumption while avoiding a full primary energy accounting. All of these situations are handled on a case-by-case basis as shown in the next section.

For the energy analysis, we adopt an aggregate hybrid lifecycle approach, meaning we conduct an input-output analysis across the various supply and industrial chains necessary to make AI operations a reality. Specifically, we assume that the final output of the process is the total energy consumed by data centers and other facilities in 2026, adjusted for the impact that can be specifically attributed to AI. Our goal is to estimate the total lifecycle energy consumption and GHG emissions from AI adoption and diffusion in 2026. Normally, one might measure or estimate energy consumption related to data centers and other facilities that happened in this year only. That would mean counting the electricity consumed by data centers in 2026, the steel and cement produced in 2026 for data centers, the water consumed in 2026 by the semiconductor industry while producing chips for data centers, the fiber optic cables laid down in 2026, and so on. Then one would attribute a portion of all this energy use to AI specifically, since not all data center capacity is geared towards AI.

But this method presents enormous challenges. One needs to determine the stage of completion for various data centers, power plants, and other facilities around the world, and to then disentangle the supply chains that are providing them with materials and resources. Obtaining the data for this kind of analysis is quite difficult, and so we therefore opted for a different approach on a case-by-case basis. For certain sectors, we’ll be able to estimate the 2026 footprint directly, like the primary energy required for data electricity consumption. For other sectors, we’ll infer their 2026 footprint through indirect means, usually by first calculating their lifecycle footprint and allocating a share of that overall total to this particular year. Industry analysts expect roughly 100 GW of new data center capacity to come online between now and 2030.64 JLL has estimated that over 35 GW of new data center capacity are under construction in North America alone as of 2026.65 These figures imply that at least 65 GW of data center capacity are under various stages of construction globally, over half the operational total in 2025. It’s clear from these numbers that the AI infrastructure boom is backloaded towards recent periods, which implies that a huge share of any sector-level lifecycle total is concentrated in 2025 and 2026.

And that’s only a small part of the story. The bigger part is that the latest data center components have far higher energy intensities than prior ones, considering the shift over the past few years towards liquid-powered cooling, high-performance microchips like GPUs, and other energy-intensive devices. For these reasons and the ones above, we’ll allocate somewhere between 50-60% of any total sector-level lifecycle footprint towards 2026 alone. This is the basic assumption that we’ll carry going forward in our analysis.

To estimate the GHG emissions associated with global AI diffusion, we decided to largely adopt the standards of the GHG Protocol, a widely used methodology.66 That means we aim to track Scope 1, Scope 2, and Scope 3 emissions, with certain caveats and adjustments that are explained along the way. The goal is to have a comprehensive lifecycle analysis of the emissions impact from all the global energy networks and industrial supply chains involved in creating and building the physical infrastructure necessary for AI deployment and diffusion.

Following the GHG Protocol ensures we don’t get lost in the weeds about carbon offsets or other greenwashing gimmicks from Big Tech. We count operational and lifecycle emissions and that’s it. Scope 1 emissions are direct emissions linked to core business or household activities. Think of the energy used by a manufacturing business in its factories to produce goods. Scope 2 emissions are indirect emissions associated with the energy generated to keep a company or household operational, like emissions required to generate electricity that companies and families use to keep the lights on. Scope 3 emissions are upstream and downstream emissions across the lifecycle chain. Think of the emissions caused by the employees of a large company as they drive to work. Those emissions are assigned to that company under the GHG Protocol, since the workplace schedules and requirements of the company are the fundamental reason why the employees are driving in the first place.

Our analysis breaks new ground in many areas that haven't received much attention, from land use change emissions caused by data center construction to downstream rebound effects from AI adoption. AI doesn't just live in data centers. It affects logistics, manufacturing, retail, banking, and so many other economic sectors around the world. For the first time ever, we're going to quantify the impact on many of these sectors on a global scale, while reiterating that we are by necessity making reasonable estimates and that these calculations must be understood accordingly.

Quantitative Analysis

In the sections below, we calculate the various AI-driven energy and emissions footprints of different industries and sectors, then we add up everything at the end to arrive at our global totals.

Primary Energy of Data Center Electricity Consumption

The International Energy Agency offers a wide array of assessments on what share of total data center capacity is devoted to AI operations and acknowledges the inherent uncertainty of determining this figure given the paucity of the available data.67 A widely quoted source is the Lawrence Berkeley National Laboratory’s flagship 2024 report on electricity demand from data centers, which suggested that roughly one-third of total data center capacity in the United States is devoted to AI, although with large uncertainties around the estimates.68 But regardless of the source, what nearly all of them have in common is that they track the amount of electricity consumed by GPU-accelerated servers, divide that by the total amount of electricity consumed by the data centers (or scale it with a facility-wide PUE, effectively the same thing), and call that the AI share of consumption. But this approach is very misleading for several reasons.

First, preparing the training data for AI model development, everything from text extraction and deduplication to labeling and tokenization, usually relies on CPU-powered servers and computers.69 Second, AI computing in modern data centers is widely distributed across various nodes and servers, requiring complex network switches to relay data across various servers. And moving all these signals also requires optical transceivers, which themselves consume lots of electricity. Third, scaling GPU loads with a standardized PUE ignores the non-linear dynamics of AI computing demands, which can exhibit extraordinary volatility. There are additional concerns as well, but they all lead to the same conclusion: the AI share of data center electricity consumption is much larger than what appears in traditional sources. This is especially true once we consider what’s coming down the pipeline in the next few years, when the majority of the growth in data center power demand is expected to be driven by AI workloads.70

Given the considerations above, we assume that 40% of global data center capacity is being devoted to AI. We acknowledge the inherent uncertainty around this assumption, and we think a great case can be made that the AI share should be even higher. In any case, it’s not a figure that can be definitively proven with the currently available data, in large part because hyperscalers and many other data center companies deliberately withhold and obscure the necessary information that would allow researchers to more accurately determine these distributions.

The average carbon intensity of the global electric grid is 460 grams of CO2-eq per kWh.71 Given the 788 TWh of global data center electricity use estimated below, the direct emissions from the data center sector would account for about 362 million metric tons of CO2 emissions in 2026, and the AI sector specifically comes out to roughly 145 million tons of CO2 emissions (40% of the data center total), an enormous sum equivalent to the annual emissions of roughly 30 million gas-powered cars.

There were roughly 12,000 operational data centers worldwide at the end of 2025, with a total global peak power capacity of roughly 122 GW as of Q1 2025 (equivalent to 122,000 MW), but we’ll assume 125 GW of capacity to account for the passage of time since Q1 of last year.72 For a narrow metric counting only commercial data center facilities, total data center energy consumption worldwide in 2025 has been estimated at roughly 470 TWh, though it’s rising quickly and, given current trajectories, will certainly be higher in 2026.73 A broader metric from S&P Global that includes micro data centers, enterprise server rooms, modular data centers, crypto facilities, and also accounts for the major inflection points in 2025, including the avalanche of AI-accelerated servers coming online, puts total global data center electricity consumption at 788 TWh in 2025.74 It’s this 788 TWh figure that will serve as our baseline starting point for global data center electricity consumption in 2026, simply because it's more accurate and comprehensive than the data from the IEA, which is largely focused on consumption from hyperscale and colocation data centers. Given our assumption that 40% of global data center capacity will go towards AI workloads in 2026, the estimated AI-driven component of global data center consumption is 315 TWh (40% of 788 TWh). Considering that overall global electricity consumption stands at roughly 32,000 TWh, our estimate puts the AI data center share alone at 1% of all global consumption.

The crypto industry itself has been estimated to annually consume roughly 140 TWh in electricity for Bitcoin mining.75 However, an increasing portion of this consumption is now covering AI workloads, as the crypto industry is making a concerted effort to shift additional capacity towards AI.76 The pinnacle of this structural shift was the $10 billion deal that Bitcoin miner IREN signed with Microsoft last year, agreeing to provide NVIDIA GPUs for high-performance AI computing needs.77 The sharp distinction that once existed between crypto mining facilities and commercial data centers is therefore rapidly evaporating, and for the first time ever it makes sense to lump them into the same category of electricity consumption. The estimated AI component of electricity consumption shown above thus implicitly includes the crypto sector as well.

The 315 TWh figure above is quite something, but it’s not the final answer. That’s because it takes upstream energy to produce electricity, and it’s this primary energy that we really care about. Since a major share of data center electricity is coming from coal and gas power plants as well as nuclear power plants, we need to determine the primary energy associated with that consumption. Based on global grid averages, 60% of global electricity was produced from fossil fuels, 31% from renewables, and 9% from nuclear power.78 We’re going to assume that this distribution holds for data center electricity use as well.

Now we can calculate the first major numbers we’ve been chasing all along. Of the 788 TWh globally, the mix above means that 473 TWh came from fossil fuels, 244 TWh came from renewables, and 71 TWh came from nuclear power. Based on prior research from the IEA, we’re going to assume a traditional efficiency of 36% for the fossil fuel plants and 33% for the nuclear plants.79 That implies the fossil fuel component of data center electricity consumption has a primary energy footprint of 1,314 TWh, the renewable component stays the same at 244 TWh (since we’re the using physical content method), and the nuclear component comes out to 215 TWh. In total, the global primary energy footprint of data center electricity use is 1,773 TWh. Allocating 40% of that total sum to AI means that AI-driven primary energy consumption for data center electricity use is 709 TWh globally. This is our first major concrete figure, and it’s already at 0.42% of total global energy consumption, since the latter stood at roughly 166,670 TWh in 2025, and we estimate will reach something like 168,000 TWh in 2026 to account for expected growth.80 Although this figure is a major component of the total quantity, we still have a long way to go.

Steel and Concrete

The global steel industry is responsible for roughly 7-9% of global GHG emissions, perhaps slightly more.81 Hyperscale data centers often require 10,000 metric tons of steel or more.82 However, the majority of the world’s data centers are not gargantuan facilities, so we assume that the average data center embodies roughly 2,000 metric tons of steel. According to the World Steel Association, producing every ton of steel leads to the emission of 2.18 metric tons of CO2-eq.83 And each ton of steel required 21.27 GJ of energy on average to produce.84 Using these numbers, we estimate total data center lifecycle emissions from steel at roughly 52 million metric tons and the total energy required for all that steel at 142 TWh. Allocating 40% to AI yields 57 TWh of energy and 21 million metric tons of emissions. For the 2026 footprint, we assume 60% of the lifecycle total, thus yielding 34 TWh and 13 million metric tons of emissions, respectively.

Data centers contain enormous amounts of concrete, which is a mixture of sand, gravel, water, and cement. The vast majority of concrete’s emissions footprint comes from the cement component, and so that’s what we’ll focus on. Based on industry reports in the United States, the typical American data center contains roughly 1,700 tons of embodied cement.85 But considering that the typical American data center is larger than the typical data center in the rest of the world, we assume that the global average of embodied cement per data center is roughly 1,200 tons. With 12,000 data centers, that implies a total lifetime cement footprint of just over 14 million tons. Using data from the EPA, which typically indicates that every ton of cement produced releases roughly 0.78 metric tons of CO2-eq gases, we estimate that the total lifecycle emissions footprint from data center cement is about 11.2 million metric tons.86

Metals and Materials

Data centers are full of metals, from copper and aluminum to tin and graphite. For example, copper is widely used for wiring and graphite is widely used for thermal management. Let’s start with copper as a small gateway to our broader analytic approach. Based on figures from data centers in the United States, we estimate that the average data center contains roughly 12 tons of copper per MW.87 With an estimated 133,000 MW of peak allocated capacity worldwide in 2026, that implies a total embodied copper footprint of 1.6 million tons.88 Based on lifecycle assessments performed for Chinese copper production, and accounting for improving efficiency in copper production, we estimate that it takes roughly 60 GJ to produce one ton of copper on average.89 Converting that to TWh for all the embodied copper in global data centers, the total lifecycle energy footprint comes out to roughly 25 TWh.

Data from the World Economic Forum shows that the typical metals footprint in data centers is roughly 70 tons per MW.90 To avoid performing a separate analysis for dozens of different metals and materials, we’re going to follow a similar approach like the one we explained for copper above, just at an aggregate level across all metals and materials. Given the total power capacity above, the final estimated mass footprint for the all the metals in global data centers is roughly 9.3 million tons, but we can take away the 1.6 million tons from copper above to avoid double-counting, so the real metals mass footprint for which we need to measure energy consumption and emissions is about 7.7 million tons.

Assuming an average production run of 50 GJ per ton, the total lifecycle energy footprint is 128 TWh. Adding that to the 25 TWh from copper gives a total of 153 TWh for the energy footprint of all data center metals worldwide. Allocating 40% of that total to AI leaves the AI share of the data center metals energy footprint at about 61 TWh. We again assume that 60% of the lifecycle total will come from this year alone, so 37 TWh of energy is allocated towards AI-driven metals and minerals production. We estimate the emissions footprint of this 37 TWh of energy at roughly 17 million metric tons of CO2-eq, by assuming that the carbon intensity of the global electric grid also applies to the carbon intensity of metals and minerals production.

Semiconductor Fabrication

Global silicon wafer shipments totaled almost 13,000 million square inches in 2025.91 It has been estimated that the lifecycle electricity intensity of semiconductor fabs is roughly 1.43 KWh per square centimeter.92 After converting to square centimeters, multiplying by 1.43, and converting from KWh to TWh, the total electricity footprint of semiconductor production in 2025 comes out to roughly 120 TWh. Considering total global electricity consumption was roughly 31,780 TWh in 2025, that means roughly 0.4% of the entire world’s electricity consumption is devoted just to producing semiconductors.93 We allocate 40% of this total to AI, so the AI share of semiconductor consumption is 48 TWh. The global CO2 emissions footprint of the semiconductor industry has been estimated at roughly 190 million metric tons for 2026.94 Allocating 40% of this total to AI means that the AI share of semiconductor emissions is roughly 76 million metric tons of CO2.

But we're certainly not done yet, because on-site electricity consumption is only one aspect of the semiconductor industry's energy and emissions footprint. For example, before any silicon wafers are processed in fabs, the wafers have to be produced in the first place, and that requires energy-intensive methods like the Czochralski process (see Figure 4). The Department of Energy has estimated that as much as 60% of an industry's manufacturing footprint lies upstream from direct on-site operations.95 We assume 60% for the semiconductor industry, meaning that its total global energy footprint comes out to 300 TWh in 2026. Deloitte has estimated that AI-focused chip sales alone will comprise roughly half of all semiconductor industry revenues in 2026.96 We shouldn’t allocate a full half of the energy footprint to AI on the basis of this result, since AI chip sales carry extraordinary profit margins based on limited supply and monopoly power. But we believe it’s reasonable to allocate 40% to it, especially because producing advanced semiconductors for the AI industry is far more energy-intensive than producing microcontrollers or other simpler devices. With that assumption, the AI-driven share of the semiconductor industry’s energy footprint in 2026 is 120 TWh. This is a large figure, but it's also likely conservative and understated.

Industrial Robotics

Robots are the pinnacle of physical AI (see Figure 5). There were about 4.6 million operational industrial robots globally in 2024.97 We estimate that roughly 5 million industrial robots are operating worldwide in 2026. Industry data from the previous decade shows that the average robot consumes roughly 22,000 kWh of electricity a year.98 Although it's true that newer robots have achieved greater power efficiency, the rise of lights-out manufacturing has also meant that they're running longer on average every single day, so we'll still use the average figure cited above. For 5 million robots, that unit-level average consumption implies roughly 110 TWh of aggregate global electricity consumption. Since the average carbon intensity of the global electric grid is 460 grams of CO2-eq per kWh, the total operational emissions footprint of industrial robots comes out to roughly 51 million metric tons of CO2-eq a year.

Assuming that robots and data centers are being powered by the same global fuel mix (see above), the primary energy associated with this electricity generation equals roughly 248 TWh. Since we define robots inherently as physical AI, this entire sum is included in the AI energy ledger. This is a staggering amount of energy that most people don’t think about when it comes to AI, because they’ve come to understand AI in very limited terms, as being just about neural networks and LLMs. Here we can see that even slightly broadening the range of the concept has major implications for how we assess its global energy and emissions footprint.

Transmission Networks and Telecommunications

In addition to data centers, data transmission networks involving cell towers, switching centers, and fiber-optic injection nodes also consume vast amounts of electricity, at least 360 TWh in 2025 if not much more.99 These networks are absolutely critical for AI workflows, since instructions to AI models could not be sent and answers could not be delivered to end users without them. Using Cisco data about the AI inference share of network traffic, we estimate that roughly 1% of this 360 TWh is devoted specifically for AI operations, meaning that AI deployments consume annually about 3 TWh of electricity worldwide through data transmission networks alone.100 It’s not a big number at all, but we did want to surface it for public attention because we expect it’s a number that’s going to rise rapidly over the next few years.

Transportation and Logistics

The manufacturing sectors powering the AI infrastructure buildout have globally extended supply chains. Minerals are first mined in a given location, then they might be shipped halfway around the world for refining, shipped somewhere else for final assembly, and the finished product then gets routed to halfway around the world again to be sold to consumers in a given country. The total transportation sector accounts for roughly 30% of global energy consumption, which comes to about 50,400 TWh for 2026.101 What we’re really interested in is freight and logistics, and that’s about 40% of the global transportation sector, or about 20,160 TWh in 2026.102 We assume that the global primary energy share for data center electricity use, roughly 1%, should also be allocated to this logistics energy footprint, so that means roughly 201 TWh of energy from the logistics industry in 2026 was devoted to shipping and moving material components and computer hardware for data centers, including steel, concrete, minerals, and microchips. Allocating 40% of that to AI implies that the AI-driven energy footprint for logistics in 2026 was about 80 TWh.

To perform a sanity check on this figure, we’ll pick an alternative methodology. Data centers had a total global power capacity of roughly 125,000 MW in 2025, and we’ll stick to this more conservative figure for the analysis rather than the 133,000 MW of capacity we estimate for 2026. A typical server rack draws about 1,000 W, so that implies 125 million hardware units, from server racks to storage arrays, are operating in global data centers as of this year.103 Research indicates that transporting a single server has an emissions footprint of roughly 25 kg of CO2-eq emissions.104 Transporting a heavier storage array takes up to 486 kg of CO2-eq emissions. We assume a blended average of 300 kg of CO2-eq emissions, which implies a total carbon footprint of 37.5 billion kg of CO2-eq emissions. We’ll take diesel fuel as our proxy for determining the freight energy footprint from these carbon emissions. Studies show that the emission factor of diesel is roughly 0.074 kg CO2/MJ, which in plain language means that burning 1 MJ of fuel releases 0.074 kg of carbon emissions.105 Another way of saying the same thing is that for every kilogram of carbon emitted, 13.5 MJ of energy are released. Given the carbon footprint above (37.5 billion kilograms), the total energy footprint comes out to about 506 billion MJ, which equals roughly 141 TWh.

This is somewhat lower than our official figure above, but at least they are both in the same ballpark. Our alternative methodology would likely yield a higher estimate if we accounted for the distribution of logistic channels, like what travels via air or rail or sea, and how long things are traveling in each of those components. Using the diesel emission factor from above, 80 TWh of energy consumption from the AI component of the freight and logistics industry would yield roughly 21 million metric tons of CO2-eq in annual emissions.

Renewable Energy Manufacturing

Older sources held that renewables provided just over 25% of data center global electricity consumption, but this share is almost certainly much higher for 2026, so we’ll stick to the 31% assumption used above.106 Considering that so many renewable energy systems are rapidly being built specifically to facilitate the expansion of data centers, it makes sense to include them as part of the lifecycle footprint of AI.107 While renewable systems like solar arrays and wind turbines have a negligible emissions footprint during their operational phase, it takes energy to produce them in the first place, and that manufacturing process does indeed release significant emissions. As we calculated at the top of the section, renewables were responsible for about 244 TWh of global data center electricity consumption.

Lifecycle assessments from 2024 suggested that the cumulative energy demand of average photovoltaics in the United States was roughly 13,000 MJ (oil-equivalent) per kW.108 Assuming that average figure holds globally and converting to TWh per GW, we get about 3.6 TWh per GW of lifecycle energy demand required to produce photovoltaics. According to the International Energy Agency, annual global additions of renewable energy capacity reached 800 GW in 2025, of which 75% came from solar alone.109 That implies 600 GW of new capacity was due to solar, and using the footprint cited above (3.6 TWh per GW), we estimate that 2,160 TWh of energy was required to produce all of these solar components. Using a very conservative footprint of 0.5 TWh per GW for the wind and battery systems together (the remaining 25%), we estimate it took roughly 100 TWh of energy to produce the new annual wind and battery capacity installed in 2025. The final tally for the lifecycle energy demand to produce all new renewable capacity installed in 2025 is 2,260 TWh. Rounding this out and accounting for further growth in 2026, we assume the total annual manufacturing energy footprint of renewable energy is 2,300 TWh.

Since US tech companies alone are driving almost 50 GW of new renewably energy capacity demand for data centers, it’s reasonable to assume that 60 GW of new renewable energy capacity is meant for data centers on a global basis, which is about 7.5% of the total 800 GW from above.110 Applying this percentage to the total renewable manufacturing footprint of 2,300 TWh yields 172 TWh. Allocating 40% of that to AI means that the AI-driven energy footprint of renewable energy manufacturing in 2026 is 70 TWh. Assuming that the same carbon intensity as the global grid average applies to this energy footprint, we estimate that AI-driven renewable energy emissions footprint for 2026 was roughly 32 million metric tons of CO2. A critical note of caution about this analysis: we have decided to exclude the upstream energy and emissions footprint of the roughly 80 GW of natural gas power plants under construction in the United States that are meant to support the data center expansion. There was too much noise and uncertainty in the data, and in any case this particular footprint will not be that large for 2026, probably no more than 20 TWh. However, we want to flag it as a major issue going forward, since these plants will consume vast amounts of energy once they become operational.

Land Use Change Emissions

Data centers have an enormous land footprint, and that land is usually occupied with plants and wildlife before construction begins. Many data centers are built on sensitive wetlands, marshes, and peatlands that store vast amounts of carbon dioxide. Ecologists estimate that the world's peatlands store twice as much carbon as all the world's forests put together, so building data centers in these ecologically sensitive areas is an especially terrible idea.111 When these pristine natural habitats are destroyed, the underlying carbon is suddenly exposed to atmospheric oxygen, triggering a wave of carbon dioxide emissions which can last for years. Damaged peatlands account for roughly 5% of all global anthropogenic CO2 emissions a year, so conserving them is of paramount importance for global ecology.112

Let’s begin our estimate of AI-driven land use change emissions at the 12,000 operational data centers in 2025. The largest hyperscale data centers can sometimes exceed 50 acres, but most data centers worldwide are much smaller (typically only a few acres). However, we're going to adopt a lifecycle perspective, so we'll also include the land footprint of renewable energy systems and natural gas power plants that are being constructed, or have been constructed, specifically to support the data center buildout. With this in mind, we’ll adopt an average land lifecycle footprint of 100 acres per data center in our analysis, which yields a total global land footprint of 1.2 million acres (or 486,000 hectares). In addition to the lifecycle off-site footprint, this global footprint includes not just the direct footprints of the facilities but also the land occupied by adjacent facilities and infrastructure, such as parking lots, retention ponds, and electrical substations.

To simplify our calculations, we’ll assume that all of this land was cleared evenly across the past 20 years, which we know is inaccurate as data center construction picked up significantly in the early 2020s. We’ll also assume for simplicity’s sake that that the total carbon sequestered in these affected environments enters the atmosphere at equal annual chunks across time, which is also inaccurate, as research has shown that about 50-60% of CO2 emissions in a particular year from forest conversions come specifically from land conversions that happened in that year.113 But as first approximations, these will do.

Next, we’ll assume that a typical hectare of land impacted by data center construction holds plants and soils that collectively sequester 450 metric tons of CO2, based on research studies about forest biomes and other environments.114 That implies an annual release of 22.5 tons of carbon dioxide per hectare per year from typical land and regular biomass. But peatlands are not regular by any definition. They store vastly more carbon than most plant biomes above ground, so we’ll assume they release roughly 45 tons of carbon dioxide per hectare per year (twice as much as the figure above for regular biomass). We’ll also assume that the data center buildout mirrors the characteristics of the global land surface, meaning that 3% of the data center land footprint is on peatland soils (14,600 hectares) and 97% of it is on regular land (471,400 hectares). Putting it all together, we see that land use change emissions from regular biomass contributed roughly 10.6 million tons of CO2 emissions in 2026 (471,400 multiplied by 22.5 from above) while land use emissions from the peatland component were roughly 660,000 tons of carbon dioxide (14,600 hectares multiplied by 45 from above). In sum, we estimate that global land use change emissions for 2026 as a result of data centers totaled roughly 11.3 million tons of CO2, an astonishing figure equivalent to the annual emissions of over two million cars.

Construction Industry Footprint

The construction industry is a major component of the AI-driven infrastructure buildout. Total US spending for data center construction will likely reach roughly $29 billion this year.115 Since the US has roughly half of the world’s data centers, we’ll assume total global spending will reach $59 billion. Average global energy intensity was roughly 3.87 MJ/USD in 2022.116 We’re going to assume it’s about 3 MJ/USD for the construction industry, so with that assumption we get an estimated 177 billion MJ for the global data center construction energy footprint in 2026, which comes out to 49 TWh. Allocating 40% to AI means the AI-driven data center construction energy footprint was 20 TWh in 2026. Since construction equipment like bulldozers, excavators, and tandem rollers overwhelmingly use diesel engines, which have an emissions intensity of about 265 grams of CO2 per kilowatt-hour (kWh) of energy (265,000 metric tons per TWh), that implies the total emissions footprint from data center construction in 2026 was about 13 million metric tons.117

Downstream Consumption and Rebound Effects

Up until now, we've been talking about the upstream factors necessary for AI diffusion: mining, industrial production, logistics, and so on. But we haven't yet considered the downstream effects. Once households, businesses, and governments adopt AI systems, they may also adopt different behaviors and strategies that require energy to implement. Although AI can be used to optimize smart grids and shipping routes, researchers like Lynn Kaak and others have pointed out that it can also be used to accelerate oil and gas exploration and extraction, along with other energy-intensive activities.118 The goal now is estimate the energy and emissions footprint associated with these downstream activities and rebound effects.

It’s abundantly clear from the research we already cited when we discussed the Jevons Paradox (also known as the backfire effect in economics) that capital-intensive industries usually produce rebound rates exceeding 100% over time. Thus, the real question isn’t if AI will lead to more downstream global energy use over time on a net basis, but how much. To get a sense for that, we need to first have a sense of the expected energy savings from AI deployment. Using data from GeSI, we adopt an abatement factor of 10 for our analysis, meaning that for every unit of energy consumed by the AI sector, we expect it to save 10 units of energy over the following five years in the absence of backfire effects.119 This gross abatement assumption is also supported by figures from the Boston Consulting Group, which estimated that AI could “mitigate 5% to 10% of global greenhouse gas emissions by 2030.”120 With a roughly 1% footprint, that would suggest a gross abatement of around 10. Of course, we do think that a strong backfire effect will materialize from global AI deployments, so the next task is to provide an estimate for it. Based on research for industrial sectors worldwide and in the United States, macroeconomic rebound effects can range anywhere from 110% to 800% over various timescales.121 We’re going to adopt a lower-end rebound rate of 140% over the following five years. This is actually a fairly conservative figure, given the rising entanglement of AI with virtually all industries and economic activities around the world; the real backfire effect over the following five years will likely be higher.

We use the following equation, where R is the estimated rebound effect (140%), to calculate the net extra energy consumption:

We estimated the global primary energy of AI electricity use at 709 TWh above. Per above, we use an abatement factor of 10 over the following 5 years. That means we expect AI to lead to global energy savings of roughly 7,100 TWh over the next five years. However, since AI deployment is expected to backfire, the net extra global energy consumption over the next five years is roughly 2,840 TWh, given the rebound formula above and the rebound rate of 140% (so R is 1.4). But we cannot naively add this to the total in 2026. It has to be annualized first, yielding 568 TWh of extra AI-driven energy consumption from the rebound effect in 2026. Using the average carbon intensity of the global electric grid from above (460 grams of CO2 per kWh, or 460,000 metric tons per TWh), we estimate the total emissions footprint of AI-driven rebound effects for 2026 at about 261 million metric tons of CO2.

It's true that rebound effects from any technological deployment generally don't scale in a linear way over time. There might be efficiency improvements at first and then the rebounds kick in towards the end of the observation period in question. That's likely the case with AI as well, so from that perspective it might seem unfair to allocate all of that 568 TWh to 2026. But we argue that this concern is mitigated by the fact that there are actual rebounds happening this year associated with AI deployments from prior years. AI didn’t magically start in 2026 with gigawatts of capacity appearing out of nowhere. The AI industry was already mature in many ways well before 2026, and all that diffusion has inevitably produced major rebounds by now.

One may object that these extra emissions we have calculated may have happened anyway, either partly or entirely, even if AI had never been invented. Suppose you have an idea of what to make for dinner and you ask an AI system about the ingredients, the recipe, which grocery store to go to, etc. Going to the store takes energy from your car, and also produces tailpipe emissions. Cooking the meal may require using a stove or oven, and they also consume energy and have an emissions footprint. The question then becomes: to what sector should these consequences be allocated? If your actions and behaviors are driven by the information you received from the AI system, we think there is a strong case to be made that the energy and emissions footprint should be allocated to AI. Of course, you may have driven to the grocery store to get the dinner ingredients even without using AI at all, and that activity may still have used up the same amount of energy and caused the same emissions. But that’s no problem at all. In that case, the energy and emissions footprint should be allocated to some other sector.

The fundamental issue isn’t just about the aggregate total, but about the allocation between the sectors that collectively determine that total. Simply put: if you did something because of AI, the energy and emissions footprint of that activity should be allocated to the AI sector, either as a downstream impact or a rebound effect. We understand that some researchers, including critics of the AI industry, may object to this conclusion, but we believe it’s fully consistent with established accounting standards from major international organizations, especially the GHG Protocol.

Final Results and Concluding Thoughts

Adding it all up, we estimate that the AI sector is responsible for almost 1,900 TWh of global energy consumption in 2026. And after accounting for additional factors that we did not explicitly analyze, like the construction of new electrical substations, robotic-centric warehouses, water reclamation facilities, and other infrastructure specifically designed to support the data center buildout, it’s likely that total AI-driven global energy demand is already closer to 2,000 TWh. Even at 1,900 TWh, the AI share of total global energy use in 2026 would be roughly 1.1%. We also estimate, after adding up all the sectoral numbers and landing at roughly 650 million metric tons of AI-driven CO2 emissions, that the AI share of global GHG emissions in 2026 stands at about 1.2%, with total annual global emissions sitting at roughly 55 gigatons of CO2-eq, albeit with a wide range of figures from EDGAR to UNEP (anywhere from 53 gigatons to 58 gigatons depending on a broad array of different assumptions). As we’ve discussed already, these numbers are much higher than many others in the media and academia have guessed on the basis of simplistic calculations and naïve assumptions.

The projections up to 2030 are obviously far worse and imply a catastrophic impact to global ecological stability. ChatGPT alone “processes approximately 2.5 billion prompts per day worldwide.”122 The scale of AI growth and diffusion globally is truly astounding, and the industry’s corresponding ecological footprint will soon rival that of major countries like India. It's not farfetched at all to think that the AI share of total global energy demand could rise to something like 5% by 2030, rivaling and in some cases surpassing the footprint of established dominant industries. The potential consequences of this extra resource pressure on biodiversity, global warming, freshwater scarcity, and biogeochemical cycles could be absolutely staggering and devastating for the stability of modern civilization.

There are a few factors that could alter this otherwise dreadful trajectory. First, political pressure could snowball and force the AI industry to stabilize or downscale. Second, in the absence of effective political pressure, the industry itself may slow down or self-implode if the scaling laws start breaking down. Third, investments may decline even if AI models continue making rapid progress, especially if the returns and profits just don’t show up, like they have failed to do for the major AI frontier companies so far. None of this means that AI goes away. At this point, it’s a permanent feature of our lives to various degrees. But it does mean that the AI infrastructure shockwave could slow down significantly, and with all the ecological challenges looming in the rest of the century, we argue that humanity must reconsider its priorities.

The latest infrastructure shockwave is already strongly contributing to global energy use and GHG emissions. At what point do societies or governments evaluate the long-term cost-benefit of this sector’s runaway growth? AI may one day be able to live up to the ideals espoused by its proponents, but that value is called into question today by the simple fact that it is also contributing to the conditions that will render Earth hostile—even uninhabitable—for billions in the coming years. Unless our societies and governments can cooperate and effectively regulate or control this industry, we worry that the entire human species will pay a heavy price.

I have an email from February that is promoting the Pixel 10; you can get £610 back with an eligible phone trade-in, which includes a £200 bonus.

This represents a shift away from the classic contract model, where you get your device and airtime plan together from a network provider like EE or Vodafone.

In February 2026, Google introduced “instant trade-in” for Pixel purchases via the Google Store, where the estimated trade-in value is applied as an upfront discount at checkout rather than a refund later.

This is designed explicitly to remove as much friction as possible from buying a handset directly from Google, in what is called a “direct-to-consumer” (DTC) model.

DTC brings lots of benefits for Google: they hold the customer relationship, can get a higher margin on each unit, and they have direct control over their pricing and promotions. For a company like Google, this is essential, as their phones are a crucial component in their wider service ecosystem, such as Gemini, Photos, email, etc.

Now, on the surface, this looks like a sensible, savvy, and sustainable practice, right?

Sustainable Marketing

If Google were to sell me a discounted new phone that would incentivise me to upgrade, I’d have to figure out what to do with my old phone, which means I’m probably going to leave it sitting in a drawer in the kitchen until I next move house. Google’s own website states:

Building products

for the planet.

And for everyone.

“From product design to manufacturing and across our supply chains, we are working to address our environmental and social impact at every step”.¹

They have 2 schemes,

1. The Recycling Programme

The website states: “Our recycling programme keeps reusable materials out of landfill. That means less waste for the planet and less clutter for you.”²

2. Refurbish and Re-sell

What exactly they do with your old phone is quite opaque; Google’s website doesn’t actually say.

Google make a phone, and if their marketing and trade-in mechanism has worked, you will give it back to them 12 months later, at which point they will “refurbish” it and sell it on to someone who didn’t get caught up in the 12-month marketing trap.

But how sustainable is this practice? How effective is recycling, and is this model actually doing anything to make smartphones more environmentally friendly?

The Anatomy of Smartphones

The number of phones in circulation is increasing exponentially year on year, until eventually they’re too old to function and are broken down into parts. As of 2022,. there were more than 8.58 billion mobile subscriptions in use worldwide in 2022, compared to a global population of 7.95 billion halfway through the year.³

According to the Environmental Protection Agency (EPA), for every million smartphones you recycle, you can recover:

16.2 tonnes of copper

0.35 tonnes of silver

0.034 tonnes of gold

0.015 tonnes of palladium⁴

This represents a return of roughly 90% of the initial inputs, which sounds great on paper, but these materials make up a tiny proportion of a phone’s physical weight, as they are primarily used as conductive materials like solders, plating, and wiring⁵. They account for ~5–10% of the phone’s physical weight.⁶

The rest of the phone is made up of chips, screens, and batteries, and each of these parts presents a new problem.

The manufacture of chips “bakes in” the resource use; these tiny components are part of energy-intensive and complex manufacturing processes that involve hundreds of steps. You can recycle them for metals, but not for function⁷; once a chip is broken down, it will never be a chip again.⁸

Similarly, the screens are made of glass and plastics derived from crude oil; they are constructed in such a way that, once they are destroyed, they cannot be reused. They are made from several inseparable and highly breakable layers that cannot be separated.

Finally, it’s estimated that recycling magnets and lithium batteries yields a tiny return on materials, and as batteries degrade and the chemical composition changes, they become less effective over time.

That’s not to mention the horrific conditions endured by those mining cobalt for batteries in the Democratic Republic of the Congo (DRC), which have been well documented. Reports from organisations like Amnesty International have documented cases of child labour and dangerous working conditions, with miners facing frequent tunnel collapses, toxic exposure, and highly exploitative wages. Practices like this should have been outlawed decades ago, but for a market as big as smartphones, capitalism not only endures it, it fundamentally relies on it.

Most of a phone’s resource use comes from the initial production; it’s estimated that 70–85% of a phone’s carbon footprint comes from its initial manufacture.⁹

Stagnating Technology

Google has a deep imperative to keep making phones every single year. Pixel 10 came out in August 2025; Pixel 11 is due to be released in August 2026.

When your business model revolves around making a new product every year, you have to find a way to keep selling them.

This seems at odds with what they claim to be a core value of theirs:

“Our devices are designed to stand the test of time, so that people can enjoy them for longer.”¹⁰

A study in 2025 looked at how the changes introduced in annual smartphone releases are perceived by expert reviewers, and whether they truly represent meaningful improvements.

It states that “Between 2007 and 2016, rapid advances in touchscreens, app ecosystems, biometric authentication, mobile payments, and organic light-emitting diode (OLED) displays significantly reshaped digital life”, and through the study concluded that “in the flagship smartphone industry, annual product cycles have shifted innovation from technological breakthroughs to marketing-led perception strategies”, and that “many annual ‘innovations’ are perceived less as genuine advancements and more as strategies to sustain the appearance of novelty”.¹¹

This is where AI comes in. When you see adverts for Pixel phones, pay attention to what their marketing team consider to be features: it’s making sure you get the prettiest version of a picture, it’s making sure you can take pictures of the stars, and it’s making sure Gemini can tell you not to put paprika on your granola.

This is an increasing divergence between the use-value and the exchange value of a Pixel phone. As we see the use-value decrease, as the technology isn’t iterated in a meaningful way, Google’s marketing department ensures the exchange value continues in line with market trends.

Look at this advert to see what I’m talking about.

The particular irony of that advert is that Google is pitching the new Pixel 10 as a phone that bucks the trend, by having it do something as mundane as pick out some flowers for someone.

The Battle for Market Share

In my lifetime, I have gone from cassette tape player to portable CD player to MiniDisc (RIP sweet MiniDisc) to MP3 player to smartphone.

Each advance in technology drove demand because it was actively better than what came before. Each change was an improvement.

Today, the technology doesn’t drive demand, production, or output. As we’ve seen, the “improvements” are acts of marketing, and it’s not even being driven by the profit motive - the growth imperative and the importance of market share are what’s driving it.

Take Samsung and Apple, for example, which between them control 52% of the market, with 32% and 20% respectively¹²; however, it’s estimated that Apple receives around 75% of the profits¹³. Since market share doesn’t correlate with profit share, profits can’t be the primary driving force.

Google creates products every year because it needs to keep growing, and its goal is to eat into that market share, not because consumers benefit or the technology has improved. It is the definition of waste.

If the technology is built to last, then products should only be iterated when technology becomes available to make a significant enough improvement, or released at arbitrary fixed gates, such as every 5 or 10 years.

If you’re truly committed to sustainability, you don’t make things people don’t need and then spend heavily to convince them to buy them.

Google is committed to driving market share for phones in order to compete, to keep growing, and to keep shareholders happy.

The Tendency of the Rate of Profit to Fall

Karl Marx posited that, as capitalism develops, the rate of profit has a tendency to fall¹⁴ - this helps us to understand Google’s situation here.

Google doesn’t make its own phones; Foxconn does, a Chinese company. It also makes Apple’s iPhones - this is an important point that we’ll return to shortly.

Foxconn has invested heavily in automation, machinery, and raw materials - these are not variables that it can influence effectively in the same way it can with labour. This is called constant capital.

If Google were your classic Victorian factory and it had a bunch of workers on a production line to make a phone, and it bought a bunch of raw materials and pumped them through, the main variable it has for capital is wages, so it can pay its workers less. As a result, its profit is variable, and it can control labour. This is called variable capital.¹⁵

As technology in automation improves, this improves for everyone, so the constant capital becomes the same for every company - remember, Foxconn makes Apple’s phones too, so it has the same constant capital that Google does. Improvements in production are equalised in the market by competition.

As a result, companies need to keep producing more and more phones to stay competitive. Market share becomes crucial to selling phones because, ultimately, all these phones pretty much do the same thing; the actual technology isn’t really a factor in consumer choice.

Finally, the other way to compete is through clever financing - similar to the way we finance cars through PCPs and HPs, which I plan to write about in the future - Google has created an incentive in the way you finance your phone.

We’d already done this to some extent with the creation of phone contracts, bundling the device and the plan together for the carrier to sell on. This gives tech companies like Google a guaranteed mass block of sales, as the mobile network operator buys the device wholesale from the phone company.

However, the trade-in mechanism effectively causes the cost of upgrading to fall. It captures the second-hand market value itself, and it shifts sales away from carriers. It is incentivising direct purchasing, allowing it to offset the standardisation of constant capital across the market. It can’t compete on production, so it competes on realisation.

The Sustainability Myth

In a world with dwindling natural resources, a climate crisis that isn’t coming (it’s here), poverty increasing globally, and terrible conditions for the miners who extract the very resources Google claims to take seriously with its stance on “sustainability”, it is creating a sea of waste and actively incentivising the consumer to buy into it through marketing and financial incentives, rather than actually delivering a better product.

Under capitalism, we cannot be sustainable. If Google were to release a new phone every 5 or 10 years, it simply wouldn’t be able to operate in the smartphone market - no one would.

There is no business model where this ends up profitable, particularly in the long term, as Foxconn automates its processes even further.

I’m not going to feed into the individualist myth and leave you with a sense of guilt and responsibility, or tell you to run your phone into the ground; it’s not on us to ensure Google is sustainable.

What I will try to leave you with is an understanding of how companies are having to increasingly rely on subtle marketing and sales tactics in order to keep up this façade.

Tech companies want to believe in the idea of endless growth and endless innovation, and we’re reaching the limits of what we are innovating. The old proverb goes, “necessity is the mother of invention”; I believe that stopped being true for most industries quite some time ago, but shareholders are obligated to pretend like it still is.

The American criminal justice industry is one of the most corrupt in the world. One might think that the main point of a justice system is to deliver justice. But that’s not how it works in America, where innocent people who don’t have a trust fund are chewed up and spit out by the system every single day. The justice industry in the US has become little more than a blunt instrument for the ruling classes to dump millions of people in jail while saddling them with excessive levels of debt. There are many types of corruption in the American legal system, from bribes and conflicts of interest to coercive plea bargaining and corporate money flooding state and local judicial elections (see Figure 1). But what all of them have in common is that they’re subsumed under the broader class-power dynamics of late-stage capitalism, and thus they all point to an industry that manufactures crime, fear, and obedience, functioning as a de facto enforcer of the capital order.

The ideological cosmology of crime serves the interest of the ruling classes. From the nightly news to the tabloids, the dominant coverage of criminal behavior focuses on violent crime and property crime, like shootings, murders, and burglaries. But financial crimes are largely marginalized or even ignored. Only a single banker was ever jailed in the United States for the financial crisis and its role in the Great Recession, a period of rampant criminal behavior from the country’s financial elites.1 The FBI does keep some databases on white collar crimes like embezzlement, but they’re very limited and incomplete, absolutely paling in significance to the more well-known violent crime databases within the broader Uniform Crime Reporting (UCR) system.2 Indeed, many kinds of financial crimes are not explicitly tracked in the UCR at all, including insider trading. As scholars like Jeffrey Reiman and Paul Leighton have argued, the criminal justice industry has been deliberately stacked against the poor, at the levels of policy and rhetoric.3 When politicians talk about crime, especially on the right, their overwhelming focus is on violent crimes committed mostly by the lower and middle classes. The systematic looting of society by the rich and powerful is largely absent from political discussions in legacy media controlled by dominant corporations.

It’s not surprising why elites talk about crimes this way: financial crimes with the biggest impacts are committed by the rich, and the rich want us to look away from their criminal behavior and divert our gaze towards the poor and the lower classes instead. But it’s not just that capitalists distract our attention through their control of the media and other institutions. It’s also critical to remember that the ruling classes use law enforcement to actively fabricate and engineer crimes in poor communities so that they can justify further repression and law enforcement operations in those same communities. Street crime is not just something that happens in certain communities because they’re poor. Rather, targeted communities are deliberately engineered to stay poor, and waves of organized oppression are forced upon them to ensure they remain a reliable and expendable source of cheap labor.

These class-power dynamics implicate even the racial dimensions of crime. The United States imprisons large numbers of black people as a way of continuing slavery and Jim Crow through other means.4 Hundreds of thousands of prisoners provide critical labor and many of them are not paid a single dime for their work. One of the critical functions of the American justice industry is to therefore serve as a funnel for a steady supply of cheap labor to the ruling classes. To sustain this decrepit system, the elites authorize police and law enforcement officials to aggressively target black communities through overpolicing and entrapment operations, contributing to higher arrest rates and artificially inflated crime statistics, which are then used to justify further repressive measures. From the War on Drugs to the modern era, the American state has resorted to organized terror as a way of destabilizing and undermining black life in this country. In addition, the capitalist system actively creates poverty and economic devastation in black communities, which then further fuels the fundamental conditions that incentivize more street crimes and gang behavior.

Prison for the Poor, Freedom for the Rich

The biggest problem with the American criminal justice industry is that it’s designed to imprison poor people and to avoid touching the rich. A 2023 report from the Prison Policy Initiative found that, on any given day, roughly 70% of individuals imprisoned in jails have not actually been convicted of anything; they’re mostly there because they cannot afford bail.5 In the United States, jails are incarceration systems operated by cities and counties whereas prisons are incarceration systems operated by state governments and the federal government.

Cash bail is a notorious legal weapon that keeps many poor people imprisoned in local jails. For an infamous example, consider Sandra Bland, a 28-year-old black woman who was arrested in 2015 for a routine traffic violation.6 Her family could not afford the $500 necessary for Bland to be released from prison. Bland was found dead from an apparent suicide in her jail cell three days after being arrested. Contrast her fate to that of Robert Durst, a wealthy real estate heir who in 2001 posted a $250,000 bond and fled the state of Texas after being accused of murder. Durst was later found by police in Pennsylvania and went on trial for his crime. However, he was acquitted in 2003 because of a corrupt jury that believed the verbal rubbish of his high-powered attorneys. It took over a decade before law enforcement arrested him again. He was finally found guilty of murder in 2021 and died the following year.7 These two stories highlight a common pattern: poor people languish in jail, forced to be absent from work and at risk of eviction, while the rich post bond and go free, returning to their normal lives and preparing a solid defense case with their attorneys.

It’s common in the United States for people to languish in jails long before they ever get a trial. A report from the Institute for Crime and Justice Policy Research found that American defendants detained before trial spent a median of 68 days in local jails, a truly shocking figure, and one which is heavily exacerbated because of cash bail.8 Another study found that assigning cash bail produces a 6% to 9% increase in the rate of criminal recidivism.9

The most pernicious aspect of America’s cash bail system is the commercial bail bond industry, which posts bail for defendants in exchange for a fee that usually runs from 10% to 15% of the entire bail amount. The system works as follows. First, defendants pay the fee to the bond company. Second, the company pays the full bail amount to the court, but will require the defendant to offer up collateral in case the defendant skips the court date. If the defendant does show up on the specified court date, the bond company returns the collateral back to the defendant but keeps the initial fee, thus earning a profit since the initial bond payment is refunded to the bond company by the court once a case concludes.

In the early 2020s, there were roughly 15,000 bond agents responsible for bailing out more than two million people every year.10 In 2021, six large insurance corporations were responsible for underwriting 76% of the bail bonds issued that year, including giants like Fairfax Financial Holdings Limited and Bankers Financial Corporation.11 It’s been estimated that an astonishing $15 billion in bail bonds were issued in 2019, with the industry as a whole potentially scoring over $2 billion in profits.12 But for this system to remain consistently profitable, it’s imperative that people actually show up on their court dates, which is why the commercial bail bond industry hires bounty hunters to find absent defendants. These bounty hunters actually have the authority to make arrests and bring people to court, and they’re quite aggressive about doing so, sometimes resorting to outright beatings and physical assaults.13

Only the United States and the Philippines have a large and organized commercial bail bond industry.14 It’s true that other countries also use cash bail, but only under very limited circumstances. In the vast majority of countries around the world, courts might establish steep financial bail terms for serious crimes like murder or terrorism, but routine or mundane crimes, like petty theft and traffic violations, are almost never subject to cash bail. Other countries generally impose non-financial bail terms and restrictions for pre-trial release, such as electronic monitoring, mandating regular visits to police stations, surrendering passports to the court, and so on. These systems can be just as effective as cash bail, if not more so, at ensuring that people show up on their trial dates. And obviously because they’ve mostly taken money out of the equation, they don’t a produce a situation where poor people languish in jail while the wealthy go back to their private yachts, which is how things work in America.

Let’s consider France as an example. A suspect in France is usually held in custody only for 24 hours when the crime in question is punishable by less than a year of prison time.15 In cases where the crime is punishable by more than a year of prison time, the prosecutor may ask for an extension to 48 hours of custody. If the crime you’re accused of can be punished by more than 10 years of prison time, then you’re typically held in custody for a total of four days. Those suspected of terrorist acts can be held in custody for up to six days. In most of these cases, a suspect will appear in court before a judge, who then decides the bail terms. Typical conditions include avoiding contact with victims or witnesses. Cash bail does sometimes occur in France, but it’s extremely rare compared to the United States. By and large, what happens to the vast majority of criminal suspects in France is that they’re detained for a certain period of time after being arrested, assigned a trial date, then eventually released without having to pay any money at all.

Few things epitomize America’s tendency to jail the innocent more than the plea bargain, which has become a standard tool of prosecutorial tyranny. A plea bargain is an agreement between defendants and prosecutors designed to avoid a trial. A defendant will typically plead guilty as a way to avoid a longer jail sentence if the trial results in conviction. Over 95% of all state and federal cases end with plea bargains.16 Although it’s illegal for prosecutors to threaten, pressure, or coerce defendants into accepting a plea bargain, hardball tactics are deployed so frequently that hardly anyone within the criminal justice industry seriously questions them anymore. Prosecutors routinely insinuate that if defendants don’t accept their terms, they could face much harsher charges at trial, a notorious tactic known as “charge-stacking.” Some prosecutors have even threatened to jail a defendant’s family members unless the defendant pleads guilty.17 An inevitable result of this prosecutorial aggression is that legions of innocent people plead guilty, since they’re cowed into submission by the prospect of harsher punishments if they’re mistakenly convicted.

In 2020, ACLU attorney Somil Trivedi wrote an article about the abuses of the plea bargaining system.18 Trivedi opens the article with a discussion about how the legal system destroyed three different lives. He begins with George Alvarez:

In 2006, George Alvarez was charged with assaulting a prison guard while awaiting trial on public intoxication. He knew he didn’t do it, the guards actually jumped him, but the ten year mandatory minimum sentence at trial scared him so much that he pled guilty. Little did he know that the government had a video proving his innocence, but they buried it long enough for prosecutors to extract the plea first. George spent almost four years behind bars fighting for his innocence before finally being exonerated.

Next, he talks about a woman who pled guilty after spending 14 months in prison for simply being involved in a fight:

In 2015, Lavette Mayes got in a fight with her mother-in-law. She had no criminal history and had sustained injuries herself, yet a prosecutor charged her with aggravated assault, and a judge set her bail at $250,000. Unable to pay, she spent fourteen months in jail awaiting trial, unable to see or touch her children once. She lost her job. She developed health problems. Even after getting released on bond and downgraded to an ankle monitor, she struggled to resume her previous life due to the trauma she’d experienced while incarcerated. So instead of spending years fighting her case in failing health, she pled guilty to end the ordeal.

Finally, he talks about a person executed for murder who maintained his innocence until the end:

Just last year, the state of Georgia executed Ray Cromartie for a 1994 murder. The case against him was paper thin and Cromartie maintained his innocence until the end, but Georgia denied every request for DNA testing that could have set the record straight. A lesser-known fact about the case is that 20 years ago, Georgia prosecutors offered Cromartie a plea deal under which he could have been paroled after seven years and free by now. But Cromartie refused to admit guilt, and so the state retaliated by seeking the death penalty and ultimately killing him.

Cases like that of Ray Cromartie remind us that the judicial system doesn’t just imprison poor people in large numbers; it also imprisons many innocent people so it can inflate and distort statistics about its efficiency.

Prosecutors aren’t the only ones who abuse the legal system they’re supposed to protect. Police departments across the country have a notorious habit of extracting forced confessions from innocent people who don’t know about their full legal rights and often don’t even understand the significance of what’s happening. Police have sometimes questioned potential suspects for hours and hours, trying to grind them down psychologically so that the suspects will ultimately confess to their “guilt.”19 A decision from the Supreme Court even allows police officers to make false statements to suspects about any details pertaining to the case. It’s common for the police to claim that they have DNA evidence proving a suspect’s involvement in the crime. Police also pretend to have witnesses who spotted suspects at the scene of the crime. These fabrications are a core part of the interrogation strategy, precisely because they’re designed to break down psychological resistance from the suspect. Roughly 25% of all DNA exonerations involve innocent prisoners who originally confessed to their “crimes” under police pressure.20

And that brings us to judges, the final pillar of the Unholy Trinity in the criminal justice industry after prosecutors and police. In a major investigation published in 2020, Reuters looked at misconduct from judges in America.21 The reporters analyzed thousands of cases:

In the first comprehensive accounting of judicial misconduct nationally, Reuters identified and reviewed 1,509 cases from the last dozen years, 2008 through 2019, in which judges resigned, retired or were publicly disciplined following accusations of misconduct. In addition, reporters identified another 3,613 cases from 2008 through 2018 in which states disciplined wayward judges but kept hidden from the public key details of their offenses, including the identities of the judges themselves.

The results were stunning:

All told, 9 of every 10 judges were allowed to return to the bench after they were sanctioned for misconduct...They included a California judge who had sex in his courthouse chambers, once with his former law intern and separately with an attorney; a New York judge who berated domestic violence victims; and a Maryland judge who, after his arrest for driving drunk, was allowed to return to the bench provided he took a Breathalyzer test before each appearance.

Perhaps the most notorious case was that of Les Hayes, a judge in Alabama who jailed hundreds of poor people simply because they couldn’t afford to pay fines. In 2012, he sentenced a single mother, Marquita Johnson, to 496 days in prison for her failure to pay parking tickets! That’s more time than would be given for negligent homicide. While she languished in jail, Johnson’s three children were taken to foster care, where one was molested and another physically abused. Johnson later remarked that “Judge Hayes took away my life and didn’t care how my children suffered.” The report also describes some of the other people sent to jail by this corrupt judge:

In 2016, the state agency that oversees judges charged Hayes with violating Alabama’s code of judicial conduct. According to the Judicial Inquiry Commission, Hayes broke state and federal laws by jailing Johnson and hundreds of other Montgomery residents too poor to pay fines. Among those jailed: a plumber struggling to make rent, a mother who skipped meals to cover the medical bills of her disabled son, and a hotel housekeeper working her way through college.

Judicial corruption in America is something that reaches up to the highest possible levels. Consider the case of Supreme Court Justice Clarence Thomas, who has turned out to be one of the most rotten and debauched federal judges in history. As a major investigation Thomas has long accepted free trips and vacations from his billionaire pal Harlan Crow, without disclosing them to the public.22 For example, Thomas and his wife traveled to Indonesia back in the summer of 2019, taking Crow’s private jet to get there and then island-hopping around on his super yacht. In 2014, Crow actually bought several real estate properties from Thomas and some of his family members for roughly $130,000.23 The mother of Clarence Thomas continued to live in one of the properties purchased by Crow for years after the sale. Crow even pumped $36,000 for renovations into the house shortly after buying it.

As if all that wasn’t bad enough, Crow also funded the private school tuition for Mark Martin, the grandnephew of Thomas, who took Martin in his legal custody when Martin was six years old.24 The elite private boarding school in Georgia, Hidden Lake Academy, had a monthly tuition exceeding $6,000. Martin spent about a year in total at the school. And Crow isn’t the only billionaire bankrolling Clarence Thomas. The health care industry mogul Anthony Welters once loaned Thomas roughly $267,000 to purchase a luxurious RV, and it’s unclear of Thomas ever fully repaid the loan.25 Thomas frequently used the RV to highlight his man-of-the-people persona, without telling the people that his corrupt rich friends had bought it for him. Another investigation uncovered that multiple billionaires have lavished Thomas and his wife with plenty of gifts and favors.26 Some of those lavish favors included the following:

At least 38 destination vacations, including a previously unreported voyage on a yacht around the Bahamas; 26 private jet flights, plus an additional eight by helicopter; a dozen VIP passes to professional and college sporting events, typically perched in the skybox; two stays at luxury resorts in Florida and Jamaica; and one standing invitation to an uber-exclusive golf club overlooking the Atlantic coast.

Thomas may be the worst in terms of corruption, but he is not alone. Justices like Samuel Alito and John Roberts have also received numerous free vacations and other gifts from corrupt capitalists looking to influence the Supreme Court. has an excellent recent post outlining the ethical lapses and conflicts of interest from Roberts and his wife.27

This evidence collectively shows that the American criminal justice doesn’t just need to be reformed. It needs to be overhauled from root to branch. But the fundamental problem is that judicial corruption is ultimately linked to the broader corruption of the capitalist system itself. Judges, prosecutors, and police officers don’t just act in isolation. They are constantly pressured, empowered, and incentivized to do the bidding of capital. And nowhere is this corruption more apparent than in the racial dynamics that plague America’s criminal justice industry.

Race and the Criminal Justice Industry

Prison labor has always been a popular strategy for locking up racial minorities and forcing them into de facto slavery.28 Indeed, the Thirteenth Amendment to the United States Constitution explicitly says that slavery is legal for people who have been convicted of a crime. Prison workers in America make everything from office furniture to military equipment. In 2019, the Bureau of Justice Statistics at the Department of Justice conducted its Census of State and Federal Correctional Facilities. The census found that roughly 1.2 million inmates, nearly all, were held in prisons that offered at least one work program.29 The ACLU estimated in a 2022 report that about 800,000 of these prisoners were actively working and that they, in a single year, collectively produced over $2 billion worth of goods and provided over $9 billion of prison maintenance services.30

The report makes clear that prison workers generally make pennies on the dollar for every hour they work; the lucky ones might make anywhere from $1 to $2 an hour. In many states across the South, prisoners usually earn nothing at all. These absurdly low wages don’t allow them to save up or plan for the future; most of what they make is quickly spent on expensive prison services, like food and phone calls. The companies that employ prison workers might pay them a nominal wage on the one hand, but almost every penny is squeezed back out by forcing the inmates to pay high prices for necessary services. The basic reason why these workers are de facto slaves is because they can’t really refuse to work. If they do, they are ruthlessly targeted and punished by prison authorities, either by being placed in solitary confinement or by being denied basic services.

Prisons are designed to constantly drain wealth from prisoners, and the biggest problem isn’t necessarily the private prison system. The real private sector threat to prisoners stems from all the third-party vendors and contractors that provide things like phone calls, commissary snacks, and health-related services. In a major 2026 report, the Prison Policy Initiative tracked the sources of spending in the criminal justice industry with the most recent annual data.31 Roughly $3.1 billion a year went towards spending for private prisons, less than 1% of the total (see Figure 2). The total direct annual costs to incarcerated individuals and their families was almost $28 billion, encompassing everything from fees and fines ($15 billion) to other costs, like bail and civil asset forfeiture (almost $13 billion). The Fines and Fees Justice Center summarized the report as follows:

Of the $445 billion spent annually, only $34.3 billion goes towards direct costs for the care and legal representation of people who are incarcerated: $7.9 billion for public defense, $14.6 billion for healthcare, $3.2 billion for food, $5.9 billion for construction, and $2.7 billion for utilities. That is less than 8 percent of total spending, and only 16 percent of the policing budget alone. The overwhelming majority of spending flows to those who run the system, with nearly half of correctional spending going to staff salaries.32

As the criminologist Margaret Kovera argued in a 2019 paper, “racial disparities in the criminal justice system are well documented and widespread.”33 Conservatives often claim that black people are disproportionately represented in prison populations because they commit crimes at disproportionately higher rates, but this argument ignores the fact that the number of people in prison at any given time is a function of many other factors besides crime rates. Kovera points out that “although racial disparities in arrests account for between 70% and 75% of the racial disparities in incarceration, decisions made after arrest by judges and attorneys regarding postarraignment detention, plea deals, charged crimes, and sentencing also contribute to differential incarceration of Black and White offenders.”

And it’s actually far worse than that. Criminal statistics are not just innocent records of criminal behavior. They are often artificially created and engineered by the ruling classes as they target ethnic minorities. Stash house stings from the Bureau of Alcohol, Tobacco, and Firearms (ATF) are some of the most notorious examples, and even local police departments have gotten in on the act. A typical stash house sting works like this: law enforcement officials or confidential informants reach out to some unsuspecting people and hatch a plan to steal a stash house full of drugs. The stash house is completely fake, it doesn’t exist. It’s purely a fiction designed by law enforcement to entrap desperate people into committing a crime. Law enforcement promises a huge payment to these individuals, who are usually poor and desperate for money, after they carry out the operation. Once the entrapped “criminals” arrive at the site, they are arrested. Data shows that roughly 90% of the people arrested in these stings are categorized as either Black or Hispanic.34

Although stash house stings have only led to a few thousand arrests at most in the past few decades, reverse stings have been far more consequential, with thousands of people arrested annually by local police departments. Tens of thousands were arrested in New York City alone during the aggressive policing era of the late 1990s and early 2000s, when the broken windows theory was used to justify repressive state measures to ostensibly bring crime under control.35 In a reverse sting, the police play the role of dealer and provider, trying to entice and entrap random people into buying illegal drugs, or sex in the case of prostitution stings. But as the evidence from New York City and elsewhere has shown, racial minorities are heavily targeted in these sting operations, thus artificially inflating their arrest statistics.

Precisely because law enforcement targets racial minorities so intensely, the odds of tragic and fatal outcomes increases sharply. Mohsen Naghavi and his collaborators published a major research study in 2021 that exposed the severe undercounting of police-related deaths in America.36 The study determined that black people were 3.5 times more likely to be killed by police than white people, a much higher rate than had been reported in some earlier studies.

The researchers described their methodology as follows:

We compared data from the USA National Vital Statistics System (NVSS) to three non-governmental, open-source databases on police violence: Fatal Encounters, Mapping Police Violence, and The Counted. We extracted and standardized the age, sex, US state of death registration, year of death, and race and ethnicity (non-Hispanic White, non-Hispanic Black, non-Hispanic of other races, and Hispanic of any race) of each decedent for all data sources and used a network meta-regression to quantify the rate of under-reporting within the NVSS. Using these rates to inform correction factors, we provide adjusted estimates of deaths due to police violence for all states, ages, sexes, and racial and ethnic groups from 1980 to 2019 across the USA.

Their analysis led to the following conclusions:

Across all races and states in the USA, we estimate 30,800 deaths…from police violence between 1980 and 2018; this represents 17,100 more deaths…than reported by the NVSS. Over this time period, the age-standardized mortality rate due to police violence was highest in non-Hispanic Black people (0.69 per 100 000), followed by Hispanic people of any race (0.35), non-Hispanic White people (0.20), and non-Hispanic people of other races (0.15)…Between 1980 and 2018, the NVSS did not report 55.5%…of all deaths attributable to police violence.

The journalist Radley Balko published a famous compilation and summary in 2020 of roughly 300 studies dealing with racial bias in the criminal justice system.37 The studies show racial disparities in sentencing, police searches, jury selection, plea bargains, bail amounts, and many other areas. For example, a 2017 study using data from the US Sentencing Commission, a federal agency, found that black men received almost 20% longer sentences than white men for crimes that were covered by identical sentencing guidelines. The study accounted for age and past criminal history. For another example, a 2020 ACLU report found that black people are 3.5 times more likely to be arrested for marijuana offenses than white people, even though the two groups use marijuana at similar rates.38

The National Institute on Drug Abuse reported in 2000 that white students used crack cocaine at 8 times the rate of black students.39 But they are not punished with the same level of intensity, not even close. If the War on Drugs had been about targeting drugs, then authorities should’ve focused on areas with notoriously high drug use, like college campuses. Instead they went after inner city neighborhoods because their real targets were just poor black people all along.

The total penal population in the United States went from around 300,000 in the 1980s to about 2 million by 2010.40 In many major American cities, it’s not uncommon to see roughly 80% of young black men with criminal records.41 Drug convictions accounted for the majority of this increase. In some cases, first-time drug offenders can face 10 years in prison. To avoid this terrible fate, thousands of black people take plea bargains for short periods in jail, even for crimes they did not commit. The United States has the largest prison population in the world, even higher than China.

Drunk driving, most of it done by white men, results in many more deaths than cocaine use. But drunk drivers are often charged with misdemeanors. The recent opioid epidemic in rural America, which is largely white, brings this point into stark focus. When inner city black neighborhoods struggled with poverty and crime, the solution was to put millions of black men in jail. But when rural whites started struggling with poverty and drug abuse, the proposed solutions emphasized funding for jobs and mental health services.42 In other words, the crack epidemic and other drug crises for blacks were viewed as a criminal issue by the ruling elites. But those same elites reframed the drug crisis in rural America primarily as a mental health issue. The cultural and political response to the struggles of these different groups has varied dramatically. One received far more compassion and “understanding” than the other.

The problems don’t end when people get out of jail either. Former inmates are often treated like second-class citizens when they are released. They are often disqualified from food stamps, public housing, and student loans. They face discrimination when looking for jobs. They are frequently barred from serving in juries. Aside from the 2 million people in jail, another 5 million people are on probation or parole, predominantly for nonviolent offenses.43 And millions more have to live with the harmful legacy of their prior imprisonment. As scholars like Michelle Alexander have noted, the prison label is often just as important as the prison time.

A study from the National Employment Law Project (NELP) showed how employers routinely violate the standards of the Equal Employment Opportunity Commission (EEOC), which has stated that flat bans based on a prior criminal history may be illegal.44 In 1987, the EEOC issued guidelines that said employers may consider past criminal history when hiring people only if they take into account certain factors associated with the specific crime, like the gravity of the offense or the length of time that has passed since the crime was committed. But the NELP study of job ads on Craigslist found flagrant violations of EEOC guidelines. One ad demanded “No arrests or convictions at any time in the past 7 years.” Others flatly said that applicants could not have any kind of criminal history. Such rampant discrimination makes it extremely difficult for convicted black men to enter the legal economy, since they are barred from employment opportunities right off the bat.

Conclusion

Justice is never impartial in any system, but especially in the current era of neoliberal capitalism, the judicial system has been turned into yet another playground for the rich, another market that can be distorted and manipulated at will with hardly any consequences for the puppeteers. What remains central to this analysis is the cultural hegemony of capital and the way it frames our “common sense cosmology” around the meaning of crime.

Crime in our public institutions is typically presented as something poor people do, thus justifying the resulting political and economic suppression of the working classes via the full force of the state, embodied in the Unholy Trinity of judges, prosecutors, and police forces. By keeping the public focus on crime as a phenomenon that only applies to the lower classes, the ruling elites preserve their power and constantly recreate the economic conditions for an endless supply of cheap labor. In a new era, we would come to recognize that the capitalists themselves are the greatest criminals in history, and the working class is the only historical agent capable of delivering real justice.

Below I’ve included YouTube links for three different podcast interviews, with hosts , , and . They are all amazing thinkers and individuals, and you should go follow them right now! In the sections below, I also provide brief descriptions of each podcast and the main themes in our discussion.

1) Dynamic Complexity with The Blueprint, hosted by Marley Finnegan

is the host of The Blueprint and an all-around amazing human being. Her podcast explores themes around meaning, spirituality, our deeper connections with nature and one another, and the struggle for long-term sustainability in a chaotic modern world. In this interview, we talked about some of my main ideas and the unifying themes behind them, chiefly my emphasis on seeing the world through the lens of entangled and interdependent networks interacting chaotically and unpredictably, and what consequences follow when we understand the world this way. Something I really liked about this interview is that we discussed my childhood in Albania and early influences on my life, especially my grandmother, at the beginning of the discussion. It’s a good opportunity to get to know a little bit more about me and why I turned out the way I did!

2) Empire & Exodus with Macro N Cheese, hosted by Steven Grumbine

is the host of Macro N Cheese at and one of the most wonderful and effective communicators I know. His podcast is focused on MMT, socialism, class analysis, and geopolitics from a left-wing perspective. I’ve done many interviews with Steve by this point and I always enjoy talking with him. In this interview, we took a deep dive into one my recent Substack posts on imperialism and global migration flows, which made the argument that imperialism is the main driving force of global migration patterns in modern times.

3) We Must Decarbonize with Reality Studies, hosted by Jesse Damiani

is the host of Reality Studies and Urgent Futures. I’m a huge fan of Jesse’s podcasts and I watch and listen to every single one that comes out. Jesse is incredibly friendly and approachable, but also piercing and and perceptive in the questions he asks. He has an uncanny ability to bring out the best in his guests and get to the heart of what they really care about. His podcasts are focused on understanding our planetary crises and emergencies, from climate change to the chaotic complexity of modern civilizations. In this interview, we talked about decarbonization: what it means, why it’s important, who’s trying to do it, and how to do it right. This interview covered the global dimensions of the renewable energy transition, along with the challenges and opportunities that it’s facing in different places.

I called the Introduction of my book Nature and Society, a title meant to capture the fundamental dialectical dance between the natural world and human civilization that’s at the heart of the book. It’s a fairly lengthy introduction that summarizes many of the book’s main themes, including the core one: that nature and civilization are a chaotically interactive and integrated system linked together by dynamical feedback loops, bifurcations, tipping points, and critical transitions (see Figure 1). Global civilization is a vast thermodynamic system embedded in the planetary biosphere and is utterly dependent on the stability of that biosphere for its own survival. With the expansion of capitalism causing thermodynamic chaos throughout the planetary biosphere, the time has come to seriously reflect on the direction we’re going, and then to mobilize humanity to change course and chart a path for a better world.

Introduction: Nature and Society

Nature is embroiled in politics, often in the most unexpected ways. The Netherlands is the ultimate master of agricultural efficiency; it produces 6 percent of Europe’s food with just 1 percent of Europe’s farmland.1 But not everything is rosy with the Dutch. In October 2019, farmers in the Netherlands made headlines around the world when they took to their tractors and started blocking roads and highways throughout the country.2 They were outraged by recent court rulings and government proposals designed to reduce nitrogen emissions. The overabundance of nitrogen in Dutch ecosystems was sabotaging the country’s biodiversity, leading to the rapid growth of grasses and harmful weeds that could devastate insect and bird populations, in addition to the country’s pristine landscape. Even worse, many of the country’s shallow water systems had experienced horrible levels of pollution; some urban water areas in the Netherlands started to look like dumping grounds for hazardous waste. The government responded by mandating new standards for limiting the excessive nitrogen pollution being produced by farms and factories. But a Dutch court ruled in May 2019 that the government’s plans didn’t go far enough and violated European Union regulations; the decision immediately halted some 18,000 construction projects worth roughly 14 billion euros.3 All of a sudden, farmers could no longer make critical repairs, add more livestock, or expand their operations in any significant way. Their anger finally boiled over in the fall. Three years later, in the summer of 2022, Dutch farmers started protesting again by dumping garbage and manure all over national highways. Where did all this excess nitrogen come from, and why was it such a big deal, to the point where it brought a country like the Netherlands to a screeching halt on multiple occasions?

Modern agriculture relies heavily on artificial fertilizers, which usually contain large amounts of nitrogen and phosphorus. Plants need these critical nutrients to power their growth, especially since nitrogen and phosphorus are both essential components of many proteins and other biomolecules. But the fertilizers don’t always stay in the fields where they’re dumped; rain and seeping groundwater can carry them to other areas. And that’s where the challenges begin. The excessive concentration of nutrients in a particular environment is known as eutrophication.4 When fertilizer waste makes its way into rivers, estuaries, lakes, and oceans, bacteria and other tiny organisms use the nutrients to grow large colonies of algae, sometimes called algae blooms. These blooms block sunlight for aquatic organisms that live below the surface. And once the algae themselves die, the decomposition of their remains sucks out the rest of the oxygen in the water, turning the area into a “dead zone.” There are hundreds of dead zones around the world by now, with the largest ones reaching the size of small countries. Together, eutrophication and dead zones can harm fishing stocks, reduce tourism, and disrupt water supplies. In 2014, an algae bloom in Lake Erie disrupted the water supply of 500,000 people in the city of Toledo for several days.5 Algae blooms can look curiously harmless from a distance, but they’re not so innocent when they bring down entire ecosystems.

The effects of nature on human civilization are often gradual and cumulative, but the natural world can sometimes deliver a sudden blow that completely reshapes a particular society. In November 1970, the intense winds and storm surges of Cyclone Bhola struck East Pakistan, a large enclave of Pakistan created after the British partition of India in 1947. The brutal cyclone killed up to 500,000 people, making it one of the deadliest recorded natural disasters in human history.6 Entire villages were wiped out and large parts of the region were utterly devastated. Public anger boiled over, especially at the government’s lethargic response to a massive humanitarian tragedy. The political consequences were felt immediately. In December, the pro-autonomy Awami League won a landslide victory at the polls, but officials in West Pakistan refused to accept the results because they didn’t want to hand over the reins of power to a Bengali political party. The rising political instability among the various factions contributed to the outbreak of war in 1971. By the end of the conflict, East Pakistan was gone, and a new nation was born in its place: Bangladesh.

Cyclones and algae blooms can be massive systems stretching over vast distances, but nature can also impact humanity through much smaller biophysical vectors. The tsetse fly lives in sub-Saharan Africa and carries dangerous microbes capable of killing cattle and other livestock in large numbers. The disease produced by these microbes, characterized by fever and anemia, is known as nagana when it affects animals and “sleeping sickness” when it affects humans. It’s estimated that bites from tsetse flies kill at least three million cows and other livestock in sub-Saharan Africa every year, causing billions of dollars in financial damage along the way.7 It’s bad enough that some farmers have historically avoided large parts of sub-Saharan Africa where agriculture could otherwise thrive, simply so they don’t have to deal with tsetse flies. Many people have argued that the deadly flies are a major reason for Africa’s economic underdevelopment relative to other continents, though certainly other historical factors have also played a major role. Nagana is far from the only disease to impose devastating effects on humanity. From smallpox and cholera to typhus and Covid, epidemic diseases have always been some of the most decisive drivers of human history. In the twentieth century alone, smallpox may have killed up to 500 million people, about as many as have died in all the wars humanity has ever fought.8 The eradication of smallpox by the 1970s, following a massive global vaccination campaign, was by far the greatest public health victory in human history.

The complex interactions between nature and human society do not just unfold on the surface of the planet. In many large cities around the world, people extract water from underground water reservoirs called aquifers. But some of these cities are at least partially built on soft soils like clay-rich lake beds, so when too much water is extracted, the soft sheets of land underneath start to compress and undergo a process known as subsidence, meaning that the ground literally sinks. Subsidence is a huge problem that’s currently affecting many places around the world, especially dominant metropolises like Mexico City and Jakarta. In Mexico City, differential rates of subsidence across various districts have led to broken pipes and roads along with collapsing houses and buildings.9 Neighborhoods that were once sitting on flat ground now look like they were built on gently rolling hills. In Jakarta, the capital of Indonesia, subsidence and rising sea levels have cleared out numerous coastal neighborhoods and contributed to massive flooding problems. Throw in excessive pollution levels and extreme traffic congestion and you get a situation where it’s becoming almost unbearable to live in certain parts of the city. Jakarta’s ecological problems have gotten so bad that the Indonesian government has decided to run away and build an entirely new capital, called Nusantara, on the island of Borneo.10 Indonesian officials claim that the new capital city will be “green” and environmentally sustainable, but its very construction has led to the devastation of pristine forests and other critical habitats, to say nothing of the harmful pollutants and greenhouse gases that are also being released into the atmosphere. This example nevertheless underscores a common pattern that repeats throughout human history: when people make a mess of things in one area, they either migrate to other areas or try to somehow shift the locus of ecological degradation. For example, rich countries are notorious for selling much of their trash to poor countries, in effect exporting some of their ecological problems away by taking advantage of desperate countries that need money or other resources.11

Cyclones, epidemics, dead zones, and subsidence are just some of the many ways that the natural world intersects with human civilization. Even though we often think of our social lives as being somehow separate from nature, there’s a profound connection between the world “in here” and the one “out there.” We sometimes have this sense of the natural world as an obscure force swarming with uneventful distractions. We do not consider ourselves as physical systems guided by natural conditions; we see ourselves as independent agents doing whatever we want. We have a gut-level belief in the power of our infinite will, and the idea that other things can stop us or challenge us seems laughable. If we wish to “tame” nature, we build dams to control rivers. We blow holes through mountains and construct tunnels. We like to brag about the impact that our actions have on nature, by showing off our elegant skyscrapers, gorgeous bridges, and lengthy canals. When we wish to “protect” nature, we establish parks and start exploring them through hiking, climbing, and camping. In this frame of mind, we are active and nature is passive; we do things and nature just sits there. But if we are truly so detached from nature, there would be no reason to seek shelter from the heavy rain, no reason to flee from a hungry tiger, no reason to consume oxygen from the atmosphere, no reason to take long walks on the beach, and no reason to enjoy the comfort of shade from the soaring heat. Our collective humanity very much depends on the natural world, for joy, for comfort, and for sheer survival. Nature is full of complex and dynamic systems that are constantly interacting with our societies. The natural world is not simply active in some abstract sense; its collective physical interactions guide and forge many fundamental features of human societies and civilizations. Humanity does not exist on a magical pedestal above the rest of reality; we are just one slice in a grand continuum of physical systems that interact, combine, and transform over time. We too belong to the natural world, and we too experience its interactions and conditions, just like everything else. The wonders of the world are waltzing to the rhythm of restless atoms and oscillating fields.

The fact that we belong to nature means that understanding nature is the key to understanding ourselves, and the historical journey of humanity more broadly. One of the major curiosities about the development of our species is why it took so long for human beings to establish energy-intensive civilization. About 50,000 years ago, our ancestors had roughly the same anatomical features as we do today. Their skulls and bodies were similar to ours. They almost certainly had similar intellectual capacities; some human groups were probably speaking fully developed languages.12 But for all these important similarities, they lived in a very different world than ours. They didn’t have agriculture, roads, writing, and large buildings. They didn’t have government bureaucracies. Why did they lack these things, even though they had the intellectual capacities to create them? To answer this question, it’s useful to know something about their natural environment. For the past two million years, the climate of planet Earth has been dominated by large temperature variations that have produced everything from warm and flourishing conditions to alarmingly frigid patterns in which massive glaciers engulfed much of Europe, Asia, and the Western Hemisphere. This climatic chaos came to an end roughly 15,000 years ago, with the conclusion of the last Ice Age. What followed was a period of remarkable and unusual climatic stability in which the world experienced not just warmer temperatures, but stable temperatures, stable rainfall patterns, and stable volcanic activity for prolonged stretches of time.13 Even this world had its share of sudden variations and cataclysmic natural disasters, of course, but at least people didn’t have to deal with most of Europe being buried under massive stacks of ice.

Humans almost certainly knew about the basic patterns of agriculture long before the establishment of fully developed agricultural civilizations. But for tens of thousands of years, that knowledge didn’t really matter because we couldn’t do much with it. The transition to agriculture was messy, complex, and unfolded over thousands of years; most hunters and gatherers actually resisted such a massive societal change at different points in time.14 But shifting ecological dynamics provided the possible operating space for agriculture to develop in the first place. The Ice Age had deprived people of the resources and ecological conditions to generate the necessary food surpluses for a sedentary community, which is not to claim that sedentary communities are what people wanted. Many of them were perfectly happy being nomadic. But even if they had wanted something else, nature was not exactly cooperative. In a topsy-turvy world of radical climatic shifts, counting on a sedentary location for a reliable food supply was just wishful thinking. A promising settlement in one year could be obliterated the next year by a powerful blizzard or some other natural disaster. People needed to have something far more regular and reliable to survive, much less to establish a successful agricultural community. Before the dawn of agriculture and the domestication of animals, the chaotic climate forced people to rely on food sources that were often seasonal and sporadic. As a result, widespread climatic instability probably delayed the emergence of energy-intensive civilization by tens of thousands of years, even though humans had all the physical and intellectual capacities required for its realization. By contrast, the climatic stability of the last 10,000 years has served as the central ecological primer that allowed human beings to form complex societies and civilizations. A predictable climate allowed people to grow food surpluses from the same settlement. It allowed them to organize their lives and communities without the fear that everything would collapse in a few years. In short, it afforded them the first real opportunity to move away from their previous nomadic existence, although this process came with its own set of problems and complications.

Nature evidently imposes powerful limits on what societies can do. Our civilizations are complex biophysical systems that interact with the wider natural world, and none can be fully examined apart from their underlying material conditions. The main goal of this book is to develop a better theoretical framework for thinking about the links and relations between the natural world and human economies, in the hope that such knowledge can help us plan for a better future and wrestle with the complex problems of the intensifying ecological crisis in our own age. A secondary goal is to offer a specific and alternative worldview for how to construct the broader energetic, economic, and political parameters of future societies. The dynamic structure of economics rests on physical and social conditions. Economies are complex dynamic systems in which people exchange energy with their natural ecozones and interact with one another as they produce, consume, and distribute goods and services. Economic systems can expand or contract depending on underlying ecological conditions, class structures, labor relations, social conflicts, and technological adaptations. These material factors are all locked in a vast web of chaotic interdependence, meaning that changes in one factor can produce highly complex and non-linear effects on the other factors. Ecological conditions may retain their causal primacy in the long run, but social and economic factors can predominate over critical periods that constrain the distribution of financial wealth as well as the intensity of commercial and productive cycles.

I will call this basic summary the ecodynamic synthesis. It’s not meant to be a formal scientific theory with rigorously quantifiable predictions. The larger aim is instead to present a new worldview that is informed by various intellectual traditions, including scientific and economic sources. The ecodynamic synthesis borrows and combines insights from ecological science, network theory, complex systems theory, and various economic theories, chiefly Marxist and institutional theories. Underpinning the ecodynamic synthesis is a flow-cycle model involving nature and society. Nature is the flowing river and society is the rotating watermill. Dynamical flows are the collections of energy and materials that economies absorb from the natural world. Some portions of these flows are used for production and distribution, but others are lost as wasted energy that can lead to natural instabilities that are highly disruptive to the prevailing social order. Dynamical cycles, on the other hand, are the recurring collective modes of production and distribution facilitated by the flows. The flows constrain and facilitate the cycles by providing a resource base for energy extraction, but the cycles can also impact ecological flows through various chaotic feedback loops associated with the energy losses of our economic activities, and other non-linear perturbations.

Before getting into more details, I should add a few notes about the philosophical orientation of the book. First, it would be wrong to see the message of this work as anything approaching environmental determinism, a controversial term that’s difficult to define. It roughly means something like the belief that the economic, cultural, and political structures of human society are mostly, if not exclusively, caused by environmental factors and conditions, including the prevailing climate and the available resources in a particular place. Such a reductionist framework will inevitably leave out a lot of important details. The climatic conditions of the United States can never tell you why there are specifically 50 states in the country, or 435 members in the House of Representatives. The river systems and arable lands of the United States cannot explain why we need to file our tax returns with the Internal Revenue Service by April of every year, nor can they explain the specific tax rates, credits, and deductions that applied to us in those returns. There are many legal, institutional, and economic features of society that are simply emergent, that cannot be reduced to deeper scientific principles through naïve models of causation. However, it’s still true that the global ecosphere sets the broader constraints in which human societies and civilizations can successfully operate. And if these dynamical constraints experience profound and rapid disruptions, as they are in our age, then the natural world can indeed compel and pressure human civilization into reorganizing its emergent properties. Likewise, the emergent features of human society can also be helpful or disruptive to the rest of the natural world, depending on what we do. That’s the basic point of this book: nature and society can affect each other in highly complex ways, which means that we need to think carefully about what kind of society we want to have in the first place. To speak about “nature” and “society” might seem to imply that these two terms signify fundamentally different things. But isn’t society just a part of nature? Society is embedded in the natural world, as are all human bodies, feelings, thoughts, and ideas. In that case, why would we talk about them dialectically, as if they’re two separate things? The answer is simple. We need to distinguish between a system and its surroundings if we’re going to say anything empirically and scientifically useful about the system itself. Even though a car engine forms one part of a larger vehicle, and that vehicle is embedded in a particular environment, it’s still useful to distinguish between the engine as a system and the wider environment in which the engine needs to operate. Only then can we start making claims about the efficiency or the work output of the engine, among other things. If we want to understand the long-term evolution of human society, then we can only do so by exploring how society interacts with the rest of the natural world. Much of this exploration will be done from the perspective of physics, but it’s important to emphasize that this book is not just about physics. It’s also about the social, economic, political, and historical relations that inform the wider scientific discussions regarding humanity’s future on this planet. The goal is to aim for a holistic analysis that ties together different causal factors. Physics will be the centerpiece of this analysis, but the book will also relate these scientific ideas to other emergent features of human society.

Energy

The fate of all economic systems is written in their energy flows. The exchange of energy between different systems has a decisive influence on the order, phase, and stability of physical matter. Energy is typically defined as the ability to do work, but this definition is not ideal when considering things like radiation or quantum physics, so one can better think of it as any conserved state of motion, such as work or heat, that can be exchanged among different systems.15 Kinetic energy and potential energy are two of the most important forms of energy storage. The sum of these two quantities is known as mechanical energy.16 A truck speeding down the highway packs a good amount of kinetic energy, which is energy associated with motion. A boulder teetering at the edge of a cliff has great potential energy, or energy associated with position. If given a slight push, its potential energy transforms into kinetic energy under the influence of gravity, and off it goes. When physical systems interact, energy is converted into many different forms, but its total quantity always remains constant. The conservation of energy implies that the total output of all energy flows and transformations must equal the total input.

Energy flows among different systems represent the engine of the cosmos, and they happen everywhere, so often that we hardly notice them. Heat naturally flows from warmer to colder regions, hence our coffee cools in the morning. Particles move from high-pressure areas to low-pressure areas, and so the wind starts to howl. Water travels from regions of high potential energy to regions of low potential energy, making rivers flow. Electric charges journey from regions of high voltage to regions of low voltage, and thus currents are unleashed through conductors. The flow of energy through physical systems is one of the most common features of nature. As these examples show, energy flows require gradients, which are differences in temperature, pressure, density, or other factors. Without these gradients, nature would never deliver any net flows, all physical systems would remain in equilibrium, and the world would be very boring. Energy flows are also important because they can generate mechanical work, which is any macroscopic displacement in response to a force.17 Lifting a weight and kicking a ball are both examples of performing mechanical work on another system.

Energy flows are central to the existence of life. For example, living organisms rely on the conversion of chemical energy into thermal and mechanical energy for survival. Living things usually store chemical energy in large biomolecules called adenosine triphosphates. Every single day, life is hard at work producing vast quantities of ATP molecules from proton pumps powered by electrochemical gradients across cell membranes, a fancy way of saying that energy flows in our cells produce the things we need to survive. Proteins and enzymes then break down these biomolecules and exploit the resulting energy to perform vital biological tasks, such as the contraction of muscles or the replication of DNA. Another example is the radiation zooming out from the Sun, which is generally responsible for some of the most important energy flows in the circle of life. A long time ago, the ancestors of many special organisms learned how to harness the power of sunlight and other basic substances for survival. Today these organisms include plants and algae; biologists call them autotrophs.

In the energy pyramid of life, the autotrophs are the base because they sustain all other living things. Part of the energy they process and absorb makes its way to heterotrophs, the organisms that need the nutrients provided by autotrophs to function and survive. Herbivores, carnivores, and omnivores are all heterotrophs, ultimately reliant on plants and other autotrophs for their basic needs. Herbivores consume plants directly while carnivores evolved to consume herbivores as a way of acquiring the substances they need from plants. Nuclear fusion inside the Sun is what keeps the whole system going. A byproduct of nuclear fusion is radiation, packets of photons that wend their way through the Sun before being released into space. These photons carry tiny amounts of energy individually, but vast quantities collectively. As they strike the Earth, they excite electrons and other particles to higher energy states, setting off a series of reactions and energy flows that help to sustain the chemical and biological features of life. The Sun is the chief sustainer of all the relevant energy flows in the biosphere.

Entropy and Dissipation

Although energy flows can produce work, they rarely do so efficiently. Large macroscopic systems, like trucks or planets, lose or gain mechanical energy through their interactions with the external world. The lead actor in this grand drama is dissipation, defined as any process that partially reduces or entirely eliminates the available mechanical energy of a physical system, converting it into heat or other products.18 As they interact with the external environment, physical systems often lose mechanical energy over time through friction, diffusion, turbulence, vibrations, collisions, and other similar dissipative effects, all of which prevent any energy source from being converted entirely into mechanical work. The presence of dissipation makes it necessary for dynamical systems to regularly absorb energy from the external world, as a way of overcoming the energy losses associated with dissipative interactions. Car engines burn through gasoline to sustain the motion of the wheels, which need periodic bursts of energy to overcome the dissipative effects of friction. Another simple example of dissipation is the heat produced when we rapidly rub our hands together.

Dissipation is not limited to things like cars and machines. It can have stunning and profound effects on planetary and cosmic scales. For example, the friction between the tides and the surface of the Earth is gradually slowing down the rate of rotation of the entire planet. Around 600 million years ago, a day on Earth lasted about 22 hours, and 300 million years later it had gone to almost 23 hours. As the tides move across the crust, the resulting friction produces heat. Over the eons, the ultimate effect of this process has been to slow down the rotation of the planet, which loses rotational kinetic energy as it produces more heat. In the natural world, macroscopic energy flows are often accompanied by dissipative losses of one kind or another. Physical systems that can dissipate energy are capable of rich and complex interactions, making dissipation a central feature of the natural order. A world without dissipation, and without the interactions that make it possible, is difficult to imagine. Think about what would happen if friction suddenly disappeared from the world. People would be slipping and sliding everywhere. Our cars would be useless, as would the very idea of transportation, because wheels and other mechanical devices would lack any traction with the ground and other surfaces. You would never be able to hold hands with your loved ones. You could never rock your baby back and forth. Our bodies would rapidly deteriorate and lose their internal structure. There would be no basic concept of time and place, not in a traditional sense anyway. The world would be very alien and unrecognizable.

Dissipation is closely related to entropy, one of the most important concepts in thermodynamics. While energy measures the motion produced by physical systems, entropy tracks the way that energy is distributed in the natural world. Entropy has several standard definitions in physics, all of them essentially equivalent. One popular definition from classical thermodynamics states that entropy is the amount of heat energy per unit of temperature that becomes unavailable for mechanical work during a thermodynamic process.19 Another important definition comes from statistical physics, which looks at how the microscopic parts of nature can join to produce big, macroscopic results. In this statistical version, entropy is a measure of the various ways that the microscopic states of a larger system can be rearranged without changing that system.20 For a concrete example, think of a typical gas and a typical solid at equilibrium. Energy is distributed very differently in these two phases of matter. The gas has a higher entropy than the solid, because the former’s particles have far more possible energy configurations than the fixed atomic sites in solids and crystals, which have only a small range of energy configurations that will preserve their fundamental order.21 We should emphasize that the concept of entropy does not apply to a specific configuration of macroscopic matter, but rather applies as a constraint on the number of possible configurations that a macroscopic system can have at equilibrium.

Entropy has a profound connection to dissipation through one of the most important laws of thermodynamics, which states that heat flows can never be fully converted into work.22 Dissipative interactions ensure that physical systems always lose some energy as heat in any natural thermodynamic process, where friction and other similar effects are present. Real-world examples of these thermodynamic losses include emissions from car engines, electric currents encountering resistance, and interacting fluid layers experiencing viscosity. In thermodynamics, these phenomena are often considered irreversible. The continuous production of heat energy from irreversible phenomena gradually depletes the stock of mechanical energy that physical systems can exploit. According to the definition of entropy, depleting useful mechanical energy generally implies that entropy increases. Formally stated, the most important consequence of any irreversible process is to increase the combined entropy of a physical system and its surroundings. For an isolated system, entropy continues to rise until it reaches some maximum value, at which point the system settles into equilibrium. To clarify this last concept, imagine a red gas and a blue gas separated by a partition inside a sealed container. Removing the partition allows the two gases to mix together. The result would be a gas that looks purple, and that equilibrium configuration would represent the state of maximum entropy. We can also relate dissipation to the concept of entropy in statistical physics. The proliferation of heat energy through physical systems changes the motion of their molecules into something more random and dispersed, increasing the number of microstates that can represent the macroscopic properties of the system. In a broad sense, entropy can be seen as the tendency of nature to reconfigure energy states into distributions that dissipate mechanical energy.

The traditional description of entropy given above applies in the regime of equilibrium thermodynamics. But in the real world, physical systems rarely exist at fixed temperatures, in perfect states of equilibrium, or in total isolation from the rest of the universe. The field of non-equilibrium thermodynamics examines the properties of thermodynamic systems that operate sufficiently far from equilibrium, such as living organisms or exploding bombs. Non-equilibrium systems are the lifeblood of the universe; they make the world dynamic and unpredictable. Modern thermodynamics remains a work in progress, but it has been used successfully to study a broad spectrum of phenomena, including heat flows, interacting quantum gases, dissipative structures, and even the global climate.23 There is no universally accepted meaning of entropy in non-equilibrium conditions, but physicists have offered several proposals.24 All of them include time when analyzing thermodynamic interactions, allowing us to determine not just whether entropy goes up or down, but also how quickly or slowly physical systems can change on their path to equilibrium. The principles of modern thermodynamics are therefore essential in helping us understand the behavior of real-world systems, including life itself.

All life forms are engaged in the physical process of avoiding thermodynamic equilibrium with the rest of their environment by continuously dissipating energy. This was one of the key insights that the physicist Erwin Schrödinger suggested in the 1940s, when he used non-equilibrium thermodynamics to study the central features of biology.25 We may call this vital process the entropic imperative. All living organisms consume energy from an external environment, use it to fuel vital biochemical processes and interactions, and then dissipate most of the energy consumed back to the environment. The dissipation of energy to an external environment allows organisms to conserve the order and stability of their own biochemical systems. The essential functions of life critically depend on this entropic stability, including functions like digestion, respiration, cell division, and protein synthesis. What makes life unique as a physical system is the sheer variety of dissipation methods that it has developed, including the production of heat, the emission of gases, and the expulsion of waste. This sweeping capacity to dissipate energy is what helps life to sustain the entropic imperative. Indeed, the physicist Jeremy England has argued that physical systems in a heat bath flooded with large amounts of energy can tend to dissipate more energy.26 This “dissipation-driven adaptation” can lead to the spontaneous emergence of order, replication, and self-assembly among microscopic units of matter, providing a potential clue to the very dynamics of the origin of life.

Organisms also use the energy they consume to perform mechanical work by, for example, walking, running, climbing, or typing on a keyboard. Those organisms with access to many energy sources can do more work and dissipate more energy, satisfying the central conditions of life. Like all other biological organisms, humans consume resources from some external environment in order to survive, reproduce, and expand. The consumption of energy is crucial because it allows animals to perform mechanical work, which means that they can run away from predators, move to other areas in search of food, or build tools to solve different problems. In addition, the biological mechanisms that are central to life, like protein synthesis and cell division, all require energy conversions in order to unfold at a regular pace. And the energy that we consume from the external world is precisely what allows our cellular components to perform all of these critical tasks. Energy and work are measured in the same units, the standard one being the joule. One joule is roughly the amount of energy your hand uses to lift an ordinary apple up to your mouth. A food calorie, also known as a kilocalorie, is equal to about 4,184 joules. A central empirical focus of our future analysis will be the rate of energy use, mostly because we want to track how the exchange of energy between different systems can change over time, and how those changes then affect the structural organization of the systems themselves. The rate of energy use is known as power, but to avoid any confusion with the more familiar understanding of the word, we shall use terms like “power output” or “rate of energy conversion” whenever we talk about power. In addition, we want to remove population effects in our analysis by using the term “energy conversion” to mean per capita conversion, unless otherwise indicated.

Physics and Economics

The thermodynamic relationships among energy, entropy, and dissipation impose powerful constraints on the behavior and evolution of economic systems.27 From a thermodynamic perspective, economies function as non-equilibrium systems capable of rapidly dissipating energy to some external environment. All dynamical systems gain strength from some energy reservoir, reach peak intensity by absorbing a regular supply of energy, then unravel from internal and external changes that either disrupt vital energy flows or make it impossible to keep dissipating more energy. They can even experience long-term undulations by growing for some time, then shrinking, then growing again, before finally collapsing. Interactions between dynamical systems can produce highly chaotic results, but energy expansions and contractions are the core features of all dynamical systems. The energy consumed by all economic systems is either converted into electricity and mechanical work, along with the physical products derived from that work, or it’s simply wasted and dissipated to the environment. Economies that increase the amount of electricity and mechanical work they generate can usually produce more goods and services. Historically, electricity and mechanical work have comprised a relatively small fraction of total energy use in any economy; the vast majority of the energy consumed by all economies is routinely squandered to the environment as waste, dissipation, and other kinds of energy losses.

Throughout history, economic growth has depended heavily on people converting more energy from their natural environments and concentrating the resulting energy flows for the completion of specific tasks.28 When humans were hunters and foragers, the primary asset that performed mechanical work was the human muscle.29 Our muscles were great for running, walking, collecting fruits, and hunting animals, but they did not generate enough mechanical work for energy-intensive production. This nomadic way of life lasted for some 200,000 years before undergoing significant disruptions after the Ice Age. Over the next few millennia, new lakes and rivers were born from the receding glaciers. Flora and fauna rapidly multiplied; this bonanza of wild crops and animals made it possible for some groups of people to settle down on specific plots of land, leading to numerous pastoralist and agricultural strategies that would come to define much of our history. Agrarian economies relied heavily on cultivated plants and domesticated animals to help generate surpluses of food and other goods and resources. These agrarian modes of production and consumption dominated human societies for almost ten thousand years but were eventually replaced by a new economic system: capitalism.

Capitalism emerged and spread from a complex set of converging conditions: the extraction and development of energy-dense natural resources, waves of colonial expansion, intensive periods of industrialization and technological innovation, the proliferation of epidemic diseases, and genocidal campaigns against Indigenous populations accompanied by massive land theft. It was the most chaotic and violent transformation in human history. The global economy has since become an interconnected system of finance, computers, factories, vehicles, machines, wage labor, and much more. Creating and sustaining this system required a major upward transition in the rate of energy throughput from our natural environments. In our nomadic days, the daily rate of per capita energy consumption was around 5,000 kilocalories, perhaps even less.30 By 1850, per capita consumption in Britain had risen to roughly 80,000 kilocalories per day, and the rate has since ballooned to about 200,000 kilocalories a day among the most advanced energy-intensive economies, such as the United States.31 From a scientific perspective, the fundamental feature of all capitalist economies is a high rate of energy use focused on boosting productivity and economic growth. The collective deployment of machines, vehicles, and electronic devices requires the production of vast amounts of useful energy, which in turn allows people to make more stuff, travel farther distances, and lift heavier objects, among other tasks. Capitalism is far more energy-intensive than any previous economic system, and it has wrought unprecedented ecological consequences that may threaten its very existence.

We can think of capitalism, from a biophysical perspective, as a supercharged entroplex, a mega-dissipative system dumping massive amounts of gases, liquids, and solid waste into our natural environments. The biggest effect of this spasmic energetic release has been the degradation of the planetary ecosphere to a more entropic and chaotic state, with profound implications for the future of humanity. In 2004, a group of research scientists with the International Geosphere-Biosphere Programme summarized these enormous changes and the great acceleration that started in the middle of the twentieth century:

A profound transformation of Earth’s environment is now apparent, owing not to the great forces of nature or to extraterrestrial sources but to the numbers and activities of people—the phenomenon of global change. Begun centuries ago, this transformation has undergone a profound acceleration during the second half of the 20th century. During the last 100 years human population soared from little more than one to six billion and economic activity increased nearly 10-fold between 1950 and 2000. The world’s population is more tightly connected than ever before via globalization of economies and information flows. Half of Earth’s land surface has been domesticated for direct human use. Most of the world’s fisheries are fully or overexploited. The composition of the atmosphere—greenhouse gases, reactive gases, aerosol particles—is now significantly different than it was a century ago. The Earth is now in the midst of its sixth great extinction event. The evidence that these changes are affecting the basic functioning of the Earth System, particularly the climate, grows stronger every year. The magnitude and rates of human-driven changes to the global environment are in many cases unprecedented for at least the last half-million years.32

The increasing order of human civilization is producing increasing disorder in the broader ecosphere, and this growing imbalance, if allowed to continue, will inevitably cripple the stability of global civilization itself. The huge energy losses of our modern economies have become an energy reservoir for other dynamical systems in nature, such as viruses, bacteria, hurricanes, storms, wildfires, and algae blooms, among others.33 Paradoxically, the “useless” energy that human civilization dumps into the natural world powers the formation of other physical systems, and these systems collectively are forming a new ecological order that will be incompatible with the necessary conditions for sustainable human development. It remains uncertain how long we can sustain such an energy-intensive path, but there is no doubt that the fantasy of endless growth and easy profits cannot continue. Our main goal is to chart a new path that will allow humanity to thrive within the natural constraints and parameters of the planetary ecosphere.

Our Great Challenge

In the 1980s, the world had a major ecological problem. The emission of gases like chlorofluorocarbons (CFCs), widely used as coolants in refrigeration, had punched a giant hole in the ozone layer. If that hole continued to get bigger, it would mean that dangerous levels of ultraviolet radiation from the Sun would reach the surface of the Earth and wreak havoc on human health worldwide. The world’s nations got together and passed the Montreal Protocol in 1987, which gradually phased out the use of CFCs in refrigeration and other commercial or industrial uses. Even though it still hasn’t fully gone away, the hole in the ozone layer gradually became smaller over the next few decades. Overall, this episode is often cited as prime evidence of humanity’s collective ability to solve hard problems. Here were the countries of the world all coming together in the face of adversity and doing something for the common good. What’s often neglected in this feel-good story is that CFCs were then immediately replaced by substances like hydrofluorocarbons (HFCs), which are extremely potent greenhouse gases. Humanity essentially traded one problem for another: it got rid of CFCs and saved the ozone layer, but only at the cost of dumping vast amounts of HFCs into the atmosphere and causing a rapid acceleration in global warming.

The most dangerous recurring theme of human history is civilizational collapse and dysfunction caused by widespread ecological disruption. We are now entering an age of profound and unprecedented ecological crisis, a bionomic disruption that threatens not just the viability of our economic systems, but large portions of the planet’s biological fabric. Although global warming is a major long-term challenge, it’s just one of many fundamental problems that have been caused by the capitalist acceleration of the past two centuries. For just one notable example, recent studies have suggested that roughly 8 to 10 million people die prematurely every year from air pollution caused specifically by the combustion of fossil fuels.34 In addition to global warming, humanity has adversely affected the planetary ecosphere by polluting the world’s oceans and atmosphere, starting the sixth mass extinction event in our planet’s history, causing massive disruptions to the nitrogen and phosphorous cycles that sustain all plant growth, boosting the incidence of global pandemics through agricultural expansion, and expediting the shortage of critical natural resources, including drinking water, for much of the world’s population. To the extent that human civilization will buckle and bend over the next few centuries, it will be from the cumulative pressure of all these factors, not from any single factor by itself. But since our problems are largely framed in the context of climate change, the corresponding solutions are narrow and technical, amounting to little more than hoping for rapid technological change to replace fossil fuels with renewables like wind and solar. For this school of thought, the goal is simply to decarbonize the global economy and continue as if nothing else matters. This approach is misguided precisely because our ecological problems don’t boil down to one simple thing. Substituting fossil fuels with renewable energy sources is an important and admirable objective, but not if it’s done without thinking about the broader context of our ecological challenges.

The dominant narrative in this entire debate is the idea that humanity can overcome the ecological problems of late-stage capitalism through technological innovation.35 In this book I will make a very different argument. Technological innovation has produced more energy-intensive societies with destructive ecological consequences. Improving energy efficiency tends to lead to more energy use over time, not less. Having a good understanding of the physics of capitalism will provide a better foundation for radically changing the politics of capitalism. Instead of advocating for technological change, we should advocate for social and political transformation. We should change how power is wielded in society, how decisions are made, how labor is organized, how wealth is distributed. Revolutionary transformation is what we need to secure our future. The main thesis of this work can be stated as follows: if modern civilization is going to survive the upcoming millennium, then we need to thoroughly reconstitute and revolutionize the social and economic relations that currently govern our lives. There are no quick technological “fixes” for what we’re facing; technological changes can supplement a comprehensive solution to our bionomic dilemma, but the only realistic way to save civilization in the long run is to radically change how we organize our societies.

In the first four chapters, I will lay out some of the concepts, problems, and narratives that have been used to understand the interconnections between human economies and the natural world. In chapters 5 through 9, I’ll unveil a new theoretical framework for better understanding economics and our place in the natural world. In the last three chapters, I’ll use this framework to identify some of the possibilities that are worth pursuing in our future efforts to construct a genuinely post-capitalist world.

Among all the living organisms that have called this blue marble home, humans are a very recent species. In that short period of time, we have managed to become one of the most dominant life forms in the history of the planet, creating powerful civilizations with elaborate cultures, large populations, and extensive trade networks. We have been nomads and farmers, scientists and lawyers, nurses and doctors, welders and blacksmiths. Our achievements are both astonishing and unprecedented, but they also carry great risks. The economic and demographic growth of human civilization over the last ten thousand years has profoundly impacted natural ecosystems throughout the planet. Global civilization now stands at a critical tipping point that deserves closer scrutiny. The goal of this work is to explore the deep ecology of human life and existence: to develop a fundamental sense of how human society relates to the wider natural world and then to thoroughly examine the possible scenarios for the future of human civilization. If we are to have any hope of addressing the difficult challenges we face, we must begin by understanding them and appreciating their complexity. And then, we must act.

My understanding of imperialism has been shaped by many sources, including Marx and Lenin, but is chiefly influenced by the theorist Immanuel Wallerstein, whose world-systems approach divided the world into a core, periphery, and semi-periphery.2 For Wallerstein, the imperial core is defined by powerful nations with strong militaries and large corporations that extract value from the rest of the global economy by exploiting cheap labor and grabbing natural resources and raw materials for industrial manufacturing. In this sense, empire is conceptualized through its external extractive relations with other countries and civilizations, not its internal political systems and institutions.

The Roman Empire was an empire, but so is the United States, as it too has a rigid sphere of influence that it maintains through military and financial coercion. The United States has a larger military budget than the next five highest spenders, operates hundreds of military bases and installations around the world, extracts money from host nations to finance its military presence, places severe economic sanctions on those who challenge its authority, and leverages the adoption of the dollar as the world’s de facto reserve currency in order to borrow cheaply from other countries. Within this theoretical paradigm, it qualifies as an empire. Of course, empires have existed throughout history, but the biggest focus of this post will be Western imperialism in modern times since that has been the dominant core of the global system for centuries, although I should stress that it won’t be the exclusive focus, as various characters like the Ottoman Empire will also appear in the narrative. This critical Western core likely had the largest and most intense effect on global migration dynamics in human history, outside of natural disasters and ecological conditions, which are also absolutely essential for understanding global migration flows in history but will not be covered in this post.

When we talk about immigration and its underlying causes in popular discourse, we tend to overemphasize the motivations of those who are moving and the reasons for why they're going to a particular place, a kind of push-pull analysis that has become a staple in much of the literature and the media. They’re looking for work and economic opportunity. They wanted to flee from war. They didn’t want to live under a corrupt or tyrannical government. These are the kinds of ideas that people often repeat. Rarely do we bother to examine the structural factors that are driving these motivations. Why don’t they have economic opportunity at home? Why is their country perpetually mired in conflict and chaos? Who installed the corrupt government that’s depriving them of a decent life? In this post, I’m going to explain why this is the central framing we need to adopt if we ever hope to grapple with global migration dynamics in modern history, in addition to the downstream political and economic upheavals that these migrations may cause.

People don’t migrate simply because they want a better life. If that were the case, the entire world would be moving almost every other year. Modern populations move because they’re forced to do so by the broader imperatives and constraints of the global economy, which is controlled by dominant capital in the imperial core. Migration is the cost and result of imperial subjugation in whatever form it comes, be it through economic means or from outright warfare (see Figure 1). And even when the imperial core exploits the global labor arbitrage and finds cheap labor abroad, there always remain economic sectors in the imperial core itself that require a constant stream of cheap labor to remain profitable, like agriculture in the United States, a sector where half of all workers have traditionally been undocumented.3

Migration dynamics in the imperial core are important not simply because migrants provide a source of cheap labor for capital to exploit. Immigration can affect domestic politics and and economic institutions in profound ways. For capital, immigrants are both a source of cheap labor and a convenient scapegoat whenever it becomes necessary to divert and redirect popular working class anger away from the ruling classes. Housing is not expensive because of investors, monopolists, and bad regulations; it’s expensive because of all the illegal immigrants that came in! Wages for the working class aren’t low because capitalists are grabbing all the income; they’re low because we have to compete with illegal migrants! From the 19th century with the Chinese Exclusion Act to the present day with the likes of Trump and Farage, capitalists have always inflamed these kinds of tensions at critical moments of stress for the system. In that sense, migration provides a double function for capital accumulation: it provides cheap labor that keeps production costs down and it helps to divide and distract the working class, an essential precondition for ensuring the dominance of the capitalist class.

What I want to emphasize in this post is that the imperial and colonial policies of the Western core obviously shaped the migration dynamics of non-Western peoples, but also the migration dynamics of Westerners themselves, sometimes in ways that you might find surprising. We all know that places like Australia began as penal colonies where the British Empire sent domestic convicts for forced labor, but did you know that the Irish Famine was exacerbated by deliberate British neglect meant to heavily depopulate the island so the British aristocracy could grab more land? If you didn’t, then read on!

Modern Slavery and the Atlantic System

Human beings have practiced slavery for thousands of years around the world, but never to the scale observed in modern times, and never with such global consequences for the demographic composition of entire continents (see Figure 2). Slaves provided labor power to replace the massive demographic losses caused by the European arrival in the New World, as a result of recurring epidemics and large-scale warfare. Slaves also produced many of the critical raw materials, like cotton, which facilitated further economic development and industrialization back in Western Europe.4

The demographics of the African slave trade are shocking in their magnitude. The historian Patrick Manning estimated that roughly 12 million slaves from Africa were shipped to the Americas from 1500 to 1900.5 About 1.5 million of them died during the voyage itself, since conditions on the ships were absolutely horrific: cramped and dirty with disease roving around everywhere. That meant about 10.5 million people made it the other side of the Atlantic alive, although many of these individuals would end up dying within a few weeks and months of their arrival.

In addition to the African slave trade, Manning estimates that about 6 million slaves in total were sent to the Middle East, mostly to the Ottoman Empire, and to other parts of Asia. Some 8 million slaves were traded within Africa itself, and about 4 million Africans died as a result of slave raids, forced marches, famine, and disease. These numbers are all historical guesswork and the real figures could be much higher. But they highlight the global scale of modern slavery and the extent to which that scale substantially shifted the demographic composition of so many territories, colonies, and countries.

The Atlantic System controlled by the Europeans became the political and economic engine for this massive forced transfer of human beings. Europeans would bring manufactured goods like guns and textiles to Africa and trade those goods for enslaved people. The slaves were then shipped across the Atlantic to the Americas, where they worked under brutal conditions producing commodities and raw materials for European manufacturing, which then flowed back to Africa, completing the circle of evil and domination. Apologists for colonialism and imperialism overemphasize the role that Africans themselves played in seizing and kidnapping the slaves that they sold to the Europeans. But the only reason why this historical phase happened in the first place is because the Europeans provided the demand. Without that overwhelming demand, there would be no Atlantic slave trade. The key piece of evidence for this argument is that the scale of African slave raiding was sharply smaller before the arrival of Europeans in the 15th and 16th centuries. Historians have estimated that roughly 7 million African slaves were traded across the Sahara Desert from 800 to 1900, a period of over a thousand years.6 By contrast, Europeans smashed through that benchmark in just 400 years, as indicated by the numbers above.

Imperialism in Latin America: On Revolving Doors and Immigration Boomerangs

The history of US imperialism in Latin America is outrageously long and could cover multiple volumes to the Moon and back. Capitalism in the United States has always been a black hole sucking up the cheap labor of the Western Hemisphere, and the examples are legion, but nowhere is this dynamic clearer than with Mexico, which has always provided a steady labor supply for the needs of US capital. In the Bracero program from the 1940s to the 1960s, the United States brought in millions of Mexican workers to staff various economic sectors facing severe labor shortages during and after World War II. But US policy towards Mexican labor, and migrant labor in general, has always operated on a revolving door principle: bring them in when necessary, kick them out when politically convenient. The US government deported over a million Mexican workers in 1954 following a recession triggered by the end of the Korean War.7 This basic dynamic has repeated across time for all migrants who have come to the country, Mexican or not.

One of the lesser known effects of US imperialism, however, is that it inadvertently creates the conditions for migration through its aggressive imperial policies, a kind of boomerang effect that fully echoes the law of unintended consequences. By routinely sabotaging and destabilizing governments and economies across Latin America, the United States has ensured that the region remains impoverished, relatively deindustrialized, and structurally incapable of providing a decent life for its citizens. Unsurprisingly, the result of all this imperial devastation has been that millions of people in Latin America end up fleeing north in search of a better life, a recurring consequence in modern history.

In the 1950s, the United States overthrew Jacob Arbenz, the elected leader of Guatemala. Arbenz had been pursuing a modest policy of land reform and redistribution meant to give more productive agricultural land to poor peasants. But the United Fruit Company, the corrupt American monopolist that was in charge of the country, didn’t want anyone messing with that land, so it convinced Washington to stage a coup. For American officials in the Eisenhower administration, even marginal attempts at reform in Latin America smacked of communism, so naturally Arbenz had to be removed from power.8 Arbenz’s policies were reversed shortly after the coup, and the high-quality land which had been distributed to half a million peasants was suddenly back in the hands of wealthy elites and foreign corporations. That didn’t go down well with the Guatemalan working class, and eventually Guatemala descended into a civil war that would last almost three decades. The chaos and devastation caused by the civil war prompted some 200,000 Guatemalans to flee north and head to the United States by 1990.9 This wasn’t the first time that US imperialism in Latin America caused widespread migration and displacement, and it wouldn’t be the last.

In the 1980s, the United States funded and trained numerous right-wing armies, paramilitary groups, and insurgents across Latin America, fueling gruesome civil wars that collectively forced hundreds of thousands of people to flee to the United States itself. Nicaragua and El Salvador were both devastated by right-wing death squads and paramilitary bandits as the United States actively sabotaged left-wing forces and governments that it believed harbored communist tendencies.10 In the late 1970s, there were roughly 30,000 Salvadorans in the United States. By 1990, that number had soared to over 450,000, a stunning 1400% increase, as people fled the brutality of the Salvadoran Civil War in droves.11 Likewise, there were a few thousand Nicaraguans in the United States in the late 1970s, but that number had swelled to about 170,000 by 1990 as desperate civilians fled the violence for other safe havens in the Americas.

South America has not been spared either. Venezuela was in the news recently after the United States attacked the country in early January and kidnapped its sitting president. The decline of the Venezuelan state over the past decade is a classic example of neo-imperialism at work. When Maduro came to power after Chavez died, the Venezuelan economy was generally stable, though far from perfect. But in 2014, global oil prices crashed in response to the shale revolution in the United States, which produced a global supply glut, and declining demand from the European Union and most importantly from China, where Xi Jinping steered the country towards a direction of lower growth rates.12 Since Venezuela was heavily dependent on oil exports for its budget and for obtaining dollars, the falling oil prices really hurt the Venezuelan economy, sending it into recession. But it was US foreign policy that ultimately tipped Venezuela over the edge and caused a massive crisis.

When Trump first came to power in 2017, he proceeded to impose brutal sanctions on Venezuela that effectively tanked the country’s economy, leading to a full-blown depression. Venezuela’s foreign trade volume collapsed, and the country even had trouble obtaining critical life-saving medicine for its people. The sanctions alone are estimated to have directly and indirectly killed tens of thousands of Venezuelans in 2017 and 2018 alone.13 Facing few economic prospects at home, millions of Venezuelans eventually fled their country in the next few years. Since 2016, nearly a million of them would end up migrating to the United States, the same country that had caused the miserable economic conditions responsible for their initial departure.14

For the ruling classes in the United States, Latin America has always been seen as a source of cheap labor and cheap resources. At the same time, capital also recognizes that Latin American migrants provide a useful political foil in troubled economic times. As a result, there’s a perennial revolving door in migration dynamics for the United States, with immigrants allowed to come into the country in large numbers when economic times are good and then expelled and deported when economic times get rough and capital needs a way of assuaging popular working class anger, which is what’s happening nowadays with the Trump administration’s aggressive deportation push.

British Imperialism and Global Labor Transfers

From 1840 to 1940, over 100 million people migrated around the world.15 The rise of industrialization and the hyper-imperialism of the 19th century instigated massive global migration flows, as Europeans scrambled to snatch and extract all the natural resources and raw materials they could possibly find to fuel and grow their economies back home. Soaring demand for resources like rubber, tin, sugar, cotton, and guano produced massive population flows within Africa and Asia, but also outside them.

When the British Empire abolished slavery in the 1830s, Caribbean plantations faced a labor shortage and the imperial administration decided to grab cheap labor from China and India instead, so that resource extraction in Latin America and the Caribbean could continue without interruption.16 In effect, indentured servitude became the new slavery, as these workers were brutally exploited, paid like dirt, and often died early from disease and exhaustion. This global labor transfer became known as the “coolie trade” and led to millions of Asians being shipped, often under false pretenses, to the Western Hemisphere between 1834 and 1920. That’s the main reason why, to this day, there are large Chinese communities in places like Jamaica, the Dominican Republic, and Trinidad & Tobago. Likewise, over 1 million Indians were transported by the British to the Pacific and the Caribbean, altering the demographic and cultural landscape of many islands where they previously had no presence.17

The horrors and tragedies of British imperialism were not exclusively confined to colonial territories outside of Europe. In the middle of the nineteenth century, the Great Irish Famine killed roughly one million people and forced roughly two million others to emigrate, reducing Ireland’s overall population by about a quarter.18 In the popular imagination today, most people understand the Irish Famine as a tragic yet isolated disaster, an unfortunate one-off. But it’s not that hard to show, as many historians have done, that it was actually a catastrophic tragedy engineered by imperialism.

The famine started from a potato blight, which is an invasive kind of microbe, that devastated harvests across the island. Ireland had been struck by potato blights before, but it never experienced the kind of calamity it was about to endure. What made the famine truly apocalyptic was Ireland’s precarious position within the British imperial system. For the British elite, Ireland was just another economic vassal that could be readily squeezed for profits and resources.

The British aristocracy controlled vast stretches of grazing land across Ireland, which had been invaded and conquered during the Middle Ages. The cattle raised on this land provided a steady supply of beef for British consumers back on the mainland. But the expansion of grazing land for cows also forced many poor Irish people into smaller plots of land with dubious agricultural value. To compensate for decreased soil quality, Irish farmers increasingly turned to growing potatoes, a reliable and calorie-dense crop.19 The potato was originally grown by indigenous Americans, but European merchants eventually brought it back home and its popularity took off. When the potato blight struck in the 1840s and food prices started soaring, millions of poor peasants and farmers in Ireland were priced out of the market.

The British government responded to the unfolding tragedy through callous indifference. British officials in charge of the crisis believed that a market-based, laissez-faire approach would solve the problem. The theory went that the government did not need to intervene or provide much aid relief because the price mechanisms of the market would take care of everything. During similar crises in the past, the British had curtailed all food exports from Ireland as a way of keeping more food on the island. But they shamefully refused that option this time around, allowing vast amounts of food to depart the island.

As the situation grew worse in Ireland and domestic pressure in Britain intensified, the government reversed course and started to provide relief. But the relief was simply a Trojan horse to seize the land of poor farmers, exemplified by the so-called Gregory Clause in the 1847 Poor Law Amendment Act.20 In this legislation, the British government stipulated that Irish people who owned more than a quarter of an acre in land could not receive relief, like food and work. This provision, and others like it, forced hundreds of thousands of Irish people to sell their plots of land to corrupt landlords simply so they could receive basic relief to survive. After seizing the land, the landlords predictably started charging sky-high rent prices as a way of profiting from the tragedy. A major figure in charge of famine relief was Charles Trevelyan, the Assistant Secretary to the Treasury, who wrote the following about the Irish people and the ongoing famine:

The real evil with which we have to contend is not the physical evil of the Famine, but the moral evil of the selfish, perverse and turbulent character of the people.21

Trevelyan also wrote to Lord Monteagle that the famine resulted from the “direct stroke of an all-wise and all-powerful Providence.” In 1848, The Economist clamored for the “departure of the redundant population” as a necessary step for Ireland to become a more profitable part of the Empire.22 For the British elite, Ireland was just a large reserve for cattle grazing, and its people were a hindrance preventing the full exploitation of the land.

The Great Famine started off as a problem of nature, but near-genocidal levels of social engineering from the British ruling classes turned it into a human catastrophe. Indeed, many historians have actually called it a genocide, arguing that the British wanted poor Irish people to die in droves so that the aristocracy could seize more empty parcels of Irish land.23 The famine left a deep and bitter scar on Irish psychology, providing the base of anger and resentment that eventually culminated in a successful push for independence during the early twentieth century.

The consequences of Britain’s imperial past have persisted through to modern times. Today, politicians like Nigel Farage exploit immigration as a political issue, but there was a time when the British needed immigrants to keep their economy going. After World War II, Britain faced enormous labor scarcities, like many other countries. The Atlee government passed the British Nationality Act in 1948, which granted all imperial subjects the status of “Citizen of the United Kingdom and Colonies.”24 The Act allowed many British colonial subjects to come to the mainland so they could provide a stable labor supply, epitomized by the 1948 arrival of HMT Empire Windrush at the Port of Tilbury near London while carrying about 1,000 passengers. The new arrivals became known as the Windrush generation.

As Britain ran into economic headwinds in the 1960s and 1970s, the government changed its strategy towards curtailing immigration once again. We see here another major example of how immigration, like free trade or any other professed principle, has become a convenient political foil for the Western ruling classes. Immigrants are allowed in when economic times are good, but when capitalists screw up the global economy and leave the rest of us fighting for scraps, immigrants suddenly become the enemy and have to be banished at all costs. During economic recessions and the cyclical crises of capitalism, the political message from the ruling classes is always that immigrants need to go, not that capitalism needs to go, or at the very least not that it has to change in any real sense.

The Conservatives won the 1970 general election in large part on the promise to end and prevent mass migration from non-white groups. In the following year, the Heath government passed the Immigration Act of 1971, which further undid the reforms of 1948 and created a new “right of abode” for those who had parents or grandparents born in the UK.25 This new standard for so-called patrials created a racialized system, as most Commonwealth residents in places like Canada or Australia had an immediate ancestor born in the UK while most residents in the Caribbean, for example, did not. The 1971 act was the culmination of increasing economic anxiety in Britain during the late 1960s, epitomized by the notorious “Rivers of Blood” speech from MP Enoch Powell in 1968, when he warned of a racial war if Britain didn’t halt non-white immigrants from coming to the country.

The effects of the 1971 law, and a later immigration law passed in 1981, are still felt in so many different ways. For example, in 2018, the Windrush scandal exposed the bureaucratic incompetence of the British immigration system, as the 1971 law had changed the status of the Windrush generation but had not provided any documentation to substantiate their citizenship status.26 As a result, the British government under the Conservatives demanded proof-of-citizenship that many people from the Windrush generation simply could not provide, leading the government to actually expel many of these people from Britain. A massive PR scandal followed, which revealed the incompetence of the British state and the extent to which immigration has become a political football in modern British society.

Immigration in Britain today has become a major political issue for the same basic reasons it has done so in the United States: it’s a convenient distraction for the ruling classes to avoid talking about the failures of capitalism. But there’s another deep problem at play as well. Both Britain and the United States are declining powers in the new emerging global order characterized by increasing economic multilateralism. Their history of migration was deeply wrapped up in their history of imperialism. The fundamental reason why so many people from around the world ended up in places like Britain and the United States is because these countries were imperial hydras that needed a steady supply of cheap labor to keep their economic systems going. But now that they no longer control the global system the way they once did, their politicians and capitalists are increasingly turning against the very globalization that these countries themselves created. Free trade and immigration were apparently desirable when the West was on the top of the world, but now that things are heading south, these forces of the modern world have become targets on the right, a source of constant political controversy and a reliable political methodology for scaring the working class into making opportunistic and short-sighted decisions about its future. The biggest threat to the Western working class is not immigration, it’s imperialism and capitalism. Immigration dynamics are merely a side effect of imperial capital pulling the strings of the global economy.

Imperialism and Land: How Property Theft Affected Migration Dynamics

From the enclosures in England to the Land Code of 1858 in the Ottoman Empire, imperial powers have always adjusted, stolen, manipulated, and shifted property rights around as a way of controlling the labor power of their working classes. The proletarianization of the English peasant in the early modern period as a result of successive parliamentary enclosures, peaking in intensity from the mid-18th to the early 19th century, is a well-known and rigorously documented story.27 As a result, I will instead focus on a much more fascinating case study that’s less well-known in the Western world: the Ottoman Land Code of 1858.

In the 1830s, the Ottoman Empire instituted the policy of Tanzimat, meaning “reorganization,” as a way of attempting to modernize and catch up to the Western European powers. One of the most consequential reforms of the Tanzimat era was the Land Code of 1858, which fundamentally reshaped property rights across the Balkans and the Middle East, with its aftershocks still being felt today.28 Before modern times, property rights were often customary, especially for peasants and poor farmers. For example, a typical European farmer in the Middle Ages didn’t have a formal written deed saying that these two selions right here belonged to him. It was understood through social customs, local agreements, village discussions, and lord-vassal hierarchies that certain pieces of land could only be used by certain people. But the modern world, with its extensive legal standards and bureaucratic micromanagement, changed all that.

The Ottoman Land Code required peasants to register their land in exchange for a tapu, or deed. However, many peasants refused to do so precisely because they knew that formal registration would mean military conscription and heavy taxation. As a result, wealthy nobles from the cities rushed in and registered that land in their own names instead, in effect stealing land that had belonged to others for generations and centuries, even if these third-party registrations often happened by agreement with the peasants themselves. Many peasants became landless sharecroppers following this landgrab, and they were practically forced to relocate to major cities like Istanbul in search of wage labor.

These changes heavily impacted other peoples across the empire. Local migrations proliferated among nomadic groups like the Bedouins and the Kurds, which were forced to settle down on specific plots of land as a result of the Land Code.29 The Ottoman Empire also relocated millions of Circassian refugees fleeing from Russia throughout the Middle East, often using the excuse of “vacant” lands that were unregistered to anyone for the relocation process, even though these lands were most definitely being used by actual peasants engaged in seasonal grazing and subsistence farming.

In a very powerful sense, the Ottoman Land Code of 1858 was the original sin that turbocharged the modern Israeli-Palestinian struggle. For the reasons cited above, many indigenous Palestinian farmers and peasants refused to formally register traditional land in their own names and had it registered to wealthy coastal merchants instead. When Jewish and Zionist organizations later came to the region looking for land, they bought these properties directly from the merchants, completely bypassing the Palestinian farmers and increasingly shifting the distribution of land ownership towards the Jews over time, even if they only held a minority of the land (see Figure 3). For an infamous example, the Sursock family of Beirut sold roughly 240,000 dunams in the fertile Jezreel Valley to Zionist groups and organizations in the early 1920s, leading to the eviction of thousands of Palestinian peasants.30 This land transfer sparked one of the first major waves of internal Palestinian displacement, culminating with entire generations of landless Palestinian peasants and fueling the conditions of political hostility which exploded into outright war later in the 20th century.

The formula for the landgrabs of the Tanzimat was repeated throughout the world in the modern period. First, require peasants to provide written deeds and proofs of ownership that they simply didn’t have. Second, seize their land once they couldn’t provide the new arbitrary proof demanded by the authorities. And third, turn them into wage laborers, where they could either provide a source of cheap labor on large consolidated farms or in urban factories for industrial production. In Australia, the notorious doctrine of Terra Nullius was used by settlers to declare Aboriginal land as legally vacant, justifying mass violence and displacement on a continental scale.31 Similar kinds of sordid events materialized all around the world at the intersection of imperialism and indigenous land practices, often leading to organized genocides and massacres.

Conclusion

Mass migration is a costly result of empire, and its effects on political and economic life worldwide have been pervasive and multidimensional. Empires themselves produce migration to their core territories when it suits them, then they turn against it when it becomes momentarily inconvenient. Empires also scramble migration dynamics worldwide, causing people to leave ancient homelands and settle in new territories, sparking numerous wars and social conflicts along the way.

What political conclusions can we reach from this analysis? There are several.

For the left, it’s that any domestic social, political, or economic problems that may be caused by immigration are largely downstream of more fundamental geopolitical relations and hierarchies in the global system. This makes the importance of class and political struggle on a global basis all the more necessary, as local solutions to the problems caused by migration are mere bandages on a deep wound. And for the working class in any given Western nation, it’s important to emphasize political engagement, labor agitation, union building, and solidarity in the face of capital’s attempts to undermine its position by exploiting immigration as a Trojan Horse.

And there’s an important lesson for the right here as well. What the right needs to understand is that you can have a global empire or a homogeneous ethnostate, but you cannot have both. No powerful empire or superpower in history has ever managed such a feat, and the United States is not about to buck that trend. In modern times, empire necessarily leads to the mixing of populations because establishing and reinforcing an empire requires control over the world’s labor supply and the broader political and economic architecture of capital accumulation, including global reserve currencies and military bases.

For those who love humanity and want something better than the status quo, the ultimate imperative is clear: let’s end imperialism, and human beings will no longer be forced to uproot their lives and communities. None of this means, of course, that we shouldn’t encourage migration for other reasons beyond imperialism. Immigrants can provide numerous benefits for any society, including cultural diversity, entrepreneurial innovation, economic mobility, and political flexibility. And even if we do end imperialism and capitalism, their ultimate ecological impacts via global warming and other biophysical vectors may still cause large volumes of global migration for decades if not centuries to come. These are all very critical issues in their own way, though they are not the main focus of this post.

The central lesson we need to internalize is that, especially in modern times, the global movement of peoples is largely dictated by the structural demands of the imperial core. The political struggle against imperialism is therefore a necessity that will liberate the working classes of the entire world, not just from forced migration, but from all the heinous depredations of capitalism.

Supply and demand are two of the most important concepts in economics, but the way neoclassical economists often talk about them is downright delusional. In this post, I’ll explain what prices actually mean in relation to supply and demand, and how these phenomena are entangled with the broader social order.

Introduction: The Marshallian Scissors

It’s certainly true that supply and demand provide important constraints on economic activity and the dynamics of the nominal domain. If 95 percent of the world’s apples suddenly disappear tomorrow, it’s a safe bet that global apple prices will rise dramatically. In the critique that follows, I am not rejecting supply and demand as useful categories of economic analysis; I am specifically rejecting the way that these two concepts have been abused and manipulated within neoclassical theory itself. I have often employed supply and demand as explanatory concepts in my book, The Physics of Capitalism, and throughout my writings, both here on Substack and elsewhere. But I have only done so by placing these phenomena within broader causal and historical contexts.

What I’m attacking in this post is the naïve Newtonian picture in which supply and demand function like mechanical forces that pull and push prices in different directions. Too often in both academic and public discourse, it’s this silly picture that’s blindly repeated, and virtually all changes in prices, wages, and profits are automatically attributed to “supply and demand” like a Pavlovian reflex. This was especially the rationale du jour during the height of the Covid pandemic, when much of the media and many academic economists kept spitting out the phrase “supply and demand” for any unusual patterns in consumer goods prices, labor markets, real estate markets, or any other strange activity in the global economy, making hardly any further attempts at deeper explanations.1

We’ve arrived at a point where “supply and demand” has become a popular catchphrase, repeated just because everyone else seems to be saying it, but without any actual awareness of its meaning or implications. In reality, the nominal domain of prices, profits, and wages is heavily filtered and structured by intermediary social relations and various economic and political struggles over the distribution of economic resources.2 Prices don’t just move in tandem with changes in the biophysical world; their actual trajectories are heavily influenced by intervening social dynamics. In other words, prices are symbolic indicators of corporate power and class dynamics. The ruling classes establish the economic conditions that govern the world, and prices then emerge from these conditions as an artificial landscape.

In his 1890 work Principles of Economics, the American economist Alfred Marshall argued that prices are determined jointly by supply and demand, mocking people who favored one side or the other with a famous analogy about scissors:

We might as reasonably dispute whether it is the upper or the under blade of a pair of scissors that cuts a piece of paper, as whether value is governed by utility or cost of production.3

The Marshallian scissors have since become an endemic component of neoclassical theory, etched in the memory of every undergraduate through a gazillion charts showing supply and demand curves producing equilibrium prices. But this otherwise perfect analogy contains the seeds of its own destruction. Scissors don’t magically move themselves. Someone is there controlling them. Someone is there deciding how quickly or slowly to cut. Someone is deciding the lines and shapes that are being cut. And therein lies the crux of the problem. By heavily focusing on this seductive catchphrase, we miss the underlying forces that actually drive economic behavior.

The traditional neoclassical story tells us that prices are determined by supply and demand. The law of demand states that, all else being equal, prices are inversely related to the quantities demanded. If something becomes more expensive, people will want to purchase less of it. And the law of supply says almost the opposite: all else being equal, prices are directly related to the quantities supplied. If prices rise for a particular commodity, then producers will want to supply more of it, so they can make more money. The dynamic competition between these two economic levers is supposed to yield the “equilibrium” price. That’s the neoclassical story in a nutshell: scarce things are expensive and plentiful things are cheap. There are many good reasons to doubt whether this neoclassical story is universally true at an empirical level, especially for subsistence items like cheap food.4 Nevertheless, this won’t be the main focus of my criticism. There are deeper and more foundational issues at play in this discussion, and that’s where my focus will be instead.

Causation and Demarcation

The first major reason why neoclassical price theory is so vacuous has to do with causation. If changes in prices are caused by changes in supply and demand, then what’s causing changes in supply and demand? One cannot fall back on tropes about how there’s a feedback loop here, because real feedback loops are always causally embedded in a larger environment; they’re never isolated. For example, the water cycle can be seen as a series of feedback loops, but all of those feedbacks depend on gravity, chemistry, and so many other background conditions. It’s therefore not convincing to rely on circular reasoning and to claim that prices and supply and demand cause each other.

The basic truth is that supply and demand are incidental causes, at best. Price dynamics are fundamentally caused by social and political struggles over the distribution of economic resources. Supply and demand cycles are usually engineered by powerful individuals, corporations, and governments, as a strategy to yield the prices and profits they ultimately desire. There is perhaps no better example of this in American history than Standard Oil, an infamously corrupt corporation that frequently bribed entire state legislatures and toggled all the supply and demand levers imaginable as a way of manipulating prices and crushing competition.5

Gas prices in the Western world skyrocketed in 1974 because the OPEC cartel temporarily suspended oil shipments to Western countries as punishment for supporting Israel in the October War.6 The price of gas didn’t rise because a shortage magically and spontaneously appeared out of nowhere. It increased because of geopolitical dynamics, which caused the shortage in the first place. The De Beers cartel deliberately restricted the supply of diamonds in global markets throughout the twentieth century, leading to grotesque price inflation along the way. In some years, De Beers held back and warehoused up to 60% of its annual global diamond production from wholesale and retail markets.7 A typical neoclassical economist would look at the situation and blame a supply shortage for the high price of diamonds, just like they’d reflexively blame a supply shortage for the high price of anything. But there was only a supply shortage because De Beers artificially created one to cement its stranglehold over the diamond industry.

The examples go on and on. In the First Browser War, Microsoft deliberately gave away Internet Explorer for free by bundling it with the Windows OS, all in an effort to drive Netscape out of business.8 That strategy, along with other similar anti-competitive practices, eventually worked (see Figure 2). In recent times, meatpacking monopolists like JBS, Cargill, and Tyson have artificially reduced the number of cows they purchase for slaughter, driving thousands of independent ranchers out of business and using the artificial scarcity they created as a convenient scapegoat for sky-high beef prices at the grocery store.9 This is what pricing fundamentally reveals: the corporate landscape of power and domination under capitalism.

To think of supply and demand as omnipresent forces that somehow control prices behind the scenes is to ignore the active agency of powerful individuals and corporations in establishing the critical parameters of the nominal domain, from prices and wages to profits and interest rates. Imagine a child hitting a baseball that accidentally smashes the neighbor’s window and then has the nerve to tell his neighbor, “Well I didn’t break the window. The ball did.” In a silly and pedantic way, it’s of course true that the ball physically penetrating the window caused it to be smashed into pieces. But then again, it was the child who swung the bat and gave the ball its unfortunate trajectory, and this is the cause we ultimately care about. Neoclassical economists recycle ritualistic propaganda about supply and demand and the “invisible hand” of the market as a way of obscuring and marginalizing the critical factors, like social power and class domination, that collectively have a far more profound impact on the dynamics of the nominal domain.

Next, let’s turn to the problem of demarcation. It’s not easy to know how to define or measure supply and demand in specific circumstances. Take oil as an example. What is the global supply of oil? Is it all the oil on planet Earth? Is it all the oil in proven reserves? All the oil in commercial or strategic inventories? How about the finished oil products stored in refineries or marine terminals? Take housing as another example. What’s the supply of housing? Is it the stock of existing homes for sale? The stock of new homes for sale? Both combined? What about non-rental vacant properties? What about new finished homes held back for inventory? The fundamental reason why this issue matters is because we might find results that are consistent or inconsistent with the laws of supply and demand depending on how we define these terms and what we actually measure. If the global oil supply is all the untapped reserves on planet Earth, then oil extraction over time depletes that finite stock, implying that prices should get consistently and continuously higher, if the law of supply is right. But that’s not what we see; oil prices actually exhibit huge swings and variations over time.

Given these problems, economists typically understand the term supply to mean whatever is available for sale in a market. Business economists often differentiate between the stock, which is what the seller holds in reserve, versus the supply, which is the amount offered for sale in the market.10 This distinction is relevant because market supply is often poorly defined, and the available supply will depend on what we decide to include as part of the market in question. But more importantly, supply is often artificially constructed by those who have the power to do so. Capitalists routinely create artificial scarcity by controlling how many products show up in markets in the first place. As mentioned above, De Beers used to keep diamonds off the market so it could inflate diamond prices, and OPEC does the same thing with its production quotas on oil. Ticketmaster often sequesters upfront a large percentage of tickets for major events, leading to sharply higher prices on tickets sold by brokers, especially in secondary markets.11 Real estate developers have deliberately reduced and constrained the production of new single-family homes, allowing them to justify higher prices on the homes that are available for sale.12 Pharmaceutical companies in the United States use a variety of legal and political methods to limit available drug supplies, like evergreening and lifecycle management, thus pumping up prices for millions of people.13 Restricting supply is a time-honored capitalist tradition designed to make goods and services seem like they’re special and exclusive. For that reason, defining supply as just the stuff that’s available for sale in the market is often an ideological cover for the corrupt and self-interested decisions of elite capitalists.

Demand is even more notoriously difficult to define than supply. Here’s how the economist Susan Feigenbaum puts it: “The quantity of a good a person is willing and able to purchase at any given price during a specified time period . . . all other factors held constant.”14 Two other economists used a bit of mathematical caricature to define the concept: “Demand = Desire to Acquire + Willingness to Pay + Ability to Pay.”15 These definitions all sound great and intuitive, but it doesn’t take much thought to realize that they’re pseudoscientific pablum. How does an economist exactly measure someone’s desire or willingness to purchase something? The short answer is that they can’t; all they can do is measure what products were purchased, in what quantities, and the corresponding prices. Desire and willingness are neurobiological phenomena that cannot be accurately and consistently measured with our current technological systems. Speaking in these silly terms is part of the lazy neoclassical effort at explanations based on methodological individualism, the notion that economic behaviors and decisions are derived from the autonomous preferences that exist within individuals. But the reality is that people take plenty of economic actions that have absolutely nothing to do with their internal desires and personal preferences, simply because we’re social creatures whose actions are often influenced, and sometimes even forced or manipulated, by others.

Time and Equilibrium

As if these issues aren’t bad enough, there are also problems related to time and equilibrium. It’s possible to get conflicting conclusions about the laws of supply and demand depending on the time intervals under analysis. This issue points to a general problem in economics: a cherished empirical relation might hold for five years or so, then break down completely once we get out to twenty or thirty years. Alternatively, the relation in question might hold up pretty well over long periods of time, but could easily break down over short time intervals, such as weeks or months.

Another issue is the problem of stability and equilibrium. For any given price point, there exist an infinite number of supply and demand curves that could generate that price. Because it’s practically impossible to measure supply and demand, at least as they’re typically defined, it’s also practically impossible to identify which curves are responsible for generating any given feature of the nominal domain, from individual salaries to commodity prices. This is an especially important point, because even if we grant the neoclassicals all their wildest fantasies about supply and demand determining everything, the practical consequences of that admission are virtually irrelevant as there’s no way to empirically nail down the supply and demand curves to which specific markets are supposedly responding. And if we can’t do that, it becomes hard to make policy recommendations based on supply and demand considerations, since we wouldn’t know which supply and demand curves apply to households and businesses, both at a microeconomic level and at an aggregate level.

In the 1970s, the economists Hugo Sonnennschein, Rolf Mantel, and Gérard Debreu published a series of papers concerning the uniqueness and stability of general equilibrium in neoclassical economics.16 General equilibrium is a hypothetical macroeconomic state where aggregate supply is supposed to equal aggregate demand. Their work came in the context of earlier results from Debreu and the American economist Kenneth Arrow showing that general equilibrium could exist, but only under highly idealized assumptions that apply absolutely nowhere in the real world. The results of Sonnennschein, Mantel, and Debreu collectively became known as the “SMD theorem” after their last names. The SMD theorem states that general equilibrium, even if it exists, is neither stable nor unique. If an economy reaches a state of general equilibrium, it won’t be able to stay there. And what’s worse, there are multiple paths toward general equilibrium, opening up the problem of which path we should pursue. In short, the SMD theorem is a highly negative and deflationary result for neoclassical theory because it shows that even if you know the equilibrium prices that prevail in general equilibrium, that information cannot tell you anything about the underlying economy that actually produced those prices. In effect, there are many “microscopic configurations” that can produce the same state of general equilibrium. Later results from economists like Alan Kirman, Donald Saari, Donald Brown, Chris Shannon, and others have only strengthened and expanded the original conclusion.17

And finally, there’s the problem of interdependence. Neoclassical economists often think that supply and demand are autonomous and independent forces that jointly settle to some equilibrium state and determine the corresponding economic outcomes. That’s the fantasy. In the real world, however, supply and demand are not actually independent functions that magically settle on some equilibrium price. The two are fundamentally synergistic and interdependent precisely because governments and corporations usually try to control both levers when they plan for the future. Large corporations actively intervene to shape and influence the demand patterns of consumers through a bewildering array of strategies. For example, AI and semiconductor companies like NVIDIA have been recently adopting circular and vendor financing models to allow their customers to keep buying.18 In other words, they’re starting to artificially generate their own demand. More broadly, dominant corporations spend hundreds of billions every year on advertising to persuade people to buy junk they don’t need. They also spend vast sums of money to lobby politicians, to make campaign contributions, and in some cases to outright bribe lawmakers to get their desired legislative result. They exploit their monopoly power to knock out potential competitors and to erect barriers to market entry, thereby substantially narrowing and limiting the available market options for consumers.

These actions and patterns speak to a broader truth about capitalism: markets never bring about order by themselves, they are always constructed from pre-existing orders and social structures. Dominant corporations don’t just sit around waiting until consumers “like” their products; they ruthlessly manipulate the law and the political system to shape and corner the market as they wish. The notion that economic outcomes are the product of decentralized and distributed networks is a lazy and vapid fairy tale designed to excuse the failures of the status quo. All social and economic orders are constructed from the dynamic interplay of pre-existing class and power relations. But elites and capitalists are largely missing from neoclassical theories. In this fantasy world, it’s only consumers and regular people who make choices, like what house to purchase, where to go to college, or what business to start. The rich and the powerful apparently have no agency whatsoever. It’s not capitalists who set wages and prices; it’s the market, through spontaneous orders, invisible hands, and ghosts in the machine. In the neoclassical paradigm, the market has the same function as God did for elites in the Middle Ages: it’s a convenient deus ex machina for justifying and eternalizing the current structure of the world. In the neoclassical vision, the consequential decisions of powerful elites are reimagined and abstracted as mysterious market forces, both to legitimize the social impacts of those decisions and to obscure their true origins, making the resulting social order seem completely normal and natural instead of imposed and constructed.

In practice, it’s also quite difficult to empirically test most claims about supply and demand. Take demand as an example. Sure, it may be easy enough to notice simple correlations between rising prices and lower sales. But the “law” of demand has another part, its infamous ceteris paribus condition, “all else being equal.” This condition implies that quantities can change for many other reasons besides price fluctuations, and vice versa. The empirical problem for the economist is to show that any change along the demand curve was specifically caused by variations in prices, as opposed to the thousands of other factors that could have driven the exact same observation. But how exactly is one supposed to do that? How do you keep “all else” equal in any real-world situation? How does one account for all possible confounding factors and variables? There are numerous statistical methods in econometrics to deal with these kinds of issues, but absolutely none of them are foolproof, and neoclassical economists are infamous for creating juvenile models that explain nothing useful about the world. Unlike laws in the natural sciences, most of the claimed “laws” in economics are pseudoscientific gimmicks meant to rationalize the existing order.

Conclusion: The Social Construction of Supply and Demand

The ultimate constraints on supply always come from nature, but society can and does intervene in numerous ways to adjust the levels of supply and demand that are actually available. This intervention can occur through concerted government action, through strategic economic decisions from dominant corporations and other economic groups, through various kinds of class struggles, or through some dynamic combination of all these things.

The conservative British prime minister Margaret Thatcher once seriously suggested, in an effort to deprioritize the role of government in solving social problems, “There is no such thing as society. There are individual men and women, and there are families.”19 This is a bit like saying, “There is no such thing as a human being; there are only atoms and molecules.” And if you want to explain and understand human behavior, you must do so at the level of atomic and molecular interactions. Of course, even atoms and molecules are not the most basic constituents of nature. We can just keep the reduction going all the way through to strings and branes, if we care about consistency, assuming that those things exist. In parroting this nonsense, Thatcher failed to grasp that what constitutes a “thing” is not just its components, but also the interactions that underlie the components, which is why the “thing” is capable of changing in the first place. Individuals do not exist in a vacuum, sunbathing on their private islands away from everyone else. Society represents the mental abstractions and organized interactions that allow people to communicate and work together. It does exist, it does affect the distribution of goods and services to individuals, it does provide an extra lever of constraint on supply and demand, and it affects our lives in so many different ways. Economic outcomes affect individuals, but they are socially constructed based on the class-power dynamics of society.

A glib statement like “prices are determined by supply and demand” means nothing unless it considers the role that various social forces play in establishing and reinforcing those cycles. In the words of economist Mariana Mazzucato, “Prices and wages are often set by the powerful and paid by the weak.”20 Neoclassical economists largely ignore these hard realities, pretending that individual preferences somehow come out of thin air, divorced from the causal webs and structures of society and the natural world.

The Basic Thesis of Institutionalism

Institutionalism is part of a broader wave of theories and debates that seek to explain the rise and fall of nations, including why certain nations become much more powerful than others, from ancient empires to modern superpowers (see Figure 1). Two of the more prominent thinkers in the institutionalist camp have been the economists Daron Acemoglu and James Robinson. In their 2012 book, Why Nations Fail, Acemoglu and Robinson claim that “countries differ in their economic success because of their different institutions, the rules influencing how the economy works, and the incentives that motivate people.”1 According to them, “Inclusive economic institutions create inclusive markets, which not only give people freedom to pursue the vocations in life that best suit their talents but also provide a level playing field that gives them the opportunity to do so. Those who have good ideas will be able to start businesses, workers will tend to go to activities where their productivity is greater, and less efficient firms can be replaced by more efficient ones.”2

Acemoglu and Robinson are persuaded by the observation of the Austrian economist Joseph Schumpeter that creative destruction is a major driving force in modern capitalism, meaning that new ideas, firms, products, and technologies replace older ones that can no longer compete. Incidentally, Schumpeter actually admits to borrowing the idea from Karl Marx and Friedrich Engels, who argued in The Communist Manifesto that the recurring commercial crises of capitalism required the periodic destruction of older methods of industry and production to pave the way for newer methods and social relations in the next economic order.3 To their credit, Acemoglu and Robinson don’t just focus on economics, but also examine politics as central to the nature of social institutions, writing the following:

Politics surrounds institutions for the simple reason that while inclusive institutions may be good for the economic prosperity of a nation, some people or groups, such as the elite of the Communist Party of North Korea or the sugar planters of colonial Barbados, will be much better off by setting up institutions that are extractive. When there is conflict over institutions, what happens depends on which people or group wins out in the game of politics, who can get more support, obtain additional resources, and form more effective alliances. In short, who wins depends on the distribution of political power in society.4

Acemoglu and Robinson do acknowledge the importance of what they call “critical junctures” in history, such as wars and pandemics. However, they only see these critical junctures as important because of their ability to amplify preexisting small differences among institutional structures in different countries, not because these critical junctures can fundamentally affect the process of institutional formation and destruction in the first place. Their fundamental unit of explanation is and always remains the institution. According to Acemoglu and Robinson, societies can succeed by establishing inclusive economic and political institutions that encourage prosperity across the social spectrum. On the flipside, societies start failing when corrupt elites establish extractive institutions that siphon off wealth from the majority of the population and redirect it toward themselves.

This simple duality is the backdrop for their historical analysis, and they go through many historical examples in the book, from medieval Venice and Spanish America to England and North Korea. Their most important case study is the economic evolution of England from the Middle Ages to the Industrial Revolution. They hold that England’s economic rise boils down to the country supposedly developing inclusive political and economic constitutions that encouraged economic growth, market competition, and technological innovation. They cite various examples throughout English history, such as the signing of the Magna Carta in 1215 and the Glorious Revolution in 1688, for taking England toward a direction of political pluralism and away from a direction of political dictatorship. Over time, they argue, these inclusive and pluralistic institutions culminated in the Industrial Revolution, the ascendance of capitalism, and the birth of the modern world. As Acemoglu and Robinson sum up their views on the Industrial Revolution in England:

The Industrial Revolution started and made its biggest strides in England because of her uniquely inclusive economic institutions. These in turn were built on foundations laid by the inclusive political institutions brought about by the Glorious Revolution. It was the Glorious Revolution that strengthened and rationalized property rights, improved financial markets, undermined state-sanctioned monopolies in foreign trade, and removed the barriers to the expansion of industry.5

How should we judge the plausibility of the institutional theory? Are some nations rich because they have “inclusive” institutions and others poor because they’re governed by extractive elites? The short answer is no. Here are some of the theory’s fundamental flaws.

Where the Institutional School Goes Wrong: Venice as a Case Study

Responsive institutions and governments are obviously important for economic development. No one wants to live under a corrupt government. But institutions are merely a proximate cause of economic growth, not a fundamental one. They are dependent on a broader causal matrix of other material factors, many of which people don’t control at all.

Acemoglu and Robinson have no concrete explanation for what causes responsive or extractive institutions in the first place. They only pick up the story after institutional formation has already occurred. At best, institutional theories are provisional and incomplete, since they do not explain what’s driving the process of institutional formation. They merely look at effects and consequences, making no attempt to ground institutions in a wider causal setting. As a result, institutions in these theories function like theoretical danglers, convenient explanations, or a deus ex machina that just appears out of nowhere to save the show.

To understand why the institutional theory fails to explain anything relevant about the world, let’s look at one of the most prominent examples from the book: Venice in the Middle Ages. Acemoglu and Robinson argue that the initial wave of Venetian prosperity in the Middle Ages was largely driven by inclusive economic and political institutions, such as the commenda contracts that provided upward social mobility for young merchants.6 They posit that that the institutional closure of the Great Council in the late 13th century, known as La Serrata, marked the start of a downward spiral for Venice. From that point forward, only aristocratic elites with hereditary titles could elect the Doge; Acemoglu and Robinson then argue that these elites used their political power to enrich themselves at the expense of other Venetians. And that’s basically their story: the rise and fall of Venice was largely about internal and autonomous developments happening within the city-state, as if Venice existed in a vacuum relative to the outside world.

Their cloistered historical account manages to do the impossible: tell a story about the rise and fall of Venice without once mentioning the Fourth Crusade, the Byzantine Empire, or the Ottoman Empire. The economic dynamics of Venice in the Middle Ages were almost entirely driven by geopolitics. Writing a historical account about medieval Venice without mentioning the Fourth Crusade is like writing a history of the United States in the 20th century without mentioning World War II. The historian Jason Roche emphasizes the importance of the Fourth Crusade very clearly: “Venice’s prosperity in the thirteenth century was largely a product of acquiring Byzantine trading posts in the wake of the conquest of Constantinople during the Fourth Crusade.”7 The Fourth Crusade was the seminal geopolitical event of Venetian history, because the subsequent partitioning of the Byzantine Empire left Venice in control of Crete and other Byzantine territories, thus greatly enhancing its economic and military power. Thousands of Venetians would emigrate to Crete in the 13th century, lured by the prospect of huge land holdings, and the island was a major source of lucrative products that boosted Venetian trade.

The rise of imperial Venice is precisely one of the biggest reasons why the Republic’s political system became insular: because imperial expansion usually breeds greater concentration of political power in elite hands. Acemoglu and Robinson buttress their argument of institutional decline with a throwaway statistic about the population of Venice dropping to 100,000 by the year 1500. This number is highly misleading, first because it ignores that Venice was hit by recurring epidemics that killed many people and forced others to leave the city for different regions within the empire. And the growth of the empire itself is the biggest reason why the city’s population decline is irrelevant. By the 16th century, Venice was an imperial powerhouse that dominated Mediterranean trade (see Figure 2). The total population of the Venetian empire in 1560 stood at around 2.3 million people, compared to about 3 million for England in the same year.8 Venice the empire had a far bigger population by 1560 than it had in 1200. Venice the city-state was the linchpin of the entire imperial system, but it would be wrong to think of Venice in the Late Middle Ages as being just a city-state. Third, plenty of other European cities experienced massive population declines over the same period. London had 55,000 people in 1520, compared to 100,000 before the Black Death.9 Was that decline also due to institutional failure? If so, there seems to have been crippling institutional failure across Europe during the Late Middle Ages. That’s one explanation; a better one is that it simply took Europe a long time to recover from the demographic calamities wrought by the Black Death and subsequent outbreaks of bubonic plague and other epidemics.

The dominant reasons for Venetian decline in the 16th and 17th centuries can be safely attributed to increasingly ruinous and expensive wars against the Ottoman Empire. Venice lost Crete, Cyprus, and many other important colonial territories in these wars. Although certainly other factors also played a role in the empire’s slide towards irrelevance, wars and geopolitical considerations are by far the main culprits.

War and Politics as Drivers of Institutional Formation

Geopolitical dynamics are a fundamental driver of institutional dynamics. Political systems can change in major ways as they experience wars, conquests, rebellions, and other violent events. British history provides many notable examples of this powerful trend. For example, Parliament cemented its dominance over the monarchy largely as a result of the English Civil War, a brutal conflict which left behind a streak of war crimes, a beheaded king, and 200,000 people dead.

The creation of the Magna Carta is another powerful example of geopolitical dynamics driving institutional reorganization. Acemoglu and Robinson call the Magna Carta “a hesitant step towards pluralism” that they ascribe to the barons standing up to the English king, without ever explaining to the reader exactly where the barons found this incredible courage.10 The reality is that King John only signed the Magna Carta in 1215 because he had lost a disastrous war against the French in the previous year. That defeat became the latest in a string of English military blunders stretching back over a decade, leading to the French conquest of Normandy. The wars were extremely expensive and John had raised taxes to help pay for them. Angry that he’d squandered their time and money on a losing effort, English barons revolted and forced John to sign a document in which he ceded most of his control over the English economy.

This critical history is ignored by Acemoglu and Robinson, and it’s also ignored in most conservative readings of English history, which are so obsessed with emphasizing gradual transformation that they miss the violent and revolutionary moments that set the stage for the later periods of political and economic stability. Another glaring omission in the book is England’s loss in the Hundred Years War, which was actually a series of distinct but related wars that lasted more than a hundred years. Defeat against the French permanently turned the English people into a community of islanders. It’s a notable example of when losing a war can actually be a good thing over the long run. Although the defeat eliminated England’s continental possessions, it also reoriented the country’s strategic priorities towards naval expansion. British leaders gradually relegated the army in importance while heavily investing in the Royal Navy, which eventually became the world’s most powerful.

Another big problem with attributing England’s rise to institutional progress is that much of England’s economic advantage was already in place long before the Glorious Revolution of 1688. London’s population skyrocketed from the 55,000 in 1520 to 475,000 in 1670.11 England’s wool industry also displaced the Flemish and the Italians as the dominant European producer long before 1688.12

Ecological Dynamics and Institutions

The fundamental reason why Western Europe became rich first is because of highly favorable ecological dynamics, resulting from the convergence of various geographical, hydrological, geological, agricultural, epidemiological, and other conditions. Acemoglu and Robinson reject ecological explanations of historical development, focusing their fire on Jared Diamond specifically, on the basis of a supreme strawman fallacy: that only the distribution of flaura and fauna matters for economic development. For example, they argue that many crops and seed varieties critical for agricultural development were found throughout the Eurasian landmass, along with many large mammals that could be domesticated and used for production and transportation.13 Since all major civilizations in Eurasia had access to these natural resources, they should have followed very similar patterns of development, and yet that’s not what happened. That’s their argument in a nutshell, but it’s based on a highly selective reading of what Jared Diamond actually wrote in Guns, Germs, and Steel, nor is it representative of what ecologists believe in general.

Diamond was clear in the beginning of his book that he wanted to explain variations between Eurasia and other continents, a point that Acemoglu and Robinson do acknowledge, not variations within Eurasia itself.14 And ecologists recognize that economic development can be affected by many different factors acting separately or concurrently in space and time, including viruses, bacteria, volcanoes, earthquakes, droughts, hurricanes, climatic variations, rainfall patterns, soil quality, and a million other things. Furthermore, complex feedback loops often develop between human societies and the natural world. We’re seeing them in action right now with global warming, ocean acidification, water shortages, mass extinctions, air pollution, toxic pollution, and other related phenomena. To think that ecology is not the primary driver of human history is to live in denial. Ecological changes have been strongly implicated in the rise and fall of civilizations throughout history by dozens of different major studies.15 Examples include the Late Bronze Age collapse, caused primarily by drought and earthquakes, the Classic Maya collapse, caused chiefly though not entirely by drought, and the Germanic migrations into the Roman Empire, caused chiefly by a cooler climate in Northern Europe starting around the year 200.

In his book, Diamond made a useful distinction between fundamental and proximate causes. Institutional explanations of economic development are basically proximate causes; they’re provisional explanations contingent on specific environmental constraints. The fundamental causes are the material and ecological factors lurking underneath. The Industrial Revolution started in England largely because England had highly favorable conditions for initiating energy-intensive industrialization, as I explained in great detail in this post from earlier in the year. These advantageous conditions included vast and easily accessible coal deposits, high wages that gave an impetus for developing labor-saving technologies, and huge quantities of raw materials from international trade, among other critical factors. It doesn’t matter how many “institutional changes” the monarchs of Hawaii could make; their society would have never been the first to industrialize under the conditions it experienced. Of course, a new economic or technological process can start somewhere in the world and then be adopted by others; that’s totally consistent with ecological theories of economic development. People often bring up Japan as an example that ecology isn’t everything, given how rapidly Japan modernized and industrialized despite various ecological scarcities. But this argument is a red herring because Japan modernized by learning from others. The central point is that certain ecological and material factors are necessary for the initial development of a new economic trajectory, but once that trajectory gets going, it’s totally normal to expect that it might spread to other places around the world. That’s why we have scientific research stations in Antarctica right now. But the Industrial Revolution could have never started in Antarctica!

The institutional dynamics of Latin American history since the Spanish conquests also critically depended on exogenous material factors. Spain conquered the most densely populated areas of the Americas, so it had an incentive to extract as much de facto slave labor as possible from the people who survived the horrible pandemics that brought down pre-European civilizations. By contrast, much of English colonial development in North America was driven by English immigrants working as indentured servants. Consider that 350,000 English people migrated to North America in the 17th century alone, a huge figure that vastly dwarfs Spanish emigration rates.16 All of this happened because North America was sparsely populated by the early 17th century, thanks to decades of European pandemics working their way through indigenous populations. The big idea is that the English relied mostly on their own labor force in the first century of North American colonization, whereas the Spanish didn’t. Slaves didn’t start coming in huge numbers in colonial America until the 18th century; there were about 27,000 in 1700, just 10% of the population.17 The ecological and demographic dynamics of North America meant that the English had basically no choice but to export more “inclusive” institutions, whereas the Spanish adopted methods of violent extraction because that was the most convenient approach, given the vast labor supply at their disposal.

Concluding Thoughts

The notion that extractive elites always sabotage or disrupt economic and technological innovation to protect their power is too reductionist. It’s true that corrupt elites can sometimes eliminate new technologies that could threaten the base of their economic power. Just look at how quickly General Motors literally crushed the EV1 electric vehicle in the 1990s. Acemoglu and Robinson give other good examples in their book. But it’s also true that corrupt elites can aggressively invest in new technologies as a way of augmenting their power over other groups and nations. That’s why the Mongols in the thirteenth century eagerly borrowed and adopted Chinese military technologies on their way to conquering most of Eurasia, including China itself.

Technology can be used to create or to destroy; it has no inherent teleological direction. Human beings are the ones who must decide how it should be used, and usually those decisions are made by humans who have enormous social power. In 1909, the German chemist Fritz Haber invented a method for turning atmospheric nitrogen into ammonia, a critical fertilizer for plant growth. The resulting industrialization of the process by Carl Bosch, and its adoption by countries and companies outside of Germany, substantially boosted agricultural production around the world in the next few decades. Vaclav Smil estimated that 4 billion people are alive now simply because of this invention, although this figure is inaccurate in many ways, and roughly half of the nitrogen atoms in your body come from the Haber-Bosch process.18 At the same time, ammonium nitrate is a powerful explosive, and the German government extensively funded its production during World War I. The Germans, like other countries, also heavily funded the production of chemical weapons that were used to perpetrate war crimes in World War I, and also to expedite the genocidal slaughter of Jews and other groups in the Holocaust during World War II. If they can get away with it, elites will usually try to supplant and expropriate new technologies for the purpose of reinforcing and amplifying their class domination of society. The imposition of political and strategic power is a complex process; sometimes it can involve the destruction of disruptive technologies while other times it can lead to their proliferation.

In their obsession with defending the Western status quo, Acemoglu and Robinson also fail to recognize that plenty of civilizations in the past have experienced long-term economic growth with different institutional models. China under the Song dynasty is a great example. In 999, Chinese tax officials calculated total iron output in Song China to be roughly 32,500 tons, but by 1078 that number had risen to an astonishing 125,000 tons, a four-fold increase.19 The Chinese initially produced iron for agricultural tools and later started focusing more on weapons. In addition to gunpowder, they also produced crossbows and arrows. At its height, Song China may have manufactured about 16 million iron arrowheads every year. Kent Deng writes about the phenomenal performance of the coal industry in Song China by emphasizing its enormous scale: “China’s iron output increased about six-fold from 806 to 1078 with an annual growth rate of 0.7 percent…Song China probably produced the highest iron output per capita in the world until 1750.”20 Historian Arnold Pacey summed up Song China’s technological development by writing that, by the year 1100, “China was undoubtedly the most technically ‘advanced’ region in the world, particularly with regard to the use of coke in iron smelting, canal transport, and farm implements…In all these fields, there were techniques in use in eleventh-century China which had no parallel in Europe until around 1700.”21

Western-style institutions are not necessary for economic growth and development, and indeed the rise of China in modern times has shown that alternative developmental paths are possible, outside the vapid neoliberal formula of austerity, privatization, deindustrialization, and financialization that is currently dragging down most of the Western world. Human history is dynamic and complex, with multiple causal factors converging across time and space to produce a plethora of wide-ranging outcomes. Effective institutions can certainly be part of this story, but never the central core.

In general, what united most Neoplatonist thinkers was the belief in a higher, supernatural, and transcendental reality which then emanated and overflowed into lower orders of being, like the world of the senses that we apprehend in our daily lives. The ultimate spiritual goal of this philosophical project was to somehow understand or enter into union with this higher transcendent order. This post is meant to grapple with some of the fundamental concepts of Neoplatonism using a critical materialist lens.

In his influential work The Enneads, the ancient philosopher Plotinus explored the fundamental nature of being from a Platonic perspective (see Figure 1). One of the major things he wanted to explain is why the celestial order is supposedly permanent while things here on Earth are fleeting, coming and going all the time. “The reason,” Plotinus says, “is given by Plato,” elaborating as follows:

The celestial order is from God, the living things of earth from the gods sprung from God; and it is law that the offspring of God endures. In other words, the celestial soul…springs directly next from the Creator, while the animal life of this earth is produced by an image which goes forth from that celestial soul and may be said to flow downwards from it.

Unlike what Plotinus and most ancient philosophers believed, stars and other celestial objects are not eternal. They came into being at some point in the distant past and they will end at some point in the distant future as their nuclear fuel runs out. In the grand scheme of things, they too are fleeting, like everything here on Earth. The errors found in this passage are another excellent reminder about the dangers of transcendental thinking. Plotinus is dividing the world into different levels of being, with the higher levels of the hierarchy creating and sustaining the lower levels. According to Plotinus, the higher levels emanate a portion of their existence to the lower levels, much like the Sun gives off light supposedly without diminishing itself. The big idea is that the lower levels flow from the higher levels in a process that unfolds outside of time.

Introduction to The One

The highest level of reality, the pure and perfect source of all being, is The One. Plotinus argues that the rest of reality flows from The One, which first emanates the Divine Mind, which then produces the World Soul, which finally creates the natural world. Inspired by the mysticism of various cults and religions in the Roman Empire, Plotinus conceptualizes our world, the world of the senses, as a kind of imperfect image of a higher archetype, that stable and eternal realm dominated by the divine and intellectual principles that supposedly govern all of reality.

For Plotinus, The One is simple, indivisible, and has no parts. All other things in reality get their identity by somehow participating in The One. Plotinus writes as follows:

All beings are beings by The One…For if things are deprived of The One which is predicated of them, they are not those things. For an army does not exist if it is not one, nor a chorus or a flock if they are not one. But neither can a house or a ship exist if they do not have their one, since the house is one and so is the ship...continuous magnitudes, if The One was not with them, would not exist…if they are cut up, they change their being in proportion as they lose their one. And again, the bodies of plants and animals, each of which is one, if they escape their one by being broken up into a multiplicity, lose the substance which they had and are no longer what they were but have become other things, and are those other things insofar as each of them is one.

Notice how Plotinus conceives of houses and ships almost as strict metaphysical categories, as stable concepts that are grounded in The One. It’s unclear what Plotinus is talking about when he suggests that animals and plants “lose their substance” as they are broken up into multiple parts. If we remove some hair from a dog, does that mean that we no longer have a dog? If we remove some leaves from a tree, does that mean that the tree is no longer a tree? You can extend this question to other things as well. If we remove a few shingles from the roof of a house, does that mean that the house is no longer a house? You might argue that shingles and leaves are “accidental” components of houses and trees, not essential parts. But it doesn’t seem like The One can give us any clear indication about where to draw the dividing line. Which parts, and how many of them, can someone remove from a house before that house becomes something other than a house? The One cannot say.

Modern science poses other problems for Plotinus and his claim that things “change their being in proportion as they lose their one.” For example, a bar magnet can be divided into multiple pieces and all the resulting fragments remain magnetic, a property that surely must count as “essential” to the Platonists. The implication is that something can lose its “oneness” and yet still remain one. Multiplicity does not abolish identity; it simply gives it a richer variety. It’s totally fine for houses to be different from each other and still count as houses, precisely because houses are not hard or fundamental ontological categories. They’re products of human society, and the words that we have for them are merely useful social conventions for facilitating interpersonal communication and coordinating social activity.

Emanation and Matter

There are other problems for Plotinus as well. The One is supposed to be the source for all the different levels and degrees of reality, so he imbues it with extraordinary creative powers. When explaining how The One emanates the Divine Mind, for example, Plotinus flagrantly declares that the process “may be compared to the brilliant light encircling the Sun and ceaselessly generated from that unchanging substance.” It’s an audacious analogy, in keeping with Platonic tradition. Plotinus resorts to this desperate analogy as a way of solving a basic philosophical problem. The One is supposed to remain constant and unchanging as it emanates its being to the lower degrees of reality. But how does one even begin to make sense of such an idea? Wherever we look around our world, things or objects that give off a part of themselves become diminished in one way or another.

To solve this problem, Plotinus makes use of an empirical analogy from the world of the senses that he evidently believed could beat conventional wisdom. Like other analogies lost in the murky dungeons of Platonism, however, this one is completely faulty and useless because the Sun is actually a physical system that does diminish when it gives off light. It doesn’t diminish in the sense of always acquiring a smaller volume. Quite the opposite is true: the volume of the Sun is currently expanding and will continue to do so for a few more billion years, before our stellar protector finally becomes a miniscule shell of its former self as it reaches a volume roughly the size of the Earth. After another vast stretch of time, it will disappear forever. The Sun is being continuously diminished in the sense that it’s using up nuclear fuel whenever it produces light, and therefore running out of the very thing it needs to continue radiating. If we examine the other side of the analogy, the implication is that The One should also become diminished as its existence overflows to the lower levels. The One is not so fundamental after all.

Contemporary supporters of Plato and Plotinus might respond by saying that the analogy is not essential for the overall argument; the argument stands even without the analogy. Perhaps it does, but it’s difficult to see how. That’s precisely the point of the analogy: to explain what is otherwise unexplainable. Does it shock anyone to know that the Platonic corpus is overflowing with words like illumination, radiation, and emanation? When the Platonists are making some of their biggest metaphysical arguments, whether it’s Plato himself in The Republic or Plotinus in The Enneads, they’re implicitly relying on the empirical analogy about sunlight to make the substance of those arguments intelligible to their readers. The Platonists are trying to have it both ways. They’re leaning very heavily into an empirical analogy from the world of the senses, but they don’t want to accept the consequences that could follow from such reckless adventures. Transcendental principles do not appear to be so perfect and divine if they’re not permanent. In fact, they begin to look more and more like the ordinary things in the material world, the world of the senses.

Emanation is a transcendental principle in the sense that it makes certain categories of being more essential and superior than other categories of being. It sets up various hierarchies and relationships between different levels of being. In attempting to construct one of these hierarchies, Plotinus argues that matter comes in two different kinds that exist in two separate realms. One type of matter exists in the “intellectual realm” that lies beyond the world of the senses. The other type of matter is the corrupt and formless junk that we see in our daily lives. Commenting on this second kind in The Enneads, Plotinus claims that “in matter we have no mere absence of means or of strength; it is utter destitution…of sense, of virtue, of beauty, of pattern, of Ideal principle, of quality. This is surely ugliness, utter disgracefulness, unredeemed evil.” That’s quite an insult. For the “higher” type of matter, he goes on to say that “the matter in the intellectual realm is an existent, for there is nothing previous to it except beyond-existence; but what precedes the matter of this sphere is existence; by its alienation in regard to the beauty and good of existence, matter is therefore a non-existent.”

Why does Plotinus hold such a bleak and moralistic view about matter, to the point of calling it evil and suggesting that it doesn’t even exist at our level of reality? Because like Plato and Aristotle before him, Plotinus believes that matter is something like a passive receptacle, a lazy ingredient, a “permanent substratum,” as he calls it. But he goes much further than either Plato or Aristotle in denouncing the value of matter. It’s evident from the tone of his text that he doesn’t even want to talk about this useless pile of junk, but feels like he needs to do it as a way of setting the record straight. His raw hatred for the idea that his mind entertains about matter is painfully obvious. He even gets mad at the Epicureans for suggesting that the gods are composed of atoms and are subject to the laws of nature: “the school even has the audacity to foist matter upon the divine beings so that, finally, God himself becomes a mode of matter.” Once he gets to his arguments about the identity of matter, Plotinus claims that “matter must be of later origin than the forming power, and therefore must be at its disposition throughout, ready to become anything.” He also claims that “all that ever appears upon [matter] is brought in by the Idea.” Beyond lazy appeals to words like power and participation, Plotinus never actually explains how the ideas and principles can organize matter in the first place. He has several interaction problems that he ignores, similar to the issues that plague Aristotelian hylomorphism, although those issues won’t be my main focus here.

The position adopted by Plotinus implies that matter lacks any and all properties, including size and extension. But this argument leads to a problem that Plotinus himself recognizes: “how can I form the conception of the absence of size in matter? How do you form the concept of any absence of quality?” In the same vein, he asks “how can we conceive of a thing having existence without having magnitude?” These are all the right questions, to which Plotinus proceeds to give all the wrong answers. He responds that “the secret is indetermination.” Plotinus speculates that the indeterminate is not simply a mental act of negation, employing another empirical analogy along the way:

The indeterminate has some footing in the sphere of affirmation. The eye is aware of darkness as a base capable of receiving any color not yet seen against it: so the mind, putting aside all attributes perceptible to sense, all that corresponds to light, comes upon a receptacle which it cannot bring under determination: it is thus in the state of the eye which, when directed towards darkness, has become in some way identical with the object of its spurious vision.

The problem with this analogy is that darkness only appears intelligible to the eye because the eye has already seen brighter colors. It has a pre-existing reference point. Going by this analogy, the mind can only recognize the “darkness of matter” because it already knows the real state of matter as an existing thing with positive properties. This argument does not show that matter is indeterminate and waiting around for the forms and the principles to give it an identity. It shows precisely the opposite: that matter already has an identity as the forms interact with it. To drive home his point about matter being a chaotic junkyard that needs to be organized by rational principles, Plotinus resorts to the most self-destructive tendency in Platonism: an empirical analogy from the world of the senses. He claims that “a white object becomes white by the presence of whiteness; what makes an organism white or of any other variety of color is not itself a specific color but, so to speak, a specific rational principle: in the same way what gives an organism a certain bulk is not itself a thing of magnitude but is magnitude itself, the abstract absolute, or the rational principle.”

There is something about color that attracts Platonists and Aristotelians like bees to a flower. The color white is the product of all the colors in the visible part of the electromagnetic spectrum combined together. An object that appears white is uniformly reflecting light waves across the visible spectrum. A white object is not white because the magical principle of “whiteness” somehow inheres within it and turns it into a white thing. Likewise, what gives an organism mass and spatial extension, or “bulk,” is the fact that the organism is composed of atoms and molecules which themselves have some “bulk.” The organism gets its macroscopic mass from the activity of its microscopic parts. Rational principles have nothing to do with this process. As if to sum up the contradictory mess in which he finds himself, Plotinus actually writes that the mind knows the nature of matter “in a sort of non-knowing.” Further elaborating on the mystery of matter, Plotinus claims that “matter is essentially indefiniteness…the matter even of the intellectual realm is the indefinite; it must be a thing generated by the undefined nature, the limitless nature, of the Eternal Being, The One.” It has to be so, according to him, because matter cannot produce or organize itself. He presses this point by arguing that “number and rational principle is outside of boundlessness: these bestow bound and settlement and order in general upon all else…the thing that has to be brought under order, matter, is other than the ordering principle which is limit and definiteness and rational principle.”

This frivolous speculation from Plotinus prompts an important question. How can matter exist both in our world and in the intellectual realm? How does that even make sense? Plotinus informs us that it makes sense because “indefiniteness has two phases.” The matter in our world is an “image-thing remote from true being.” By contrast, “the indeterminate in the intellectual realm, where there is truer being, might almost be called merely an image of indefiniteness: in this lower sphere where there is less being, where there is a refusal of the authentic, and an adoption of the image-kind, indefiniteness is more authentically indefinite.” In other words, matter in the intellectual realm is closer to the principle of “indefiniteness” whereas matter in the sensible realm is further away, thereby being inferior. We’ve reached a curious point. Aristotle merely establishes a hierarchy between form and matter. Plotinus goes all in and starts erecting hierarchies between different levels of matter, on top of all the other make-believe hierarchies floating around in his magical universe.

Beyond all this confusion, Plotinus also directly addresses the heart of Epicurean metaphysics. He doesn’t make much progress in this endeavor either, but no one can fault him for trying. Recall that Epicurus had argued that everything in the world is made up of interacting atoms, which he conceived of as tiny, indivisible, and indestructible solid bodies. Clamoring against the existence of atoms, Plotinus maintains that it’s “an absurdity, an impossibility, to hand over the universe and its contents to material entities, and out of the disorderly swirl thus produced to call order, reasoning, and the governing soul into being…the atomic origin is, if we may use the phrase, the most impossible.” In a world governed by atoms, Plotinus claims that “nothing…is orderly; order has not come into being, though the outcome, this Universe, when it achieves existence, is all order; and thus prediction and divination are utterly impossible, whether by the laws of science…or by divine possession and inspiration, which no less require that the future be something regulated.”

Plotinus also wonders about how atomic motions could possibly affect the soul, causing it to reason and to make decisions. He asks the following question: “would anyone pretend that the acts and states of a soul or mind could be explained by any atomic movements?” This question presupposes that the “states of a soul” are something different from the collective “atomic movements” in the body, when that’s exactly the point under contention. Plotinus begs the question against the atomists by already assuming the truth of what he’s supposed to demonstrate. It’s not surprising that he commits this kind of fallacy, since it’s rampant throughout Platonism and Aristotelianism more broadly: the notion that matter needs some kind of external source with agency that can shape it into coherent forms and structures. In reality, as Epicurus and other ancient atomists recognized, matter is inherently dynamic and interactive, so there doesn’t need to be an external supernatural source to provide it with coherence. Aristotle wasted a lot of ink searching for the philosophical source of motion, poetically concluding in Metaphysics that all motion is meant for the stars. But there is no ultimate source of motion, since motion is the default state of reality. Things have always been in different states of motion and they will always continue to be in different states of motion. Nothing is ever at rest, not in any deep scientific or ontological sense. The world is all just pulsating bundles of quantum energy.

Plotinus continues his attack against atomism with another related inquiry: “How can we imagine that the onslaught of an atom, striking downwards or dashing in from any direction, could force the soul to definite and necessary reasonings or impulses or into any reasonings, impulses or thoughts at all, necessary or otherwise?” This question is based on the false premise that the atoms are somehow striking a detached and independent soul, when the atomists are actually claiming that complex atomic motions constitute the soul itself. Simply put, the organized activity of the atoms inside the body is the soul. What’s doing the organizing? The atoms are! They constrain themselves through their mutual interactions, which have settled into certain patterns as a result of broader environmental conditions involving other atoms.

Plotinus knows very well what buttons he’s pushing through his cavalier attitude against the Epicureans: he’s trying to put the atoms “in their place,” so to speak. The entirety of The Enneads is devoted to elevating the significance of the divine, the rational, the absolute, the eternal, the perfect. The world of matter is relegated to the bottom of the hierarchy, denounced as evil, corrupt, imperfect, divisible, and fleeting.

In debasing the qualities and properties of the lowest hierarchy, however, transcendence also exposes the fragility of whatever rests on top. It’s worth asking the Platonists: from where does the rot and degradation in our world originate, if not from the top? When the toilet overflows, the bathroom floor ends up covered in crap. What is The One, after all, but the humble manure fertilizing our glorious fields? If there’s evil, division, corruption, and imperfection in our “lower world,” it’s because these maledictions came from the “highest on high.” Plotinus used the very senses he slandered to conclude that the celestial heavens are permanent, since he never saw any stars vanish before his eyes. He would have been sorely disappointed to know that his precious heavens are made from some of the same materials found here on Earth. The “rot” extends all the way through the hierarchy.

Beyond Plotinus: Concluding Thoughts

Later Platonists tried to conserve the purity of their metaphysical theories by viewing emanation across the different orders of reality as a gradual progression of descending causes. The fifth century philosopher Proclus used precisely this strategy, which he probably borrowed from other Platonists like Syrianus and Iamblichus, in his Elements of Theology. In Proposition 7, Proclus writes that “the thing produced…is neither equal to nor better than its produc­ing cause: and hence the producing cause is in every respect better than the nature of the thing produced.” In Proposition 28, he writes that “every producing cause…constitutes things similar to itself prior to such as are dissimilar.” What he’s saying with these passages, in effect, is that every act of emanation produces a lower order that is similar, but slightly less perfect, than the higher order. Here’s another way to think about it. Primary causes produce similar primary effects, and these primary effects in turn become secondary causes for other secondary effects, which are closer in nature to the secondary causes than to the primary causes. Unfortunately for Proclus, this line of reasoning doesn’t help Platonism at all. Whatever categorical properties or possibilities might be inherent in The One, this gradual succession of causes merely changes the perfection and resemblance of those categories. It does not get rid of them and it does not create any new ones. If there’s division and corruption at the lowest rung of the hierarchy, then this argument implies that division and corruption still exist at the top, just in smaller or different amounts.

The One, assuming it even exists, cannot be pure, simple, perfect, fundamental, or undifferentiated. As it radiates various properties to the other orders of reality, The One is simply amplifying its internal tensions and instabilities, like a collapsing layer of snow that intensifies into an avalanche. Furthermore, the supposition that efficient causes produce similar effects is not always warranted. It might be true in certain cases like human beings, which do originate from other human beings. But it’s far more difficult to justify for something like nuclear fusion, in which lighter nuclei combine to form heavier nuclei and also produce radiation as a byproduct. The heavier nuclei might be similar to the lighter nuclei, but the radiation is very different. One can make this case even stronger by looking at interactions between matter and antimatter particles. When electrons and positrons collide, for example, they annihilate each other and produce radiation. Once again, the resulting products of the interaction are not really analogous to the original systems. The other problem with this Platonist argument is that the notion of similarity is highly subjective; two concepts or ideas that look similar to one person may appear to be quite different to another. Similarity is not a fundamental ontological category of reality; like other aspects of language, it’s a useful convention for simplifying a complex reality.

Where have all these futile arguments left the Platonists? Nowhere worth being. They set out to find order, reason, and authority, only to discover that the real world is governed by chaos, multiplicity, and revolution. The natural world does not need a higher authority to bring it into order. Nature is causally closed and natural systems are inherently dynamic and interactive. As Epicurus pointed out over two thousand years ago, natural systems bring about their own pockets of order from their collective interactions. Nature has always been supreme and autonomous.

If you’ve ever wondered about the energy and biophysical demands of AI but didn’t know the right questions to ask, then this video is an excellent place to start.

Check it out below!

1. Podcast interview with at for the Macro N Cheese podcast. Steve is an amazing host who knows how to both let his guests get their point across but can also push back effectively when the moment calls for it. I really enjoyed the sheer breadth and content that we covered in this conversation, from the flaws of neoclassical economics to why there are so many dead zones in the world.

2. Monthly Review is the publisher of my book, and this was the extended kick-off panel discussion we did back in February for the book’s publication. This conversation features John Bellamy Foster, a well-known socialist thinker in the United States, and Blair Fix, a Canadian political economist. It’s the most extensive interview I’ve done so far about the book, and we get into a lot of nitty-gritty details in this one.

3. I loved this interview with Miles for her Moral Footprint podcast. We covered a lot of fascinating ground, including the challenges with relying on carbon capture technologies and the limits of technological solutions for our ecological crises.

4. And finally, I wanted to highlight a discussion that was not about the book. This talk between Blair Fix and I focused on the recent trade war and its wider implications and historical context.

Chapter 9

The Industrialization of Britain: A Case Study

Everything about the Industrial Revolution is debated. When did it start and when did it end? Was there one industrial revolution or multiple industrial revolutions stretching across several centuries? Many historians occasionally speak of the First Industrial Revolution from about 1750 to the middle of the nineteenth century and the Second Industrial Revolution from about 1870 to the outbreak of the First World War. In recent times, some writers have even talked about a “Third” Industrial Revolution related to the rise of the Internet, digital technology, and renewable energy.[1] Many scholars prefer to emphasize the “Great Acceleration” that occurred after the Second World War, in which the global AFR witnessed a massive explosion, the human population skyrocketed, and the impact of civilization on the entire biosphere reached unprecedented heights. The main aim of this chapter is to explain the emergence of the First Industrial Revolution in Britain, since all later waves of technological development and economic growth in modern times flowed from that initial spark.

Other questions about the history of industrialization abound. Should industrialization be seen as a radical, discontinuous transition to a new kind of economy or was it a more gradual and continuous transition from the medieval economy to the early stages of modern capitalism? In other words, was it the Industrial Revolution or the Industrial Evolution? How dependent was the Industrial Revolution on slavery and the Atlantic trade system? Should the Industrial Revolution be understood as a largely internal and autonomous process unfolding within Western Europe or as a broader phenomenon in which European historical dynamics were fundamentally entangled with external developments outside of Europe? Did colonialism and imperialism predate and facilitate the Industrial Revolution or was is it the other way around, with the rise of industrialization making possible imperialism and colonialism? Were constitutional and institutional changes more responsible for industrialization, or do technological innovations and bold entrepreneurs deserve more of the credit? These are just some of the important questions and controversies swirling around the Industrial Revolution, and it’s hopeless to pretend that we can answer all of them here. But what we can do is provide a comprehensive explanation of the Industrial Revolution that provides useful context, if not necessarily strict answers, for all of these questions.

The Ecodynamic Synthesis and the Industrial Revolution

In The Unbound Prometheus (1969), the historian David Landes placed the Industrial Revolution in a wider historical context, arguing that it was part of a larger wave of modernization. According to Landes:

Industrialization is…at the heart of a larger, more complex process often designated modernization. This is that combination of changes—in the mode of production and government, in the social and institutional order, in the corpus of knowledge, and in attitudes and values—that makes it possible for a society to hold its own in the twentieth century; that is, to compete on even terms in the generation of material and cultural wealth, to sustain its independence, and to promote and to accommodate further change. Modernization comprises such developments as urbanization…the establishment of an effective, fairly centralized bureaucratic government; the creation of an educational system capable of training and socializing the children of the society to a level compatible with their capacities and best contemporary knowledge; and of course, the acquisition of the ability and means to use [updated] technology.[2]

Landes became famous for his account of the technological leaps and bounds that made the Industrial Revolution possible, especially from the late nineteenth century onward. But his basic causal explanation is naïve because, in typical fashion for cultural and idealist theories, it implicitly takes the modernization process as a given deus ex machina that just comes out of nowhere. Many aspects of modernization that Landes mentions in the passage above, such as changes in the mode of production or changes to the corpus of knowledge, were largely the products of industrialization as a historical process. It was precisely the changes to the underlying productive methods and economic forces of society that generated further changes in culture, knowledge, values, and ideas. And those deep underlying changes to the coronets of industrializing economies started off, in large part, because of unexpected ecological shifts and ruptures that human beings did not control. In this sense, only materialist theories can properly ground broad historical changes to the social metabolism in the wider architecture of the natural world. As I explained in Chapter 3, cultural theories of economic development are plagued by too much idealism, and thus have no concrete sense of causation or causal context. The reality is that behavioral changes among people, including the behavioral shifts of the inventors and investors of the Industrial Revolution, can be best explained by invoking economic incentives, ecological pressures, and the broader class dynamics that mediate the relationship between those incentives and pressures.

Marriage patterns in premodern Europe provide a good example. Historians have noticed that women in northwestern Europe during the Late Middle Ages waited longer to get married than women in other parts of Eurasia and Africa. Whereas women in these latter parts of the world would often get married in their teens, many women in Western Europe waited for marriage until their twenties. The end result of this delay was that Western European women had, on average, fewer kids. A smaller average family implies that goods and services are spread out over comparably fewer people, thereby raising living standards. But what explains this difference in marriage patterns between Western Europe and other regions around the world? A plausible explanation is the relatively high wages that developed in several northwestern European countries after the Black Death.[3] The extreme labor scarcity caused by the huge death toll in Western Europe prompted many women to continue working longer because they no longer had the same economic pressures and incentives to get married. So, what seems like a cultural explanation on the surface can be exposed as an economic phenomenon upon further examination. Cultural explanations for the Industrial Revolution also focus on supply-side constraints but deal very little with the demand side of the story. For example, greater scientific knowledge can certainly contribute to conversional spectralization, such as when people learned that the atmosphere actually has weight and can apply pressure. This discovery in the seventeenth century set the framework for the development of atmospheric steam engines in the eighteenth century. But still, new technologies can only arise if there’s some kind of demand for their existence, and that part of the story can only be understood once we examine the economic incentives that are driving people to make their decisions.

Running on a semi-parallel track to the cultural school, liberal and institutionalist perspectives emphasize the emergence of limited parliamentary governments as the important step on the road to capitalism. Many liberal historians argue that the Glorious Revolution of 1688 made the Industrial Revolution possible because it limited royal authority, consolidated the power of the English Parliament, and secured private property rights. I’ve already discussed some of the foundational problems with neoclassical institutional theories, but I can mention a few more in the context of the Industrial Revolution. First, studies of banking and interest rates “fail to detect any structural break after 1688,” meaning that no financial indicators can be found to support the idea that the investment climate improved noticeably.[4] Second, property rights in France, and even in China, were at least as secure as those in England during the eighteenth century. In fact, France might have had too much in the way of property rights, as evidenced by the example of profitable irrigation projects in Provence, a territory in southeastern France. These projects were not completed because property owners opposed the enclosure of their land for public use and did not want to construct any canals or turnpikes.[5] They only started after the French Revolution limited local liberties and concentrated more power in the national assembly. Meanwhile, the British government had no problem forcing property owners to give up certain rights. Third, taxes were higher in England compared to France during the eighteenth century.[6] All in all, institutional and constitutional theories cannot adequately explain the scale of the energetic and technological changes that happened in the Industrial Revolution.

In his major 2009 work, The British Industrial Revolution in Global Perspective, historian Robert Allen emphasized the importance of cheap energy and high labor costs in making England a unique environment for investment in new technologies. With a cheap and abundant source of energy suddenly everywhere, inventors and businesses decided to invest more time and money in labor-saving technologies. As Allen puts it, “The success of the British economy was…due to long-haired sheep, cheap coal, and the imperial foreign policy that secured a rising volume of trade.”[7] In 1800, coal output from British mines was 66 times higher than what British mines could produce in 1560. This enormous supply of cheap energy depended on technological innovations, but also on plain good fortune: England just happened to be sitting on massive deposits of coal that were strategically located and easy to exploit. The influx of silver from the New World, along with various demographic and political changes, led to wage and price inflation throughout Europe. This inflationary pressure was especially pronounced in northwestern Europe because of its high death toll from the Black Death and other recurring plagues. In the Late Middle Ages, for example, wages were roughly equivalent in all major European cities. By the year 1600, however, workers in London and Amsterdam had pulled well ahead of their counterparts in Europe and Asia. English and Dutch laborers made about 8 grams of silver a day in 1600 whereas European laborers in other major cities made somewhere between 4 and 5 grams of silver a day.[8] This wage inflation continued unabated in London, to the point where London workers were making roughly 15 grams of silver a day by 1800. Contrast these earnings with the 9 grams of silver a day in Amsterdam and the 4 grams of silver a day for workers in Beijing at the same time.[9]

The relatively high wages in Britain led to a major push for developing labor-saving technologies. Capitalists and industrialists wanted to automate the production process because they could save a lot of money on labor costs if they could get machines to do what human beings once did. But this logic only worked well in Britain because wages there had reached such comparably high levels. Other parts of the world had no equivalent incentive to make the huge upfront investment costs for technological innovation because rich elites in those countries could always find a vast pool of cheap labor to perform the necessary tasks of civilization. This is one of the most important reasons why England invented practical steam engines and Song China did not. In China, cheap labor was readily available and so there was no great incentive to develop complex machinery. The situation was completely different in England. The steam engine was first used to lift water and coal out of mines. Following Thomas Newcomen’s engine in the early eighteenth century, many inventors went on to make several marginal improvements to its fuel efficiency and power output. The engines developed by James Watt, for example, more than doubled the efficiency of the Newcomen engines. But these changes were all expensive and could only have succeeded with a regular supply of funding from willing investors, who wanted to invest in new technologies as a way of bypassing high labor costs. The owner of a coal mine in nineteenth-century England could have hired people and bought horses to extract water and coal out of his mine. But getting a steam engine to do the job faster was far more economical in the long run, and so English capitalists used steam engines more and more to replace much of the work that was previously done by miners.

We now have all the necessary pieces of the ecodynamic synthesis to coherently demonstrate how and why industrial capitalism first emerged in England, and not in places like Spain or Sweden. The ecological conditions that led to the development of capitalism spanned everything from ancient and gradual mechanisms to sudden and traumatic convulsions, including a moderate warming process in the Middle Ages that boosted European agricultural productivity and helped the continent to triple its population in about two centuries. But the Black Death, which first struck Europe in 1347, was easily the decisive ecological primer on the road to capitalism and industrialization, because it induced radical modal shifts in the economies of Western Europe, which gradually culminated in a series of conversional spectralizations that led to an unprecedented expansion in the AFR of the entire region, but especially that of the English economy. Making this claim does not commit anyone to the more radical notion that the Black Death was exclusively responsible for the rise of capitalism and industrialization. The simple fact is that the rise of industrialization in England, and its later diffusion across the world, depended on the temporal and spatial convergence of many complex ecological, political, economic, and cultural causes. Nevertheless, one can still recognize the fundamental importance of the Black Death among these convergent causal factors without being naïvely reductionist.

In 2022, the historian James Belich published The World the Plague Made, perhaps the most comprehensive analysis of the impact the Black Death had on the future economic development of Europe. Belich writes: “The Black Death fundamentally transformed…the relationship between labor and capital, creating a strong incentive to replace the former with the latter where possible.”[10] Most of the revolutionary technology used in Europe during the Middle Ages came before the Black Death, apart from the Gutenberg press. The long series of plagues starting with the Black Death, however, provided extra incentives to use and to proliferate labor-saving technologies. As Belich puts it, “There were few entirely new inventions, but rather a sharply increased uptake of three existing inanimate sources of energy: water power, wind power, and gunpowder.”[11] Before 1350, water-powered mills were typically used to grind grain and produce flour, which could then be used to make bread and other food products. But after 1350, water mills not only pounded a higher fraction of grain products, they increasingly shifted to being used for other industrial tasks, such as sawing timber and powering hammers. These initial modal adaptations eventually overcame previous energetic constraints on economic growth because England, in particular, had large concentrations of energy-dense natural resources like coal and relatively high labor wages that encouraged investments in automation. In addition, massive shipbuilding programs and high rates of urbanization induced raw material scarcities in domestic markets and persuaded many English entrepreneurs and governments to heavily invest in the expansion of industrial capacity. Furthermore, the discovery of two strategically exposed continents yielded vast new lands for mining and agricultural production, feeding an imperialist frenzy to seize forced slave labor from Africa and other regions of the world as part of the drive to minimize the costs associated with the production of raw materials for industrial manufacturing.

The growth of European industrial production after the Black Death is truly remarkable. The raw number of English industrial mills more than tripled from 1,300 to 1540, going from 600 to roughly 2,000. On a per capita basis that rise represented a sixfold increase because of the demographic hits that England sustained from numerous plague outbreaks.[12] The Black Death was also directly implicated in the rise of the silk industry in the Italian city of Bologna, which became Europe’s first proto-industrial hub. By 1371, Bologna was using “12 water-powered silk spinning mills, and the number increased rapidly thereafter to create Bologna’s main industry.”[13] This elaborate system of silk production relied on a huge underground network of tunnels that moved water around the city, known as chiaviche. The operation of this vast network required several hydraulic innovations and remained a closely guarded secret until the middle of the fifteenth century.[14] Bologna’s silk industry became famous for producing a thicker kind of silk yarn, known as organzine in English, which was then exported to other parts of Italy for processing into finished products and clothing. In the twelfth century, Bologna relied on about seventy grain mills to feed its population; by 1393, it only needed twenty grain mills for the same task, while sixteen mills were used for spinning silk and twenty-one others were deployed by industries like wool, paper, and metallurgy.[15]

The Black Death also had a strong impact on iron production. Before 1350, iron production was largely confined to small bloomeries that generated about “two or three tons of metal a year.” After the Black Death, there was a transition toward larger blast furnaces that could produce about “40 or 50 tons a year.” Blast furnaces reduced the number of smiths required for executing and managing the smelting process from 30 or 40 to just two or three in some cases.[16] It’s been estimated that European iron output “tripled or quadrupled” from 1400 to 1525.[17] The continent’s leading exporter, Sweden, experienced a sixfold increase in total iron production from 1340 to 1539, and the per capita figures are almost certainly much higher.[18] Indeed, Sweden became a major supplier of iron to England starting in the seventeenth century and would retain that role throughout much of the eighteenth century as well.

My basic thesis is that the Industrial Revolution in England came down to four major things: sheep, coal, trade, and war. That’s an exaggeration, but only a slight one. First came the ecological primer: the Black Death caused a huge population decline, especially in England, which had one of the worst per capita death tolls in Europe. Fewer peasants were available to cultivate corps, so the size of cultivated lands fell dramatically. Lords and peasants converted more land to pasture instead. The rise of pastoralism was a modal adaptation to a severe labor scarcity, since it takes fewer people to herd sheep than to till land to manage crops. Gradual improvements in grass and soil quality yielded higher-caliber wool that was used for the production of textiles and clothing. England’s woolen cloth industry became more and more powerful, eventually displacing the Italians and the Flemish as the dominant textile manufacturing center in all of Europe by the seventeenth century. The “new draperies” became responsible for a huge share of London’s exports, and they attracted more workers to the city’s burgeoning textile industry. In the meantime, England’s trade policies and its strategic rivalries with other European nations sustained a growing boom in the construction of warships and merchant ships. A large share of London’s workers became employed in the shipping sector, building ships and handling maritime trade. But war and imperialism had another major effect on industrialization in the eighteenth and nineteenth centuries: they led to a massive boom in government debt that drove structural changes in the investment patterns of the English economy. Many investors took the substantial profits they derived from selling government bonds and invested that money into the emerging industries and sectors that would come to revolutionize the country. And by the time of the Second Industrial Revolution in the nineteenth century, British imperialism had amassed large colonial territories that provided critical raw materials necessary for Britain’s domination of global trade.

War, trade, and textiles were the dominant factors behind the spectacular demographic growth of London in the sixteenth and the seventeenth centuries. But the enclosures and privatizations of previously common lands, ruthlessly turbocharged by aristocratic forces in Parliament during the seventeenth and eighteenth centuries, also contributed to the emergence of an urban proletariat as poor peasants were violently displaced from their traditional lands and communities. The rapid growth of London then spurred extra demand for timber, because households had to burn wood to stay warm, and the government needed wood to build warships, among other reasons. The rapacious extraction of timber caused local deforestation crises in much of England in the seventeenth century, sending timber prices soaring. Coal then emerged as a ready substitute for declining timber stocks, and the northern parts of England were full of it. We see here another example of an ecological scarcity inducing a shift in the resource base and modal adaptations of an economy. Growing demand for coal led to intensive mining and the search for methods to optimize the extraction process. That’s why the first steam engines were invented: to extract water out of coal mines. Steam technology then experienced a series of spectralizations throughout the eighteenth century with the addition of condensers that improved thermodynamic efficiency, new gear configurations that allowed the engines to generate rotary motion and to power machines in factories, and finally the transition from using low-pressure steam to high-pressure steam as the driving motive force, which was the breakthrough moment of the entire Industrial Revolution. Steam engines were already prime movers by the late eighteenth century and became dominant in the British economy by 1870, vastly outpacing the exergy capacity of other devices. The accelerating diffusion of steam power revolutionized all other industries and economic sectors, from textiles to transportation, leading to a massive expansion in the energy scale of the British economy. That, in a nutshell, is how and why the Industrial Revolution first happened in England.

It would be remiss to discuss the causal dynamics of the Industrial Revolution without mentioning anything about agriculture and transportation. Gradual improvements in agricultural productivity, for both crops and livestock, facilitated the expansion of urbanization and were thus critical in shaping the contours of the modern economy. It’s been estimated that the milk produced per cow in England rose from roughly “100 gallons per year in 1300 to 380 gallons in 1800.”[19] Likewise, the fleece weight from a typical sheep increased from 1.5 pounds to 3.5 pounds. Wheat yields doubled from ten bushels per acre in 1300 to about twenty bushels per acre in 1700. Agricultural productivity increased for several reasons, such as farmers making genetic improvements to their crops and animals as well as the improvements in land and soil quality that resulted from certain enclosures. Advances in transportation also boosted the process of industrialization because it became cheaper to move goods around, thus expanding the available markets for new commodities produced in homes and factories. The construction of new canals and improved roads made it far easier to connect the different regions of the English economy. In particular, Parliament authorized the creation of numerous turnpikes, nonprofit trusts that collected fees from road users and spent much of that revenue on improving the quality of the roads. Before the rise of turnpikes, most English roads had deep ruts and potholes that made it difficult to transport goods on the ground, thus constraining the level of integration between regional markets.[20] By the nineteenth century, England’s efficient road system had become a vital lifeline for virtually all businesses and households in the country.

The Adaptations and Spectralizations of the English Economy

Fossil fuels were the keystone resources of the Industrial Revolution because their chemical bonds store enormous amounts of energy. Consider that burning a kilogram of wood releases about 15 megajoules while burning a kilogram of coal releases about 30 megajoules.[21] And burning a kilogram of natural gas yields about 40 megajoules while burning a kilogram of gasoline yields about 43 megajoules. But it’s important to note that a given mass of coal occupies roughly a quarter of the volume as the same mass of wood, so the volumetric energy density of coal is roughly eight to ten times higher than that of wood, an enormous difference. These are all average estimates, and the real-world energy performance of any given quantity of timber or coal will depend on many factors. Nevertheless, they accurately capture the fundamental truth that fossil fuels contain far higher energy densities than sources of energy like renewables or biomass. In the nineteenth century, a few locomotives in the United States actually operated by burning wood instead of coal, but they could never successfully compete against coal-based locomotives because they couldn’t store the vast amounts of wood necessary for traveling long distances, a consequence of timber’s relatively low volumetric energy density. Any society that can successfully transition to burning coal over wood for most of its domestic needs will necessarily experience a massive expansion in its energy scale, and that’s exactly what happened to England.

Before England could shift to using coal, however, there had to be a reason to induce the change. The fundamental reason for that change was the spectacular growth of London from 1500 to 1700. And the main reason why London grew so rapidly is because of England’s modal shift toward pastoralism after the Black Death. As millions of acres of arable land were converted to pasture, a series of gradual improvements in land quality and nutrition doubled the typical fleece weight from the fourteenth century to the seventeenth. In the Late Middle Ages, English and Dutch manufacturers started producing a series of enhanced woolen products known as the “new draperies,” which were basically light worsteds. Unlike woolens, which are fuzzy and rough, worsteds are based on high-quality wool and are strong, fine, and smooth. These clothing products usually followed the patterns of Italian fabrics, but they eventually became so successful that they “drove Italian producers out of business in the seventeenth century.” To boost its domestic wool industry, England imposed an export tax on raw wool, essentially forcing that wool to be used in domestic manufacturing instead of letting it go abroad. That way the English could export finished clothing products instead of sending out raw materials. England’s wool industry remained the dominant core of its textile manufacturing until the rise of cotton displaced it in the early nineteenth century. Wool was so important to England that many of Parliament’s import bans in the eighteenth century were specifically designed to protect the industry. Indeed, the rise of the wool industry was fundamental to England’s growing prestige in European trade. As Robert Allen puts it:

By the end of the seventeenth century, about 40 per cent of England’s woolen cloth production was exported, and woolen fabrics amounted to 69 per cent of the country’s exports of domestic manufactures....Wool was even more important for London. The new draperies flowed out of the capital: cloths amounted to 74 per cent of London’s exports and re-exports in the 1660s...and made a large contribution to the growth of that city. By the early eighteenth century, one-quarter of London’s workforce was employed in shipping, port services, or related activities.[22]

As London grew, it needed more wood for construction and heating fuel. The overextraction of local forests led to timber scarcities in many local and regional markets, thereby contributing to a sharp rise in timber prices during the seventeenth century. Scholars now generally believe that there was no national deforestation crisis the way economic historian John Nef once argued, but it’s indisputable that smaller and more regional levels of deforestation did take place.[23] And these local deforestations had a major effect on timber prices just the same. As a result, the timber scarcities induced a modal adaptation toward the use of coal, which was much cheaper, for industrial activities like smelting and heating. The annual output of coal in Britain is estimated to have gone from roughly 210,000 tons in the 1550s to almost 3 million tons by the 1680s.[24] Coal had been used by the English for centuries, and it was even known to the ancients as a potential source of heating fuel. But it was extremely dirty when burned, so if timber was plentiful, and it was for most of human history, then people preferred to burn wood instead of coal. Once the English really got going with coal in the late sixteenth century, they encountered a bewildering array of technical challenges that they had to resolve. Households started building bigger chimneys to make it easier for the black smoke to pass through. And although chimneys had been technically around since the Middle Ages, they only became popular starting in the sixteenth century when English homes turned increasingly to coal as a heating source. In smelting, industrial producers refined the structure of the reverberatory furnace to prevent thick coal smoke from polluting the underlying metal. But the biggest technical challenges that the English had to overcome were related to coal mining itself.

The very process of finding and digging through coal mines, especially those that lay deep underground, encouraged the development of one of the most revolutionary technologies in history: the steam engine. Workers in underground mines frequently had to deal with seeping rainwater and uncertain groundwater reserves, all of which could flood the mine and disrupt the extraction process. In 1712, Thomas Newcomen invented perhaps the first practical steam engine in history to help mining companies pump out more water. Although the ancient Greeks had invented the steam engine almost two thousand years earlier, their prototypes were impractical and were not developed any further. Greek societies did not have the same ecological and economic incentives for technological growth that the English enjoyed in the early modern period. Newcomen created a low-pressure atmospheric engine that used the heat from burning coal to boil water inside a large container. The resulting hot steam then pushed a piston that was attached to a mechanical device, often a rotating horizontal beam, which performed a particular task like lifting weights when it moved back and forth. As the hot steam expanded and pushed the piston up, a small jet of cold water was sprayed into the main cylinder through an adjacent valve. That burst of cold water had the effect of lowering the air temperature inside the cylinder. Warmer air holds more water vapor, which is the gaseous phase of water, than cooler air. As the temperature dropped inside the cylinder, the steam started condensing into liquid water droplets. A liquid that forms out of condensed gas occupies a smaller volume and exerts less pressure on its surroundings than the gas. In effect, a partial vacuum emerged inside the cylinder, which means that the pressure inside was lower than the atmospheric pressure outside. The larger air pressure outside forced the piston down, and the cycle continued with a new blast of hot steam. Newcomen’s engine was an instant hit because it was far safer and more reliable than previous models.

Some decades later, the engineer John Smeaton made further enhancements to the Newcomen engine and improved its efficiency to a point where more coal mines were clamoring to use it. At roughly the same time, James Watt made other critical improvements. While fiddling around with the Newcomen engine, Watt added a separate condensation chamber so that the main cylinder could remain permanently warm, instead of going back and forth between hot and cold. Whereas the Newcomen engine condensed the steam inside the main cylinder, the Watt engine delivered it to another area through a controlled valve. This simple change more than doubled the efficiency of Watt’s steam engine compared to its predecessor. The engine could yield the same mechanical output as the Newcomen models by burning half as much coal, at least. Later improvements substantially expanded the engine’s mechanical output. In particular, the development of a “sun-and-planet” gear configuration allowed Watt’s engine to convert linear reciprocating motion into rotational motion, making it capable of being used in factories and other industrial places. Incidentally, Watt’s new rotational gear system was inferior to an already existing crank device, but a patent on the latter blocked its potential inclusion in Watt’s engines. Once that patent expired in the 1790s, Watt could finally marry the efficiency of his engine with the superior performance of the crank mechanism. Although the sun-and-planet gear allowed for rotary motion, it produced highly irregular and variable speeds that often made them difficult to operate successfully. Watt solved this problem by introducing the double-acting engine, which was probably his most important invention. In a double-acting engine, steam is supplied on both sides of the piston at different time intervals, which produces uniform speeds. Despite all this progress, the aggregate efficiency of English steam engines changed at a glacial pace because Watt and his business partner, Matthew Boulton, secured and aggressively defended a government-issued patent that allowed them to charge exorbitant prices for their comparatively efficient engine. Naturally, many coal companies opted to purchase the cheaper Newcomen engine, even though it was far less efficient. From 1775 to 1800, roughly 450 to 500 Boulton and Watt engines were produced, in contrast to the more than 1,000 Newcomen engines made during the same period. Figure 1 shows the composition of English steam engines by design type in the year 1800.

Figure 1. Number of operational English steam engines in 1800 by design type

Note: The Savery engines in Figure 1 are older steam engine designs from the English inventor Thomas Savery. These steam engines never really caught on with industry, although a few were produced and adopted by various users.

Source: Lee T. Wyatt III, The Industrial Revolution (Westport, CT: Greenwood Press, 2008), 199. See also Sally Baggott, Matthew Boulton: Enterprising Industrialist of the Enlightenment (New York: Routledge, 2016), 10.

Watt might be remembered as a great inventor today, but he spent much of his career in court proceedings defending his patents from other competitors. Many historians have blamed the patents for slowing technological progress on the steam engine at the end of the eighteenth century, but that assessment largely misses the broader picture.[25] Although Watt’s engine was a technological marvel, it still operated through low-pressure steam, which severely restricted the mechanical energy it could generate. The first engines that Watt produced generated only 6 HP.[26] In less than twenty years, his biggest engines were delivering as much as 190 HP, an extraordinary increase in mechanical power, but still less than what was necessary for huge factories and large trains. Watt spent much of his later life waging a very public campaign against the adoption of high-pressure steam engines. The real reason for his vehement opposition is that he viewed them as a threat to his business empire, but publicly he kept focusing on the dangers associated with them, especially their propensity to explode—a valid complaint. Previous attempts at building high-pressure engines had ended in failure, mostly because the energetic steam found its way through cracks and crevices in the boiler, which often led to major explosions. Most British inventors avoided trying to develop high-pressure steam engines for over a century because of these concerns about safety and reliability, even though the technical knowledge for generating high-pressure steam was understood by many.

Watt could not legally prevent anyone from designing and testing a high-pressure engine, and the Cornish engineer Richard Trevithick eventually managed to do just that, ironically as a way of bypassing the limitations imposed by Watt’s patents. Trevithick was the real “hero” of the Industrial Revolution, not James Watt, who fought against the technology that would form the basis of virtually all future engines and turbines in human history. Trevithick developed high-pressure stationary engines used by mining companies to extract water and coal out of the ground and then started veering more into steam-powered locomotion. Trevithick’s engines dispensed with Watt’s condenser and vented steam directly into the atmosphere, generating a characteristic sound that prompted people to call them “puffers.” In a major advance, he added his new light engines to carriages, thereby creating the first self-propelled passenger-carrying vehicles in history.[27] He placed the boiler at the back of the carriage and used connecting rods between the piston and the wheels to generate rotational motion. Rail tracks in his day, however, were often not strong enough to handle the weight of these new steam-powered devices, so they broke and cracked quite frequently. Reliable rail would have to wait until the 1820s, when George Stephenson and others built powerful and standardized tracks that could handle steam locomotion. Nevertheless, the derivatives of Trevithick’s engine designs gradually found their way into the trains and steamboats that powered the nineteenth century. High-pressure engines were far more efficient than their low-pressure counterparts, but their vast mechanical power is what sealed the deal. No longer reliant on the docile atmosphere to provide some extra pressure, the new engines exploited the dense and energetic steam to push around the pistons with enormous levels of force. Their power, not their efficiency, is what allowed them to move heavy things like trains and ships.

High-pressure steam engines had the advantage of being smaller, more compact, and more thermally efficient. Because pressurized steam contains a higher volumetric energy density than its low-pressure counterpart, every intake stroke of the engine could deliver more mechanical work to the piston for the same amount of steam. They did turn out to be dangerous, as Watt predicted, but far less dangerous than if they had proliferated a century before, when many critical advances in metallurgy were not yet ready. Following Trevithick and the Cornish engines, Arthur Woolf made the next big impact with the high-pressure compound engine, which used and recovered energy from a working fluid through several stages. In Woolf’s compound engine, a high-pressure cylinder vented steam to a low-pressure cylinder, the idea being to give the steam more time to perform useful mechanical work before dumping it to a condenser or directly into the atmosphere. Woolf’s engines had far better efficiency than anything made by Watt, but they were often difficult to operate because of their complexity. It wasn’t until the engineer William McNaught developed a more practical type of compound engine in the 1840s that it finally became economical for many companies to use steam. From then on, steam power infiltrated virtually all aspects of the British economy.

The diffusion of steam power across Britain was slow at first. In 1800, after a century of development, steam power could produce roughly 35,000 HP, compared to water at about 120,000 HP and wind about 15,000 HP. By 1830, the installed capacity of steam engines was about 160,000 HP, compared to 160,000 HP for water sources and 20,000 HP for wind sources.[28] But by 1870, just forty years later, steam power capacity had reached an astonishing 2,060,000 HP while water capacity stood at 230,000 HP and wind capacity amounted to about 10,000 HP. In other words, installed steam capacity grew at an annual rate of 12 percent from 1800 to 1830 before accelerating to an annual growth rate of 30 percent from 1830 to 1870.

And this brings us to a fundamental point about the early waves of mechanization and spectralization in the Industrial Revolution. Recall that in Chapter 5 exergy was defined as a thermodynamic system’s maximum capacity for useful work. Boosting exergy through the spectralization of conversional technologies was the main causal vector for the corresponding improvement in the efficiency of industrial devices. The Industrial Revolution in England, and virtually everywhere else as well, followed a path of exergy-driven efficiency gains that spurred additional gains and butterfly effects in economic productivity. The English achieved this incredible growth through a huge increase in the aggregate output of mechanical work, a process spearheaded by the spectralization of high-pressure steam engines, and eventually the spectralization of other engine types as well. Let’s highlight how this historical process worked through a simple example. In a given unit of time, let’s say a steam engine produces 1,000 joules of useful work and has a maximum capacity of 5,000 joules, for an exergy efficiency of 20 percent. However, a more powerful steam engine might generate 10,000 joules of useful work in the same amount of time with a maximum capacity 40,000 joules, for an exergy efficiency of 25 percent. Notice that the exergy efficiency has increased, but so has the amount of useful work produced by the second engine. In other words, it was not the push for higher efficiency that led to the energetic and biophysical expansion of the British economy, but rather the push to channel and concentrate more exergy, which led to higher efficiencies. The fundamental energy and economic story of the Industrial Revolution is not about efficiency-driven productivity gains, but about exergy-driven productivity gains. The corresponding efficiency increases were purely incidental to the larger project of simply scaling up the productive capacities of the industrial system, which required expanding its maximum capacity for producing useful work. In this regard, Boulton’s famous utterance to one of his visitors nicely sums up the core developments of the age: “I sell here, Sir, what all the world desires to have—POWER.”[29]

This exergy-driven expansion of the British economy had a curious effect on aggregate-level thermodynamic efficiencies. As steam engines quickly became the prime movers and dominated mechanical output, their gradual inclusion in the economy dragged down the efficiency of the economy as a whole. That’s because they started off as extremely inefficient devices compared to more established technologies like watermills and windmills, which had been refined and optimized over many centuries. At the sectoral level, for localized groups of mines and factories, the efficiency of steam engines began at a fairly low baseline with the Newcomen lineage and then gradually increased as future generations of engineers made improvements. But the opposite effect unfolded at the aggregate level. In Figure 2 , I estimate this era of divergence in the early history of capitalism by specifically tracking the evolution of aggregate efficiencies for English steam engines across time, and I compare that quantity to the aggregate efficiency of English steam engines and watermills combined. The most notable feature in the plot is a sharp drop in the aggregate efficiency of the English economy after 1770, caused chiefly by the Watt and Boulton engines coming to the market and the subsequent expansion of exergy capacity as next-generation steam engines started to proliferate. Although watermills were far more efficient at converting energy than early steam engines, and although watermills vastly outnumbered steam engines throughout the eighteenth century, a typical watermill had a far lower power output than a typical steam engine. That imbalance resulted in a large decline in aggregate efficiency as the number of steam engines increased across the eighteenth century. Note that Figure 2 actually shows the aggregate efficiency of steam engines by themselves as increasing over time, again with a much faster rate of improvement after 1770 as the Watt engines and Smeaton-enhanced Newcomen engines spread across the English economy. But because these engines were less efficient than watermills and because they start to represent a greater and greater share of all mechanical output, they actually start bringing down the overall efficiency associated with both devices. They likely brought down the aggregate efficiency of the entire English economy as well, but the effects there are probably far more subdued as heavy draft animals and other contingents would profoundly affect the results of the aggregation.

Figure 2. Aggregate thermodynamic efficiency of eighteenth-century steam engines and watermills in Britain.

Note: The methodology behind the Figure 2 chart makes a number of simplifying assumptions that will affect the final result. First, it does not include the mechanical energy supplied by draft animals, windmills, and other prime movers. A more comprehensive accounting of all these factors would show that aggregate efficiency still declines, but much less than what’s shown in the chart above. Second, the methodology assumes that all steam engines and watermills were operating for the same amount of time in ever year, which is obviously not true. This assumption will affect the estimated aggregate efficiencies shown in the figure, but it will not affect the basic conclusion that the aggregate efficiency of the English economy slightly declined as energy-intensive but thermally inefficient steam engines came into operation in greater and greater numbers. Third, the number of active steam engines is not reliably known for every single year, so I start off with years where the total number is better known and then use linear interpolation to estimate the active steam engines for the years in between. I also assume a constant 15,000 watermills are operating in every single year. Fourth, the methodology ignores the heterogeneity of the energy qualities involved, as in the case of steam engines coal is burned to generate the heat that produces mechanical work whereas in the case of watermills it’s flowing water that’s pushing the waterwheel. The annual efficiency of watermills is held at a constant 15 percent for every single year. By contrast, the annual efficiency of the steam engines is assumed to change through some average annual rate interpolated from known values. [30]

This narrative has intentionally focused on steam engines because they were by far the most dominant invention of the Industrial Revolution, and their tentacles eventually extended everywhere. By 1835, steam power provided roughly three-quarters of the power output deployed in the cotton textile industry as a whole, or 30,000 HP out of 40,000 HP, with the rest coming from water power.[31] That share kept increasing throughout the nineteenth century. It’s certainly true that inventions like the flying shuttle, spinning jenny, and the spinning mule substantially boosted textile production in the eighteenth century, but as with basically every other process in the history of English industrialization, the truly eye-popping productivity gains came in the nineteenth century, and steam engines were the main drivers of those gains. The story of the Industrial Revolution is the story of the spectralization of steam power across different economic sectors. Indeed, one can easily argue that the story of all modern civilization is largely a story about steam power. There’s a sense in the public imagination that the age of steam is over and that other technologies have since taken over. The reality is quite different. The dominant prime movers today are powerful steam turbines in power plants that can generate multiple gigawatts as they turn giant magnets and produce huge quantities of electricity. Hot and pressurized steam is still the king of modern civilization, but if we choose to spectralize in a different direction, we may someday succeed in bringing down the curtain on the steam age and raising the curtain on another age of technological progress.

In sum, technological innovation during the industrialization phase of British history was a process of intense conversional spectralization: the interdependent diversification and proliferation of conversional technologies designed to overcome engineering challenges and energetic constraints. The process was interdependent in the sense of being recursive, with outputs from one technological process serving as inputs for others, and in the sense of being socially engineered, with governments, investors, and inventors specifically looking to improve the performance of existing technologies by creating new variations. What typically transpired before modern times is that societies tried to overcome energetic constraints on economic growth by concentrating more human labor, more animals, more machines, or some combination of all these things. In other words, societies tried to overcome energetic constraints largely through modal adaptations. But during the Industrial Revolution in England, investors and inventors came to see the limits and constraints of technological devices themselves as barriers to be overcome and swept aside. They were persuaded to go down this path by virtue of the labor and resource scarcities in the early modern English economy, and by large commercial markets that offered the prospect of huge profits. Productivity gains in the early phases of capitalism relied on the expansion and diversification of mechanical energy. In other words, new technological systems were continuously refined to generate not just higher levels of mechanical work, but also new methods of applying that mechanical energy toward the achievement of different tasks. Watt himself considered the increase in efficiency to be the most important technological innovation of his engine, but efficiency is largely irrelevant during times when resources are plentiful, like coal in Britain during the eighteenth century. The real significance of the steam engine is that it offered people the ability to unleash vast amounts of energy in entirely new ways. Steam power could be used for both transportation and factory work, amplifying the productive and logistical cycles of early capitalism to new heights. After all, capitalists can only sell and distribute a huge number of commodities if they can produce them in bulk and move them where they need to go. The productive power of steam engines, along with many other machines and inventions, offered the new investor classes an incredible opportunity to financially capitalize on the resulting commodity boom. But to do that, the system also needed a vast infusion of money and credit.

War, Imperialism, and Financial Accumulation

Historians like Paul Bairoch have argued that the Industrial Revolution had very little to do with imperialism, colonialism, and global trade. Bairoch made these arguments on the basis that income from activities like the slave trade, and colonial trade more broadly, constituted only a small share of British national income in the eighteenth century.[32] According to Bairoch, Europe also had all the necessary resources for industrialization, so global trade could not have helped much anyway. Bairoch’s thesis is flatly false, but not for the traditional reasons that have been given by left-leaning historians like Eric Williams and Joseph Inikori.[33] It’s true that Europeans obtained resources like gold, silver, sugar, coffee, tobacco, and cotton from the colonies, but the early waves of industrialization were largely driven by coal and iron, and these two critical substances were sourced within Europe. Furthermore, Spain and Portugal established the first major colonial empires in the world, but they were not the first European countries to industrialize, not even close. Nevertheless, it’s impossible to overstate the significance of warfare, imperialism, and strategic competition in the emergence of industrialization. War impacted industrialization through multiple causal vectors. In general, it served as a dominant order parameter that shifted patterns of investment and technological development. The Napoleonic Wars, for example, caused a massive shortage of horses and fodder in Britain by 1811, as the British government prioritized equipping the army with an adequate cavalry force to fight against the French in Spain as part of the Peninsular War. The shortages induced a huge increase in transportation costs, which persuaded many collieries to invest in locomotion using high-pressure steam engines.[34] The first practical steam locomotive entered service in June 1812, and further advances quickly followed in engine technology and railway tracks. By 1830, the Liverpool-Manchester railway line had opened for public use, ushering in a new age in the history of human transportation. And unlike the initial waves of industrialization in the eighteenth century, the Second Industrial Revolution in the late nineteenth century was indeed critically dependent on the violent conquest and extraction of natural resources through imperialism and colonialism, especially since many of these natural resources extended beyond just bullion and cash crops. By the early twentieth century, natural resources like diamonds, oil, copper, cotton, rubber, nickel, manganese, and other minerals were largely extracted outside of Europe and played a dominant role in many major European industries. In 1854, Britain imported 19 percent of its foodstuffs and 26 percent of its raw materials from its external colonies.[35] By 1934, Britain was extracting 42 percent of its foodstuffs and 31 percent of its raw materials from the colonies. In particular, Britain stole and grabbed vast amounts of wealth and resources from India, which was deliberately deindustrialized and impoverished as a result of this systematic looting.[36]

War also played a role in accelerating energy transitions. From 1649 to 1660, in an aggressive naval expansion program pushed forward by Oliver Cromwell, the English navy added an astonishing 103 new warships to its ranks, with 69 of them carrying between 26 to 100 guns.[37] Between 1660 and 1688, the English produced another 106 warships with a total weight of roughly 74,000 tons.[38] By 1688, the navy had become “the most comprehensive and in some respects the largest industry in the country.”[39] This rapid naval expansion required huge amounts of timber for ship hulls and iron for things like anchors and nails. The subsequent rise in charcoal prices gradually led to a greater push toward coal-based coke as the main smelting agent in the metallurgical industry. In the popular imagination, Abraham Darby was the first industrialist to smelt iron by burning coke at Coalbrookdale in 1709. In reality, Darby borrowed his methods from earlier pioneers. In the 1690s, Shadrach Fox had been smelting iron with coke at the same blast furnace in Coalbrookdale that Darby would later come to operate. Fox used his coke-smelting techniques to provide cast-iron shells for English forces, which were busy fighting the French in the Nine Years’ War. The huge demand for iron and other products for the navy also forced the English to turn to foreign suppliers, in particular Sweden, Russia, and the Baltic regions. At the same time, demand for iron spurred an import substitution strategy in which the English made steady and gradual improvements to iron production techniques, culminating in Henry Cort’s famous “puddling process,” which converted molten pig iron into wrought iron inside a reverberatory furnace using coal-based coke as the heating source. In 1720, Britain may have produced roughly 25,000 tons of pig iron, a figure that by 1804 had increased tenfold to 250,000 tons, before shooting through the stratosphere in the nineteenth century to reach an astounding 2.7 million tons of pig iron by 1852.[40] Here too Britain’s involvement in the French Revolutionary Wars and the Napoleonic Wars proved decisive. Britain manufactured over three million small guns from 1793 to 1815.[41] Pig iron output increased from around 70,000 tons in 1788 to almost 400,000 tons in 1815.[42] The British military “consumed nearly 20 percent of iron output” during this period of wartime, and the iron industry actually entered a period of recession for years following the conclusion of the Napoleonic Wars in 1815.[43] For iron production, textile manufacturing, and many other industries, Britain implemented a policy of import substitution followed by aggressive export-oriented growth. In other words, it first mastered the production of goods that it was originally acquiring from abroad, then it became very efficient at making these things, then flooded continental and global markets with its industrial arsenal through brutal wars and aggressive trade policies. As part of this process, Britain deliberately destroyed and sabotaged India’s textile industry to boost its own domestic manufacturers.

From 1692 to 1815, British national debt went from roughly 5 percent of GDP to more than 200 percent of GDP, and this astonishing rise in government debt was almost entirely driven by war and imperialism.[44] Higher levels of government debt impacted industrialization via multiple causal pathways. First, many investors made huge profits in the bond markets of the eighteenth century and the early nineteenth century, accumulating a vast pool of money that was available for investment in the new industries of the era.[45] The most important debt instruments issued by the British government in the eighteenth century were the famous consols, short for “consolidated annuities.” Introduced in 1751, the consols had a fixed interest rate of 3.5 percent and no scheduled maturity date. Creditors could hold the consols until they decided to sell them on the bond market or until the government decided to pay them off. Issuing consols became a fundamental strategy used by the British government to finance wars in the eighteenth century. Britain redeemed all outstanding consols in 2015. The huge profits from selling consols incentivized the critical investments that accelerated industrialization in Britain, especially in the immediate aftermath of the Napoleonic Wars, when many investors made historic windfalls. Investors would rush to buy cheap debt during wartime, then sold their bonds once prices dramatically increased after the war. The resulting profits were widely invested for industrial manufacturing, spurring productivity and technological growth. During times of war, the British government sold consols at a heavily discounted price from their par value, which is only their face value. For example, if the government wanted to sell a £1000 bond, it would accept something like £500 or £600 pounds for it. As a result, investors rushed like crazy to buy these cheap and perpetual bonds that offered a solid rate of return. After the war ended, consol prices in the bond markets rose sharply because the government no longer issued the cheaper versions of the bonds. The war bonds wouldn’t necessarily sell at par value right away, but their price would climb much higher than what the investors had paid for them during the war. Many investors therefore made huge fortunes in the bond markets after they sold these high-value consols, and large portions of these profits were invested in new industries and technologies. A secondary yet important effect of rising government debt was that the nobility started investing heavily in government bonds, leaving less capital for their landed estates.[46] These shifting investment patterns depressed wages in agriculture, prompting peasants and farm laborers to find work in cities instead, becoming part of the new urban proletarian along the way. The capitalists and entrepreneurs active in the new industries of the Industrial Revolution strongly profited from this new wave of cheap urban labor, earning higher profits than what they could obtain in the bond markets in many cases. The financialization resulting from expanded government debt provided downward pressure on wages across the economy and allowed the new capitalists to score higher profits from a vast pool of cheap labor.

Finally, war and imperialism were essential methods of seizing and controlling necessary raw materials for industrial manufacturing that could not be found in Europe. The established formalities of daily capitalism unfold through ostensibly peaceful institutions like markets, contracts, and courts, but the system as a whole is foundationally built upon the violent acquisition of resources by the powerful from the powerless, a historical process that Marx called primary accumulation.[47] One could pick dozens of major historical examples to highlight his point, but let’s focus on the evolution of the spice trade. In the global trade system of the Late Middle Ages, spices arriving from Asia often fetched huge prices in Europe, where people used them for food preparation and preservation, medical care, and other personal purposes. Before the rise of sugar, wealthy Europeans clamored for vast quantities of cinnamon, nutmeg, cloves, and peppers. For much of the Middle Ages, the spice trade was controlled by Venetians and Arabs, whose legions of wealthy merchants scored large profits. Portuguese naval expeditions starting in the fifteenth century changed these commercial dynamics. Vasco da Gama rounded the Cape of Good Hope in 1498 and entered the Indian Ocean, establishing a major new trade route. The Spanish also joined this commercial arms race. Through military force and economic pressure, Portugal and Spain came to dominate the shipping lanes between Europe and Southeast Asia. Their trade monopolies allowed them to jack up prices on the European markets for critical spices like nutmeg. But the good times did not last. Other European countries became especially jealous of Portuguese profits, and pretty soon English and Dutch ships were sailing all over the Indian Ocean, ready to muscle out the competition. The Dutch East India Company, better known by its Dutch acronym of VOC, established in 1602 as a joint-stock company with the backing of 1,000 wealthy investors, could always count on the full support of the Dutch state, which waged war and sabotaged other European enemies as necessary in support of its crown financial jewel. Dutch involvement in Southeast Asia was cautious at first, but gradually the VOC started establishing forts and military bases throughout the islands that now comprise Indonesia.

Nutmeg was a unique product of trees that grew on the Banda islands (in today’s Indonesia), home to the Banda peoples and their ancient culture. Asian traders had kept the true source of nutmeg a secret for centuries, prompting Europeans to launch expensive and speculative journeys in search of the fabled trees. The Banda practiced neutrality and traded with all foreign powers that reached their islands. The Dutch hated that versatility and developed a simple plan: invade the islands, destroy the local population, and monopolize the nutmeg trade so they could set prices at will. The Dutch assault on the Banda islands in 1621 underscored the ruthless methods of the VOC. The Dutch killed thousands of indigenous inhabitants and ethnically cleansed the islands.[48] Dutch soldiers and foreign mercenaries beheaded civilians and impaled their heads on pikes; others were simply pushed off cliffs in large groups. In 1620, some 15,000 Banda people lived on the islands. By the time the Dutch were finished in the following year, only about 500 were left. To make up for this lost labor power, the Dutch imported slaves to the islands from abroad. These slaves worked and died on brutal nutmeg plantations, the ownership of which made several Dutch merchants superbly rich. The Banda islands became so important for Dutch commerce that, in a 1667 treaty with England, the Dutch gave up the town of New Amsterdam in exchange for keeping the remote archipelago. New Amsterdam went on to become an illustrious city; today people call it New York.

The increasing success of the VOC in the seventeenth century set the stage for the rise of the Amsterdam Stock Exchange, the first modern stock market. By issuing shares and bonds to anyone who could afford them, the VOC became the world’s first publicly traded company. The vast profits of the VOC were traded and packaged through stock options, futures contracts, and other exotic financial instruments. Whereas investors now listen impatiently to decisions from governments and central banks, back then people speculated with their money on the fate of far-flung Dutch fleets sailing with precious cargo. To protect the interests of the VOC, the Dutch state was forced into increasingly expensive conflicts against other competitors, including the Spanish, the English, the French, and the Chinese. Some of these wars turned out well and others did not. Rival European powers eventually managed to steal nutmeg seeds from the Banda islands, planting them in warm regions throughout the world, such as India and the Caribbean. Coupled with increased sugar production from Caribbean slave plantations, this development heavily undermined Dutch control over global trade and started reducing the profits of the VOC. By the eighteenth century, Britain and France had become the new masters of the world’s shipping lanes. The Netherlands, a small country that had become a world-beater, went into relative decline and never recovered. The state nationalized all assets and properties held by the VOC in 1799, ending the most successful company in the history of early capitalism.

Precisely because it set the tactics and methods for the rest of the capitalist era, the VOC is perhaps the best reminder that the creation of private property and the protection of wealth usually require systematic and organized violence against weaker groups of people and animals. This may be a difficult concept to grasp given the kinds of economic interactions we have in our daily lives. We find a property for sale, get a loan, buy a house. We try on a dress, see if it fits, pull out the credit card. We hear a good song, open the app, push the purchase button. Violence in our modern economies seems absent when we consider these common scenarios. But the point is not that violence accompanies every single economic or financial transaction. Rather, these transactions unfold in the context of economic systems that were themselves the products of coercion and violence. In order to buy houses, we must first build them on some piece of land. At some point in the distant past, someone probably took away that land from someone else through force and coercion, thus earning the exclusive right to use it and calling it private property. The same logic applies to the very first human inhabitants of that land, who had to displace other animals and plants in order to establish functional communities. Their property may have been shared and communal, but it still required some measure of coercion to actually obtain and control. The arrival of the first humans in Australia, for example, caused the extinction of numerous major animal species throughout the continent. At some basic level, almost all economic activity is predicated on the forceful extraction and isolation of natural resources. However, capitalism transformed this ordinary reality into an unparalleled battleground of violence and destruction, with vast trade surpluses plundered from the colonies and commodified through rigged global markets. Property does not suddenly appear from waving a magic wand; it emerges through warfare, economic pressure, ecological interactions, and political struggles. Then the winners establish laws and customs to make it all seem normal, but those laws are themselves products of the class struggles and labor relations that define the broader economic system. Property is always hostage to the changing fortunes of power.

Concluding Thoughts: The Future of Growth

The vast scale of modal adaptations that followed the Black Death presented new problems, challenges, and constraints for further economic development. Internal and external factors that converged at the right moment launched Europe ahead of other regions in the world. After the energy-intensive modes of capitalism started in Europe, they gradually spread to other regions around the world through trade, wars, revolutions, and migrations. By now, every continent on the planet has experienced varying degrees of urbanization and industrialization. North America, Europe, and East Asia have catapulted ahead of the rest of the world in terms of economic and technological development, but hardly anyone would dispute that capitalism is firmly established on a global scale.

As we’ve seen in prior chapters, some have claimed that the Industrial Revolution launched a period of self-sustaining economic growth that has no rival in human history. This claim is only partly true. Plenty of economies throughout history have experienced waves of growth and decline. It’s not like economic growth magically started with the rise of capitalism in the past few centuries, although this ridiculous story is often repeated by many economists and intellectuals as a way of legitimizing and defending the status quo. What was unique about the Industrial Revolution was the rapid rate of growth that it unleashed, a growth wave facilitated by a massive ecological disaster that forced a shift in modal patterns, which in turn led to the mastery and concentration of energy-dense fossil fuels and the resulting spectralization in conversional technologies. But rapid growth rates cannot continue forever because our economies are thermodynamic systems with an expiration date. On a global scale, there are already signs that the rapid growth rates unleashed by the Industrial Revolution are starting to subside. Facing a mounting avalanche of problems that it’s incapable of resolving, capitalism is becoming disoriented and losing steam. The time has come to consider a genuine alternative that would ensure the stability and prosperity of human civilization for thousands of years to come.

[1] See Jeremy Rifkin, The Third Industrial Revolution (New York: Palgrave Macmillan, 2011).

[2] David S. Landes, The Unbound Prometheus (Cambridge: Cambridge University Press, 1969), 6.

[3] See Robert C. Allen, The British Industrial Revolution in Global Perspective (Cambridge: Cambridge University Press, 2009), 13–14.

[4] Ibid., 125.

[5] Ibid.

[6] Ibid., 125–26.

[7] Ibid., 130.

[8] Ibid., 34.

[9] Ibid.

[10] James Belich, The World the Plague Made (Princeton: Princeton University Press, 2022), 123–24.

[11] Ibid.

[12] Ibid., 124.

[13] Ibid., 125.

[14] Ibid.

[15] Ibid.

[16] Ibid., 125–126.

[17] Ibid., 125–126.

[18] Ibid., 125126.

[19] Allen, The British Industrial Revolution in Global Perspective, 61–63.

[20] Stewart Ross, The Industrial Revolution (London: Evans Brothers Limited, 2008), 16.

[21] T. F. Walsh, America 2100: After Fossil Carbon (Raleigh: Lulu Publishing, 2015), 68.

[22] Allen, The British Industrial Revolution in Global Perspective, 109–10.

[23] See John U. Nef, “An Early Energy Crisis and Its Consequences,” Scientific American 237 (1977): 140–51.

[24] Brinley Thomas, The Industrial Revolution and the Atlantic Economy: Selected Essays (New York: Routledge, 1993), 16.

[25] See George Selgin and John L. Turner, “Strong Steam, Weak Patents, or the Myth of Watt’s Innovation-Blocking Monopoly, Exploded,” Journal of Law and Economics 54 (2011): 841–61.

[26] See John H. Lienhard, The Engines of Our Ingenuity, University of Houston, February 18, 1999, https://engines.egr.uh.edu/talks/powersir.

[27] See H. W. Dickinson and Arthur Titley, Richard Trevithick: The Engineer and the Man (Cambridge: Cambridge University Press, 2010), 43–60.

[28] Allen, The British Industrial Revolution in Global Perspective, 173.

[29] Bruce Hunt, Pursuing Power and Light (Baltimore: John Hopkins University Press, 2010), 10.

[30] For data on efficiencies and power outputs, see Richard Brown, Society and Economy in Modern Britain 1700-1850 (Milton Park: Taylor & Francis, 2002), 60. Also refer to Robert Balmer, Modern Engineering in Thermodynamics (Burlington: Academic Press, 2011), 454, 466. For data on watermills, see Terry Reynolds, Stronger Than A Hundred Men: A History of the Vertical Water Wheel (Baltimore; John Hopkins University Press, 1983), 123.

[31] Ronald Findlay and Kevin O’Rourke, Power and Plenty (Princeton: Princeton University Press, 2007), 320.

[32] See Paul Bairoch, Economics and World History: Myths and Paradoxes (Chicago: The University of Chicago Press, 1993).

[33] See Joseph Inikori, Africans and the Industrial Revolution in England: A Study in International Trade and Economic Development (Cambridge: Cambridge University Press, 2002).

[34] Jim Harter, World Railways of the Nineteenth Century (Baltimore: John Hopkins University Press, 2005), 2.

[35] Andrew Herod, Geographies of Globalization: A Critical Introduction (West Sussex: Wiley-Blackwell, 2009), 117.

[36] See Utsa Patnaik and Prabhat Patnaik, “The Drain of Wealth,” Monthly Review 72 (2021).

[37] Brinley Thomas, The Industrial Revolution and the Atlantic Economy: Selected Essays (New York: Routledge, 1993), 14.

[38] Ibid., 14-15.

[39] Ibid.

[40] Lee T. Wyatt III, The Industrial Revolution (Westport, CT: Greenwood Press, 2009), 53.

[41] Anthony Page, Britain and the Seventy Years War (London: Palgrave Macmillan, 2017), 95.

[42] Ibid.

[43] Ibid.

[44] Jaume Ventura and Hans-Joachim Voth, “Debt into Growth: How Sovereign Debt Accelerated the First Industrial Revolution,” National Bureau of Economic Research (2015), 2, https://www.nber.org/.

[45] Martin Hutchinson and Kevin Dowd, “The Apotheosis of the Rentier: How Napoleonic Wars Finance Kick-Started the Industrial Revolution,” Cato Journal 38 (2018): 255 -78.

[46] Ventura and Voth, “Debt into Growth: How Sovereign Debt Accelerated the First Industrial Revolution,” 2.

[47] It’s often mistranslated as “primitive accumulation” in English.

[48] See Amitav Ghosh, The Nutmeg’s Curse (Chicago: University of Chicago Press, 2021), 1–25.

There’s a lot of jargon in this post that I’ve defined and explained in prior posts and in the book. Before moving on, I would strongly recommend reading my very first post on Substack, as it will make understanding this post much easier. That post described the technological dynamics of conversional technologies under capitalism, and this post is a kind of companion post describing technological dynamics under valerism and how valerism would address the flaws and shortfalls of technological change under capitalism. For a short primer on valerism, I would recommend reading this post. Most of what's in this post is completely new work and builds on what I discussed in my book, which laid out a general vision for what the world could look like under valerism, but without taking a deep dive into one of the core issues: modularization.

Conceptual Introduction to Modularization

In the valerist order, the creation of a modular society, modularization, is one of the four great pillars, the other three being stabilization, socialization, and localization. In my book, I spent quite a bit of ink discussing stabilization and socialization, but I didn’t say enough about modularization and localization, and this post is an attempt to rectify that mistake. I will cover localization extensively in a future post. I emphasized in the book that these design principles are interdependent, meaning that we can’t pick and choose between them. If we want to create a global system based on stability and harmony with the broader biosphere, then we need to implement these principles jointly and simultaneously. Focusing on one but ignoring the other three would still lead to civilizational catastrophe. This is an important theme that will come up later in this post as well. Implementing aggregate modularization is extremely important, but not if it’s done without thinking about broader issues related to political economy, which the other pillars collectively address in a more direct and comprehensive way.

At a high level, modularization is the process of taking complex tasks or objects and breaking them down to smaller units that are more manageable. The classic toy example is a Lego set, where the same basic building blocks can be combined in an incredible number of ways to produce a wide array of different shapes and objects. The idea of modularization comes up often in many businesses processes, even for things like software, but the main focus of this post will be modularization as a design principle for the economic and industrial organization of consumer goods and conversional devices, which are physical systems that take energy from an input source and convert it to some other output. Examples of conversional systems would include smartphones, cars, refrigerators, and power plants.

Modularization can include interchangeable and reusable components for industrial manufacturing. It can include prefabricated modules for construction and housing. It can encompass interchangeable units for modular furniture and other consumer goods. The basic operating principles of a modular system are reusability, interchangeability, and prefabrication. To the extent that people pursue modularized strategies, they usually do so because of a desire to enhance efficiency and productivity. It’s also important to emphasize that modularization can encompass repeatable and automated tasks and processes, so it’s not just about physical items and their components. An object or a system can be modularized, but so can a manufacturing process.

These are the more familiar senses of the term, but the way I'm using it here is also a bit more expansive. First, modularization under valerism is meant to be universal and comprehensive, meaning that it’s supposed to cover all economic sectors and economic activities capable of being modularized, from extraction and manufacturing to transportation and products in end-use consumption. In other words, modularization is meant to be a macro-level design principle for organizing economic production, distribution, and consumption. Second, modularization under valerism is meant to change an economy’s internal dynamics of technological development, with the broader goal of more effectively constraining and regulating its energy scale, which is so important for limiting the extreme levels of waste and pollution that global civilization is currently dumping into the biosphere.

The goal of modularization under valerism is not merely to reduce waste and pollution or to recycle more materials, but to change the nature of technological development itself, as technological regimes are heavily implicated in the energy scales and efficiency dynamics of modern civilization. Third, valerist modularization is based on interoperability and integration. In a valerist framework, we shouldn't speak about modular systems so much as modular networks, technological environments where a broad range of conversional technologies are highly integrated. This way of using the term is a radical departure from previous work and emphasizes the harmonization of technological regimes by breaking down proprietary barriers and enhancing standardization. Fourth, valerist modularization places great importance on restricting the scope and scale of customization, which means creating relatively uniform product and technological standards.

A fundamental point I want to emphasize is that there are plenty of industries and sectors that are currently not modularized to the extent that existing technologies can achieve, largely because of political and financial constraints. In other words, the means and methods for modularizing these industries exist, but there are various political and economic forces that are working against the implementation of modular strategies. The most common barrier is simply high costs. Companies won’t undertake strategic modularization if they can’t afford it, which is why this shift simply cannot happen under the current capitalist framework. It’s the kind of major change that requires an entirely different kind of political economy, like the valerist system which I described in the book and which I have covered before in my other writings here on Substack.

Finally, let me mention how modularization differs from automation. The two concepts obviously overlap to a significant degree, as they are both often used to minimize human intervention in technological processes. But not all acts of automation need to be done through modularization, and not all acts of modularization necessarily lead to automation. It’s possible to modularize a particular system or process and still require a fair degree of human intervention, whereas the goal of automation is more specifically directed at removing the necessity for human action. As a design philosophy, modularization is also meant to highlight how the different parts and components of a system align and work together, in the hope of making the system more robust, resilient, and adaptable.

Concrete Examples of Modularization in Different Industries

Before going any further, it’s useful to have some concrete examples of modularization in mind. In this section, I’ll review a mishmash of historical examples, established technologies, and proposed initiatives across various industries, to see how modularization has played out in concrete reality at large scales of industrial production and end-use consumption. I will actually refer to some of these examples throughout the post as I’m making broader arguments.

Let’s begin with WindSpider, a Norwegian company that is trying to scale a clever self-assembly platform which uses an automated crane to lift the necessary components for building wind towers and turbines (see Figure 1). The company’s technology holds the promise of making offshore wind power construction much easier, but it’s still too soon to know what the ultimate impact on the industry will be. Nevertheless, it’s a compelling example of a groundbreaking modular manufacturing process, underscoring that modularizing repeatable tasks and processes should be a key initiative for the rapid deployment of renewable energy systems.

Modularized production in the auto industry is increasingly common and has been around for a long time by now. The hallmark example is probably Volkswagen’s MQB system, a modularized chassis that provides a common platform for building a wide range of vehicles, from hatchbacks to SUVs (see Figure 2). Volkswagen Group automakers have produced millions of cars with this platform, which has boosted production efficiencies and facilitated faster development cycles. The platform features a transverse engine design and other prefabricated components that collectively form a standard base on which to build a full car. The platform has enough flexibility that it can even accommodate multiple powertrain options and designs, from gas-powered vehicles to EVs.

The Nakagin Capsule Tower in Tokyo was a mixed-use building that contained offices and apartments during its operational phase (see Figure 3). Its striking features made it instantly recognizable. The building was eventually demolished and some of its capsules were repurposed for other uses. But it highlights the creative potential of modular construction, which has become increasingly popular in many places around the world. Modular housing, in particular, is often stereotyped as ugly and undesirable, but there’s no reason why we can’t achieve architectural ingenuity with modularized designs.

The Theory and Evidence on Modularization and Resource Consumption

In my book, I argued that, from a biophysical perspective, economic systems are best understood as conversional networks, meaning they consist of conversional devices, like smartphones and power plants, that consume energy from some source and convert that energy into useful tasks, from the transportation of vehicles to the production of consumer goods. These conversional devices are embedded in complex networks with dynamic feedback loops, and this interdependence ensures that changes to one part of the network reverberate to other parts of the network. When a change is made to a conversional device that affects the efficiency or power output of the device, I call that a spectralization. Some conversional spectralizations are so intense and catalytic that they diffuse widely and cause many upstream and downstream changes across the conversional networks of an economy, expanding its aggregate energy scale. Steam engines and AI chips are two major examples. This is the regime of catalytic spectralization that characterizes technological change in modern capitalism, and it’s so wasteful and energy-intensive that it’s putting the entire health and stability of the planetary biosphere in jeopardy. Universal modularization is the valerist answer to catalytic spectralization under capitalism, so let’s try to better understand how it works and why. A good place is to begin is by reviewing some of the empirical evidence on the benefits of modularization.

In a major study from 2012, Lu Aye and his collaborators demonstrated the ecological benefits of modular construction by analyzing an eight-story multi-residential building. They found that steel-structured prefabricated systems reduced material consumption by up to 78% by mass in comparison to conventional concrete construction methods. A study from South Korea found that modularized construction techniques produced 36% lower embodied carbon emissions relative to the traditional reinforced concrete process. The automotive sector is also increasingly relying on modular production platforms. In particular, the rise of Modular Electric Vehicle Platforms (MEVPs) is creating more efficient production cycles compared to traditional alternatives.

A 2016 review of the performance of modular buildings, including their energy performance, found that they did better than competing alternatives. The authors of this study noted that, in 1998, over 75% of all new annual homes in the United States were built on-site. The authors described modular buildings as follows:

Modular buildings are a set of modules that are built in an off-site fabrication center, delivered to the construction site, assembled, and placed on the permanent foundation. A modular building normally has multi-rooms consisting of three-dimensional modules. The modules are built and preassembled in factory environments…Upon completion by the manufacturer, these units are shipped to the site for installation on foundations much like a site-built project. About 85–90% of the modular construction is done off the construction site and the remaining work (10–15%), including foundations and utility hookups, is done on site. The application of modular construction is found mainly in general building construction, particularly apartment buildings, schools, hotels, student housing, hospitals, offices, single-family developments, correctional facilities, floating projects, and other buildings where units are repetitive.

Many studies have examined the energy use and emissions associated with specific modular buildings and constructions. For example, a 2018 study of a prefabricated building module in Sicily found that it had relatively low levels of carbon emissions and primary energy consumption, achieving a Net Zero Energy Building rating, defined as a building that consumes as much or less energy as a renewable source powering the building. Even modular steel bridges have been shown to have a much lower environmental impact compared to their concrete counterparts.

In a 1997 study, Rogers and Bottaci proposed creating "modular production systems" for manufacturing in the space of low-end and mid-range consumer goods, like toys and kitchen appliances. In manufacturing, principal assembly is the traditional method of production, where goods and components are processed through a fixed sequence of steps, kind of like an assembly line, that eventually culminates in a finished product. By contrast, modular assembly is a more flexible and adaptable strategy that focuses on building modules in off-site locations and the putting them all together at a dedicated facility. The authors propose creating four modules for their modular production systems: process machine primitives (physical machines doing the work), modular actuator elements (motion tasks), modular tooling (hardware to facilitate integration of primitives and actuator elements), and configurable control systems (communication network to make sure the modular systems are working well together). From this fundamental set of modules, they call for two new manufacturing principles: just-in-place parts production, focusing more on the production of components rather than final assembly, and automated production system design, for creating tools to model and perfect manufacturing processes. This is a model that won’t necessarily be effective everywhere, but holds immense potential across a wide range of industries, as it can significantly boost production efficiencies.

The implementation of common technological standards is important if we want to ensure interoperability and interchangeability, and thus directly implicates the importance of political economy in pursuing effective modularizations strategies. The adoption of the USB-C standard by the European Union is a canonical recent example, and soon enough all electronic devices sold in Europe will be required to use the same charging port. This policy shift in Europe finally forced Apple to drop its Lightning connector and to transition to USB-C starting with the iPhone 15. Now you'll be able to use one cable to charge everything, instead of keeping track of separate charging components for Apple and Samsung products.

Dominant corporations prefer proprietary products they can defend through patent litigation, to better ensure market exclusivity but also to hijack their customer base by making it difficult to switch to other products, supplemented by planned obsolescence that forces those locked-in customers to keep buying often. The main goal is to erect technological fiefdoms, which then encourages greater spectralization as hyper-specific conversional systems and technological standards proliferate within these feudal domains. Not all of these spectralizations will become mega-catalytic like steam engines or AI chips nowadays, but many of them will definitely put upward pressure on resource and energy consumption. The 1956 consent decree between AT&T and the US Department of Justice is probably the landmark example of government intervention in establishing technological standards. The agreement forced AT&T to license its patents for a reasonable royalty to any company that requested them, while also requiring AT&T to exit from general electronics and to just focus on phone-related technology. The subsequent diffusion of transistors led to the miniaturization of computers and paved the way for the era of personal computers, forever changing the world. Combined with the 1968 Carterfone decision that allowed external devices to electrically connect to the AT&T network, these regulatory changes collectively promoted interoperability among a wide range of networks and devices.

The kind of universal modularization that I’m advocating would make the spectralization of conversional technologies much more difficult. It would simply be impossible or impractical for a company to introduce major changes to conversional methods if those methods cannot be properly integrated into existing modular networks, which, while they would allow for technological variations, would only do so under tighter conditions. Consider Volkswagen’s MQB platform again. Because it’s designed to handle transverse engines, another car company can’t just come along and say they want to start using in-line engines all of a sudden. Of course, they could in principle, but then they’d have to redesign and overhaul much of their production process, and the costs of doing that would be so prohibitive that most companies wouldn’t even bother. This applies not just to companies that are owned by the Volkswagen Group, like Audi and Bentley, but also to other companies that have a strategic partnership with Volkswagen. For example, Ford has agreed to use the MQB platform for manufacturing the Tourneo Connect, which is produced by Volkswagen in Poland. And all of this leaves out potential intervention from the state, through legal mandates on new technological standards, nationalizations of existing industries and companies, enhanced regulatory frameworks, or combinations of all these things.

Modularization in Housing and Renewable Energy

Housing is a hallmark example of an economic sector that can be thoroughly and comprehensively modularized, provided the political will exists to make it happen. In the United States, modularization in the housing industry has never really taken off. Research has shown that less than 4% of the US housing stock was built using modular methods, in contrast to about 15% of homes in Japan and 45% of homes in many Scandinavian countries. As far as modularization goes, the United States housing market faces intense headwinds from a patchwork of local building codes, local architectural styles, consumer design aesthetics, and financing constraints from private banks, many of which would be unwilling to fund modular housing developments. As I’ve argued in an earlier post on housing, fixing these and other related issues would require overhauling our political economy, and I’ll bring up one of the primary examples from that post again.

When communist governments came to power after the Soviet victory in World War II, they launched a large-scale building campaign to provide housing for their devastated countries, using pre-fabricated housing components to construct things cheaply and quickly. The modularization of this housing boom created massive efficiency and productivity gains, allowing governments to build millions of units in a rapid amount of time. That historic push led to a wave of new apartment towers and city blocks, providing mass public housing to millions of people. The apartment blocks have often been stereotyped as ugly and unappealing in the West, reflecting the bourgeois sensibilities of those making the criticism. But for hundreds of millions of people, they were and continue to be a vital lifeline for cheap and stable housing, automatically eliminating one of the great stresses of life. And the market for these communist-era apartment homes remains quite strong. The Czech Republic is a great example, as the panelák building complexes from communist times still housed, as of 2005, roughly one-third of all Czechs, an astonishing 3.5 million people (see Figure 4).

Restricting customization through modular manufacturing and modular design strategies led to phenomenal economies of scale, allowing entire mini-cities to go up in months or a couple of years, a breakneck pace unheard of in much of the Western world, where the obsession with luxury aesthetics and customized designs often overrides practical issues around functionality and affordability. We need a total revolution in the meaning of housing, with the adoption of modularized design and manufacturing strategies that can produce rapid economies of scale and deliver major results in a short period of time. That means prioritizing things like functionality, affordability, and walkability over aesthetics and luxurious amenities.

Modularization in the renewable energy industry has typically taken the form of prefabricated components that are transported and installed on-site. Modularization is common for things like solar panels, wind turbines, and especially battery storage systems, which have made enormous gains and strides in the United States over the past few years (see Figure 5). The rise of modularized container designs contributed heavily to this expansion because it made these systems extremely scalable, another hallmark example of the sheer power of modularization to allow for the rapid deployment of products and resources in just about any given market. Of course, there’s no doubt that regulatory changes and additional financial support from the federal government spurred a lot of investments into battery storage systems, but modularization was a critical reason why all this new supply was activated so quickly.

Modularization, Planned Obsolescence, and Consumer Goods

Capitalists are very skilled at blabbering about efficiency and innovation as they push for financial accumulation and expand the underlying scale of their energy systems. In 1924, the world’s major light bulb companies formed a secret cartel called Phoebus, after the Greek god of light. The main purpose of the cartel was to mandate new manufacturing standards that would reduce the planned lifetime of an incandescent light bulb. By the early 1920s, rapid technological development had significantly improved the lifetime of the average light bulb, to the point where many of them could last as long as 2,500 hours. But that turned out to be a problem for the light bulb companies: better-lasting bulbs were hurting sales because consumers could wait a much longer time before buying replacements. To remedy the problem, the companies of the cartel ordered their engineers and manufacturers to produce light bulbs that would only last 1,000 hours. Their public justification for this move was an apparent desire to improve efficiency, but recently discovered internal documents from some of the companies make it apparent that the real reason for the shift was to boost sales.

The formation of the Phoebus cartel has sometimes been called the birth of planned obsolescence, which is the process of designing and manufacturing products to ensure their rapid breakdown, thus bringing consumers back to purchase additional versions of those defective products. Planned obsolescence is, in effect, a form of technological sabotage disguised as an apparent innovation. It has become a cornerstone of corporate planning under capitalism, impacting everything from personal vehicles to electronics. For example, Apple used to slow down older iPhones to encourage customers to buy newer variants. And when iPhones broke down, Apple made it very difficult to access certain critical components that had to be replaced, as a way of ensuring, once again, that customers ended up buying an entirely new phone instead of just replacing the defective component in question. Apple is also notorious for engineering a highly closed ecosystem of devices and technologies to ensure customer lock-in. Many companies play these kinds of games. Because high-tech commodities are designed to fail quickly, they must be continuously reproduced, repackaged, and reshipped so they can be sold to consumers. That process requires large amounts of energy. One of the biggest effects of planned obsolescence is to therefore reinforce the energy-intensive nature of capitalist production and distribution.

Universal modularization would end planned obsolescence by promoting greater reusability, interchangeability, and interoperability. If a component breaks down, we don’t need to throw away the entire system. We can simply replace the modularized component. This principle would extend across all economic sectors, even for things like consumer goods, which are frequently thrown away or entirely replaced when one of their components has a defect.

Consumer goods are a major area of future growth for modularization. There are many consumer goods that are already modularized, including furniture, but there are plenty of other consumer categories that could still benefit from it (see Figure 6). And even in areas where modularization is present, it’s not necessarily widespread. Many furniture pieces remain highly customized and specialized, making them difficult to repair or sustain over the long run. The success of modularization is therefore critically dependent on restricting the scope and scale of customization. If we modularize 50,000 different types of tables, the very process of scaling the manufacturing for these tables would backfire and lead to skyrocketing energy use. But if we modularize 200 different types, it’ll be much easier to master the production process for each one, to produce large numbers of spare components to high levels of quality, and to limit energy use and resource consumption from this sector. Enacting this project would require imposing common standards on consumer goods and sacrificing some desired aesthetics (see Figure 6).

Modularization can only be successful for our long-run energy and ecological goals assuming that it's paired up with restrictive customization. This goes back to the concept of a modular network: modularizing housing would also make it easier to modularize common things that go in most homes, like furniture, as well as appliances. The point is that when we think about what modularization could look like in a different kind of economy, we shouldn’t just be focusing on modularizing specific items, we should be modularizing entire conversional networks and technological ecosystems.

Modularization in Robotics

The recent public obsession around AI has been all about the generative AI wave epitomized by systems like ChatGPT, Copilot, and Grok. In the corporate world, there's now a major push towards agentic AI, autonomous systems that are supposed to help workers with everything from writing code to responding to customer questions. But in a very real sense, this is all low-hanging fruit. The ultimate AI dream is all about robotics, from humanoid robots that can function like human beings in a three-dimensional world to industrial robots that can automate and accelerate production. Robotics is the final frontier of AI. Robots are already all over the place. There are millions of them in factories worldwide. As of 2023, there were almost 2 million industrial robots in Chinese factories alone. When the global population plateaus and starts declining later in this century, robots will be necessary to perform vital tasks that human beings just won’t be available to do.

Modularization in robotics can encompass many different systems and processes, but I want to begin in a part of the industry that already has practical solutions in place but is struggling with scalability: warehouse automation. Modularizing warehouses also offers a great example to study the possibility of robotic modular networks and what they might look like in practice. Enhancing interoperability and interchangeability in this environment would require modularizing and integrating trucks, docking stations, pallets, robotic arms, mobile robots, and warehouse layouts, among other features and components. The goal is to modularize the network as a whole, not to focus on individual units while ignoring the broader picture.

Let's consider Symbotic, a major robotics company that operates AI-powered mobile robots for warehouse automation. On paper, its systems are highly modularized and offer flexibility and rapid scalability in different environments (see Figure 7). Symbotic robots are built from modules and feature many interchangeable components. And the entire Symbotic warehouse system is robust against individual points of failure. If some robots break down, others will continue operating, the kind of system-level resilience that’s a key benefit of modularization. But despite this level of modularization for its robotic fleet, Symbotic has struggled to deploy more than a few new warehouse systems per year in its recent history.

Symbotic robots are complex machines, and although they're designed with modular principles in mind, the company has outsourced part of its manufacturing to external contractors, so it can take a significant amount of time to produce a fleet of robots for a new warehouse deployment. Beyond that, every warehouse environment is a little bit different, so it might take a lot more effort to implement the automation system in one warehouse versus another. Even though Symbotic may have modularized the underlying components of its robots, the broader technological ecosystem in which those robots are operating has not been modularized, and therefore presents major challenges when it comes to rapid deployment and scalability.

Whether in robotics or any other industry, we need to be mindful about safety when implementing modularization. Sometimes it’s possible to modularize a system or a process but lose certain safety features along the way, and a great example is with autonomous driving. Waymo uses a more advanced suite of sensors than Tesla’s FSD. For example, Waymo uses LiDAR, radars, and a whole array of advanced sensors, in addition to a pre-mapped three-dimensional environment that gives the car a better sense of where it is and what it’s doing. By contrast, Tesla’s FSD uses an array of cameras and an AI-enhanced system to figure out where it’s going. Tesla’s system has the advantage of being more scalable and can operate in a broader range of environments. Waymo typically operates in more limited environments, but delivers a strong safety record in those environments, whereas Tesla has often been accused of major safety lapses, though it’s difficult to get an accurate picture of its actual safety record. Nevertheless, with autonomous driving or other areas in robotics, we’re going to learn through experience and trial-and-error where to strike the balance between the need to modularize and the need to maintain proper standards of safety and quality control.

There are many opportunities for further modularization in the robotics industry. Highly modularized systems like swarm robots are increasingly being used for construction, manufacturing, search and rescue, and exploration of challenging environments. Force-torque sensors are great candidates for further modularization across a range of robots, to a greater extent than what’s been currently achieved. In general, there are opportunities to modularize many kinds of actuators (the devices that provide motive force for the robot, like electric motors) as well as end-effectors (the devices that perform a specific task, like grippers). A major area of opportunity for enhanced modularization is in robotic charging stations. Just like phone chargers were once customized and hyper-specialized across proprietary domains, many companies in the robotics industry haven’t integrated their charging systems and ports. This is an easy place for government intervention to help set those common standards and ensure all robots are basically charged using the same set of processes, methods, and ports. But there will be limits to where modularization can take robotics. Some tasks and processes are so specific and hyper-specialized that modularizing them to a significant degree would be impractical. Electroactive polymers, increasingly being used as actuators in prosthetic limbs, are a great example, as they can be difficult to control and there’s a lack of uniform standards for ensuring their integration with other robotic systems. Like with everything else, figuring out what works in the best way is a matter of give-and-take and trial-and-error.

If it all goes according to plan, the robotics industry could automate much of its own manufacturing, turning everything into a self-organized recursive process. One day we could have robots building other robots, robots repairing each other, and robots forming an integrated technological ecosystem. Human intervention would become largely formal and perfunctory in this modular meganetwork.

Conclusion: Modularization in the Larger Valerist Order

Universal modularization is not a panacea. To be effective at achieving the broader goals of valerism at a global level, modularization must be accompanied by radical changes to global political economy. If that doesn’t happen, modularization will become just another empty slogan commandeered by corrupt capitalists who want to give the outward appearance of caring about ecological sustainability while doing practically nothing to actually achieve it.

For example, if we pursue modularization but don’t change the nature of international trade by insisting on localization, then what’s going to happen is that major corporations will prefabricate components in countries with cheap labor and ship those components halfway around the world so they can get to their final destination. Indeed, this is already happening in many industries, like fashion, but it would just happen at a far bigger scale if different countries imposed universal modularization on their economic systems but didn’t reform the broader contours of global trade and the supply chains that come with it.

Next, let’s consider how modularization would mix with socialization, or the transition from private ownership to public ownership. Sure, a government can always impose rigorous technological and product standards on manufacturers and consumers and while enforcing those standards through a tight regulatory regime. In plenty of cases, this kind of approach will work just fine. But in many other cases, especially energy-intensive industries with highly extended supply chains, where it’s not so easy to monitor the production and distribution of modular components, it actually makes far more sense to build internal state capacity and resort to nationalization, providing direct state ownership over the means of production. I explored arguments for socialization at a deeper level in the book itself, but the basic point I’m highlighting here is that modularization would often be easier to implement at larger scales if the state had more direct levers of control over production and distribution.

Finally, I want to address the relationship between modularization and the Jevons Paradox. I covered the latter extensively in my book and in my first post on Substack. Since modularization would lead to massive efficiency gains and rapid scalability, we need to be careful to make sure that it doesn’t backfire. Efficiency gains in the history of capitalism are often major drivers of more energy use and resource consumption. That’s where stabilization, the first pillar of valerism, comes in. In the book, I call for implementing caps on annual average daily per capita energy use, and I clarified that a valerist society should stick to something between 30,000 kilocalories and 60,000 kilocalories. Most industrialized economies are currently outside of this range, though some are very close to the upper bound while others are very far from it. Imposing these kinds of energy and resource constraints, as a fundamental goal of our political economy, would prevent the worst excesses of any modular strategies. As I’ve said all along, the goal isn’t for modularization to change the world all by itself. It simply cannot do that. The broader goal in valerism is to create a modular society that operates under the direction of a new kind of political economy, one that prioritizes the needs of the working class while also ensuring the stability and habitability of the planetary biosphere.

Kripke and Descriptivism

Kripke wanted to explain the relationship between a word and the thing to which the word refers. He began by attempting to refute the so-called descriptivist theory of proper names, generally associated with Gottlob Frege and Bertrand Russell, before offering his alternative. Kripke’s central thesis is that a name (like "Bob") refers to something (the actual person Bob) because there is a special kind of causal relationship between the use of the name and the thing to which it refers. The first part of the causal theory of reference states that the thing or object to which a name refers, known as the referent, is fixed by an "initial baptism," as Kripke put it. This initial baptism, also known as a dubbing, is an original act of naming. At this point, the name becomes a rigid designator for that object. As the name gets used, heard, and modified by different linguistic communities over time, it might change in various ways, but it always remains linked to the initial baptism through a causal chain. In offering this theory, Kripke wanted to claim that words really do mean something concrete. Language is not just a bunch of arbitrary conventions over how words are used and interpreted. And by extension, philosophy can actually make progress and get closer to "the truth."

The descriptivist theory says that the meaning of a proper name, like "John" or "Aristotle," is identical to a collection of descriptions that speakers have attached to that name. The referents are the objects or people that satisfy the given descriptions. For some concrete examples, consider the proper names Aristotle and Santa Claus. Aristotle might be described with statements like "the teacher of Alexander the Great," "the guy who wrote Nichomachean Ethics," and "Plato’s most famous student." If there’s someone who fits all or most of these descriptions uniquely, then Aristotle exists, or existed. Santa Claus might be described by terms like "the jolly old man with a thick white beard" and "the guy who lives in the North Pole with his elves." Again, if someone uniquely satisfies these descriptions, then that person is Santa Claus.

In his first attack, Kripke argued that a proper name is not necessarily identical to its list of descriptions, what has come to be known as the modal argument because it relies on the notion of possible worlds. Suppose there is a world where Aristotle died as an infant. In that case, none of the descriptions above would actually apply to him, but it would still be correct to call him Aristotle, according to Kripke. His basic point is that it’s not a necessity for an existing referent to satisfy its descriptions. Something can exist even if it doesn’t match up with its list of definitions.

Another argument that Kripke offered in his rejection of descriptivism is known as the epistemic argument. Consider the terms "Chancellor of Germany" as a description of the proper name "Friedrich Merz." Now the claim "Friedrich Merz is the Chancellor of Germany" seems like an a priori truth. In other words, it seems to be true just by the definition of the terms, without requiring any concrete experience to verify. But Kripke says that’s wrong; we need to go out into the world and check to identify the actual leader of Germany. Collectively, these arguments convinced many philosophers to abandon the descriptivist theory, although others merely modified it and presented new variations.

The Causal Theory of Reference

Kripke proposed another approach for understanding references. His theory is based on two principles: reference fixing and reference borrowing. The reference of a term is fixed by some initial event, the baptism. This could be something as simple as pointing to your dog and calling it "Kaycee." The unique physical actions associated with the creation of the term represent the first causal link that differentiates the reference of the term "Kaycee" from other references. Likewise, all other references are unique because their dubbing events happened under unique causal circumstances. They happened at different times, different places, and involved different groups of people. Of course, not everyone who uses a particular name was present at the baptismal event. We know what the name “George Washington” refers to even though no one alive right now was present at his dubbing. How do other speakers know what a name refers to? The basic answer, according to the theory, is that they got their ability from other speakers who knew about the name. Those speakers got it from others speakers, and so on until we go all the way back to the initial baptism. In effect, a causal chain develops that links the baptismal event to current speakers who use the name.

The causal theory had two big advantages over other rivals. First, it explained the concept of sense, defined by Frege as the "mode of presentation." One can identify sense with the causal chain that links each term to its corresponding referent. For example, the "Evening Star" and the "Morning Star" are different names for the same referent (Venus). But their sense is very different. One name means the star that appears at dusk and the other one means the star that appears at dawn. The causal theory apparently explains this difference in sense. Because the two names have different causal histories, their meanings diverged as they were transmitted from one group of speakers to the next. The theory also dissolves a puzzle about reference. Consider the statements "Venus is Venus" and the "Evening Star is Venus." Why do these identifying statements mean different things even though the names both refer to the same object? Because the causal chain from "Venus" to Venus is different than the causal chain from "Evening Star" to Venus. The different chains made different contributions to the meaning of the two sentences. In addition to these issues, the causal theory resolves or clarifies many other problems.

Kripke also argued for the necessity of certain propositions classified as a posteriori. Before Kripke, philosophers had generally assumed that only a priori statements, like "all bachelors are unmarried," were necessarily true. They made this assumption on the basis that such statements are not contingent on experience; they are simply true by definition and do not need to be verified. By contrast, a posteriori statements, like "it’s really cold outside," are considered contingent on experience and are not necessarily true.

Kripke deviated from the philosophical tradition and suggested that some a posteriori propositions must hold in all possible worlds, making them necessarily true. For example, the "Evening Star" and the "Morning Star" are different names that both refer to the same thing: the planet Venus. Thus, one can conclude that the "Evening Star is the Morning Star." This realization required experience; ancient peoples generally thought that the two celestial phenomena were unrelated. But because both names rigidly designate (or pick out) the same planet in all possible worlds, then the claim above is a necessary truth, assuming that necessity is defined in terms of applicability in all possible worlds. A similar case can be made for the statement "water is two hydrogen atoms attached to an oxygen atom." This realization also required empirical work and verification. But once this discovery was made, Kripke argued, the bond between the name and the referent would hold in all possible worlds. It could not fail to hold, making the statement necessarily true.

These claims flew in the face of Wittgenstein’s work on the philosophy of language. By the time of his mature intellectual phase with Philosophical Investigations, Wittgenstein had argued that linguistic meaning derives largely from social use and convention, not representation. This made it difficult to understand how we could speak about truth because the very word "truth" would vary in meaning depending on how people used it. Kripke was now saying that certain claims reached through empirical research could be necessarily true; they did not depend on how people used the words.

Contra Wittgenstein, he also argued that self-reference was perfectly cogent. He brought up an example that Wittgenstein himself had used: the standard meter in Paris, which defines the length of the meter. Scientists nowadays use the speed of light to set the length of the meter, but back then they just had a physical rod of a certain length, which was called "the standard meter." Wittgenstein claimed that one could not really determine whether the standard meter itself was a meter long or not. It makes sense to say that a piece of wood is a meter long, but it doesn’t make sense to say that the standard meter is a meter long. That would be a self-referential statement. The standard meter can be used to measure other things, but it cannot be used to measure itself, according to Wittgenstein. His fundamental point is this: the "meter" is a social and linguistic convention, not an objective feature of reality. The length of the meter is not open to empirical investigation. It’s just an arbitrary standard that human beings defined; one cannot "verify" it.

Kripke disagreed with this conclusion, arguing that it’s perfectly fine for an object to refer to itself. He imagined a scenario where thieves stole the standard meter and replaced it with another meter stick that happened to be slightly longer. If the authorities recovered the original standard meter, they could simply take it to a store that sells meter sticks and compare the two lengths. The “real” standard meter would be the same length as the meter stick in the store. The fake one used by the thieves would be longer, allowing people to distinguish the real from the fake. Kripke considered this kind of knowledge to be both contingent, because the standard meter rod could have been a different length, and a priori, because no experience was required to verify the length of the meter.

Critiques and Enhancements

Kripke’s work had an enormous influence on analytic philosophy. By rejecting the notion that philosophy is simply the analysis of language, as ordinary language philosophers had believed, Kripke rekindled the prospect that philosophy could actually reveal deep truths about the world, at least in some particular contexts. But this clever theory, like so many others in philosophy, definitely has a few flaws, some more severe than others.

Philosopher Gareth Evans identified one of the lesser issues: the problem of multiple groundings. He brought up a famous historical example to make his point: the name of Madagascar, which now refers to the large island just off the southeastern coastline of Africa. In the Middle Ages, the explorer Marco Polo landed somewhere in eastern Africa, probably in or near Mogadishu, during his travels. However, he mistakenly thought that he had reached an island. The chain of events is confusing and uncertain after this point, but it appears that the word Madagascar was a local designation for the region of Somalia where Marco Polo had landed. Marco Polo took this term and applied it as a reference for his imaginary island.

After Europeans reached the actual island off the southeastern coast of Africa in the early 16th century, Marco Polo’s Madagascar gradually became its name, even though it was not the island to which Marco Polo had originally referred. What we have here is a case of a reference shift. Marco Polo first baptized Madagascar as a name for an imaginary island near the Arabian Peninsula, but then the name wound up referring to an entirely different island to the south. How can the causal theory account for the secure grounding of references if references can sometimes shift by mistake? By simply acknowledging that these kinds of mistaken identity scenarios are likely to be found and eventually corrected, to say nothing of them being exceedingly rare.

The causal theory also struggles to explain instances where the meaning of words changes over time. References are dynamic and contingent, not static. For example, consider semantic drift, a well-known phenomenon in linguistics. The Albanian word möter means sister, but notice how it's very close to the English word mother. A long time ago, möter also meant mother in older varieties of Albanian, but over time the meaning of the word drifted, to the point where now it means sister. Defenders of the theory claim that it can still handle cases like these, as the transitions in meaning still happened because of various causal events chained together over time. But at the very least, situations like these weaken the importance of the initial baptism, as words can acquire meanings that depart radically from their baptismal event.

By far the biggest defect in the theory has turned out to be the qua problem. It goes something like this. Suppose you point to a dog for its initial dubbing and give it the name "Lucy." Why does Lucy refer to the dog and not to, let’s say, one of its body parts? Why does it not refer to the dog’s nose, tail, or eyes? It seems like the speaker’s intention has a big role to play here. When you point to something and give it an initial dubbing, there is some kind of implicit intention that narrows or delineates the focus of the reference. Furthermore, other people must also be able to understand the relation between the name and the referent at the moment of dubbing, or at a later moment, otherwise no one could possibly know what you’re referring to. Behind every act of reference, there are multiple layers of embedded social assumptions and presuppositions.

And this brings up an important point: reference partially presupposes intention and meaning, and meaning is generated from a complex web of social, cultural, and economic conditions. There’s nothing wrong with aiming for a causal theory of reference; the problem is that Kripke’s original theory was too narrow and restrictive. To rescue it, some philosophers brought back certain elements of descriptivism, recognizing that some prior descriptions in the mind of the speaker must apply to an object during the initial baptism. When you point to something and give it a name, people are able to understand what’s happening because they can apply their pre-existing cultural schemas. It would be silly to name a dog in a culture where dogs are not given proper names. Part of understanding the causal nature of reference implies understanding the wider causal matrix of the cultures in which we live. References are embedded within semiotic and cultural systems, and these in turn are embedded in the broader biophysical structures and interactions of the world.

Let's now revisit Kripke’s argument about the standard meter. The fact that an object can refer to itself does not mean that it can define itself. Only people can define things. Only people can look at the length of something and call it a "meter." The possible references of the standard meter do not exhaustively determine the meaning of the standard meter. Comparing the real standard meter to a meter stick from the store does not mean that the standard meter has defined itself; it just means you matched up two sticks that social convention dictated should be the same length.

The focus of this post is an extended critique of the intellectual background supporting the dominant version of libertarianism found in the United States, a version that's decidedly right-wing and very much favors corporate power. The two thinkers I will mainly focus on are Robert Nozick and Friedrich Hayek. At a superficial level, American libertarianism emphasizes the centrality of individual rights and liberties above and beyond collective or social obligations. Most American libertarians (from now on I will drop the "American" part as it's clear who I'm talking about) believe that individual rights are inalienable and fundamental. As a result, they support very limited government intervention in social life and generally hold that economic decisions should be made through free markets and decentralized networks.

In reality, libertarianism has always served as a lazy and reflexive ideological justification for capitalism. Libertarians want to make the social order look spontaneous and unorganized precisely to masquerade the ruthless power dynamics through which corrupt elites rule the world and make consequential decisions that affect everyone. Rather than being a philosophy which champions human freedom and individual liberty, modern libertarianism has been transformed into little more than ritualistic apologetics for corrupt monopolists and corporate warlords.

Nozick and Minarchist Libertarianism

In 1974, Robert Nozick published his classic book, Anarchy, State, and Utopia. The book champions a minimal state "limited to the narrow functions of protection against force, theft, fraud, enforcement of contracts, and so on." This view has come to be known as minarchist libertarianism, although it has antecedents in classical liberalism as well. Nozick thought that individuals should be seen as ends rather than means, echoing the main themes of Kantian ethics over utilitarianism. This position implies that the rights of one individual cannot be violated to avoid violations of the rights of other people. Individual rights are fundamental. He supported a “night-watchman” state that guarantees these basic rights but otherwise lets people get on with their day.

The book was primarily a response to the political philosopher John Rawls, whose 1971 work A Theory of Justice had laid out a framework for redistribution in human society that balanced between the competing ideals of liberty and equality. Nozick disagreed with Rawls, arguing that people have a right to what they produce and that taxation amounts to forced labor. He thought that the state had no right to redistribute resources to the poor. Nozick developed his entitlement theory as an alternative to the Rawlsian paradigm, arguing that redistribution is just only through consensual and voluntary exchange. Nozick maintained that justice in acquisition and justice in transfer were the only two principles by which people should be able to own property. When people deviate from these principles, by cheating the system or by enslaving other people, then justice in rectification is required to set things right. Nozick wrote that a distribution is just "if everyone is entitled to the holdings they possess under the distribution," based on the principles above.

Nozick begins his theoretical account by considering a Lockean state-of-nature, meaning an idealized collection of people living without government, perhaps in primitive economies like those familiar to our nomadic ancestors. Nozick believes that this approach offers the best way of assessing the differences between a society ruled by anarchy or by government. In his hypothetical, there is a "nonstate situation in which people generally satisfy moral constraints and generally act as they ought." Nozick disagrees with Locke that a grand social contract is necessary to establish things like money and civil society. He thinks that voluntary arrangements and interactions between people are all that’s needed to do the job.

Nozick believed that every individual in the state of nature can achieve a minimal level of welfare. However, his ideas, including justice in rectification, open the door for some kind of social assistance to the poor. Nozick wrote the following:

It is conceivable that in the normal operation of the economy, a private property regime might at some times, for some people, fail to provide access to this level of welfare when left to itself. If so, then justice, as the libertarian understands it, demands that the state act to correct the distribution of welfare generated by the spontaneous play of market forces.

Nozick continues by talking about the formation of a minimal state. He thinks that voluntary arrangements between people would lead to mutual-protection associations designed to protect groups from harm. Market pressure, economies of scale, and rational self-interest all compel more people to join the strongest association, which gets bigger and more powerful. Eventually it gets so powerful that it acquires an effective monopoly over the use of force, emerging as a de facto state. For Nozick, the two critical conditions for a state are monopoly over the use of force and the guarantee of universal protection. Crucially, he views the emergence of this force monopoly as morally permissible because no redistribution was required to bring it about. For independent individuals who choose to remain outside the confines of this "statelike entity," Nozick urges some kind of compensation as a way of balancing out the disadvantages associated with their lack of security and protection. His central point is that the minimal state should simply safeguard private property to facilitate market exchange. Anything beyond that goal would violate individual rights.

His account of utopia was extremely brief, about 30 pages. For Nozick, utopia seems to be a loose association of free and diverse communities. People can hop from one to the other as they see fit. Everyone can be whatever they want, as long as they don’t harm others. Critics charged that Nozick’s utopia was vague and lethargic.

The book also contained an attack against anarcho-capitalism, popularized by Murray Rothbard and other libertarian thinkers. Anarcho-capitalists believe that the state should not exist and that individuals should enter into contracts with private companies for all social services. Nozick held that anarcho-capitalism would eventually descend into minarchist libertarianism because some private corporations would concentrate enough power through competitive advantage to function as de facto states. The history of capitalism proves that Nozick was absolutely correct on this particular issue, as many dominant corporations, from the East India Company to the United Fruit Company, often accumulated so much power that they indeed became de facto governments.

Liberal and left-wing critics were generally merciless against this work, although it did receive praise from some conservatives and libertarians who wanted to demolish the welfare state. Nozick himself largely dropped out of the political philosophy arena and moved on to other projects, leaving most of his critics unanswered. A common line of criticism was that the book offered nothing but useless abstractions. Nozick has very little to say about history, sociology, and anthropology, even though the subject of the book seems to bear a strong relation to these fields. Instead, Nozick analyzes a bunch of disconnected individuals as if they were particles in a physical system settling towards some perfect and rational equilibrium. The complex and emergent features of human society are completely ignored.

Not all self-described libertarians approved of this book. Rothbard heavily criticized it in a 1977 article. Rothbard made several points against Nozick. He argued that individual rights are inalienable and no existing state could be justified to curtail them. He also said that the compensation principle was unbounded and would result in unlimited despotism. In 2000, the liberal scholar John Rawls also criticized Novick in his Lectures on the History of Political Philosophy. Rawls made a very simple and elegant point: all voluntary transactions come with small imperfections and inequalities. If left unaddressed, these imperfections will accumulate over time, turning into major structural deficiencies that could cripple Nozick’s utopia. His argument has received support from the work of numerous major economists, including George Akerlof and Joseph Stiglitz, who have detailed and analyzed a variety of asymmetric relations in market transactions.

The central point is that even if people are engaged in consensual and voluntary transactions, plenty of opportunities exist for lying, cheating, stealing, and manipulating. If a random caller persuades grandma to hand over her hard-earned Social Security money to invest it in some get-rich quick scheme, the resulting transaction might be consensual and voluntary, but we nevertheless recognize that it’s wrong and unfair. If a corrupt car dealer knowingly sells a crappy car to an eager yet oblivious buyer, the transaction might be voluntary and non-coercive, but we all realize that the buyer got played. If a shady financial advisor convinces a client to buy a worthless stock because there’s a huge commission riding on the trade, the resulting transaction will again be mutual and voluntary, but it’s obvious that the advisor manipulated his client by using false information. The notion that all voluntary transactions are automatically ethical is completely absurd, precisely because the real world is teeming with informational asymmetries that give certain people an advantage over others in any given exchange.

The libertarian economists Tyler Cowen and Alex Tabarrok argued in a 2015 essay that asymmetric information is becoming less relevant because of technological innovation. They brought up some examples to support their argument, such as the declining frequency of odometer fraud in used car sales. Their main point is that because increased technological innovation has reduced the importance of asymmetric information, we don’t need additional government laws and regulations to achieve the same goal. But they are flatly wrong.

Technological innovation has massively exacerbated asymmetric information, because it’s allowed the rich and powerful to protect and amplify their advantages over rivals and competitors by deploying new methods of evasion and subterfuge. For example, Volkswagen fooled government regulators by controlling its diesel engines with highly sophisticated software that could quietly cheat during emissions tests. For another example, large brokerage firms on Wall Street operate advanced and secure dark pool market exchanges that can facilitate block trades through high-frequency algorithms for major institutional investors, all away from public view. For these elite investors, ensuring organized secrecy and informational asymmetry is exactly the point, as they don’t want to alert other investors and the public at large about the prices involved in their deals, prices that in some cases can differ sharply from those in public exchanges like the NYSE or NASDAQ.

High-frequency trading also makes it far easier to commit spoofing, the practice of placing vast numbers of fake orders that are canceled before execution, all as a way of generating artificial demand for a particular stock or asset. In 2020, JP Morgan Chase paid a fine of almost $1 billion to the federal government for carrying out widespread order spoofing in precious metals futures contracts. Although JP Morgan happened to get caught in this one case, the vast majority of spoofing incidents are never publicized or prosecuted, precisely because high-frequency trading allows sophisticated investors to game the system and get away with it. Furthermore, many of the examples given by Cowen and Tabarrok are also heavily focused on informational disparities at the point of market exchange. But so many of the major informational asymmetries in modern capitalism are actually found upstream in the production process. For example, a massive global auto parts cartel in the 2000s had substantial downstream effects on car prices, but consumers knew absolutely nothing about it. Simply put, even if Cowen and Tabarrok are correct that informational asymmetries are declining at the point of market exchange for consumer goods, that would prove nothing about informational asymmetries at different points along the economic chain, especially for industrial production, logistics, and distribution.

It’s also worth questioning to what extent market transactions are actually voluntary. Our participation in capitalist markets might superficially reflect individual desires and utilities, but those desires are themselves the products of complex social relations. Capitalists spend hundreds of billions of dollars every year on advertising designed to convince people to buy useless junk. The biggest corporations often engage in monopolistic practices to crush potential competitors and maintain market dominance. They also influence the political process via campaign contributions and concerted lobbying efforts. It pays off too, since governments caught in the claws of private capital routinely pass laws that weaken regulations and reduce taxes for certain industries. The idea that such intense elite pressure plays no role in what common people want to purchase is totally absurd. Thus, the position that voluntary exchange could be the basis for social organization is wrong, because all “voluntary” economic transactions between individuals in a capitalist system are actually a product of pre-existing social rules, relations, and power structures arising from the historical dynamics of that system. Markets are created and embedded within the wider social fabric. That fabric includes class hierarchies, corporate monopolies, and the institutions of the state. All of our personal choices are constrained by these wider social realities.

Now let’s get to the most interesting part. One of the biggest flaws in Nozick’s project is a problem endemic to libertarianism more broadly. The obsessive focus on voluntary exchanges and transactions largely obscures the process of production that makes goods and services available for market exchange in the first place. Once we shift focus to the conditions of production, it becomes much easier to see the effects of corruption and injustice. Consider chocolate and coffee, two precious commodities that have become addictions for hundreds of millions of people around the world. Many researchers and journalists have extensively documented how both industries are plagued by issues with slave labor and child labor in their supply chains.

In West Africa, for example, young children are often lured into working on cocoa plantations with false promises, and many are outright kidnapped from their villages and families. Once they get to the plantations, they live in squalid conditions and are forced to use machetes to pry open the bean pods that contain the lucrative cocoa beans, from which chocolate is eventually made. Constantly handling these dangerous weapons means that severe injuries on the job are the rule, not the exception. Many children and slaves are also physically beaten if they don’t work fast enough. To ensure low prices and high revenues, the farm owners need to sell as many cocoa beans as possible to their corrupt corporate overlords in the West. The Western companies don’t really care about the exploited pool of cheap labor because that cheap labor helps keep cocoa prices down, thus reducing overall production costs for the companies.

After the cocoa beans leave Africa, they’re processed and combined with ingredients like sugar and milk to produce the finished chocolate goods sold to consumers worldwide. Unsurprisingly, these finished products are sold at much higher prices than the unprocessed cocoa beans, boosting corporate profits. Buy low and sell high; that’s the basic strategy of the Western companies that dominate the chocolate industry worldwide. When you walk into the store and buy a chocolate bar, your decision may be wholly consensual and voluntary, but you’ll never know about the blood, sweat, and tears that it took to make your precious chocolate. A former slave named Drissa was once interviewed about his experience on a cocoa farm and said that “when people eat chocolate, they are eating my flesh.”

Even if all acts of market exchange are purely voluntary and morally acceptable, it would emphatically not justify the resulting distribution of wealth, because that distribution also heavily depends on the process of production that’s feeding the market with goods and services. And in the real world, that process is rife with domination and exploitation. Contrary to the wishful thinking of libertarians, the world is not a nice playground where all the children hold hands and get along. Yes, there is plenty of cooperative behavior in the world, but there’s also plenty of conflict and struggle, and those conflicts have a huge influence on the ways in which people cooperate and interact. In the current capitalist order, no form of "voluntary" economic interaction can be decoupled from the dynamic power relations that shape those interactions in the first place. This is one of the fundamental mistakes that libertarians make in their attempts to describe the world as it actually exists today.

Hayek and Spontaneous Orders

If there's one concept all good libertarians know by heart, it's the idea of spontaneous order, first popularized by the Austrian economist Friedrich Hayek. In an economic context, a spontaneous order is a social order that emerges from the numerous interactions of individual people following their own rational self-interest. Spontaneous orders are not supposed to be designed or planned; they are presumably the product of distributed and decentralized networks sharing goods, ideas, and information. Libertarians often say that society is the product of human action, but not of human design. In The Fatal Conceit, a book released in 1988, Hayek puts it this way:

To understand our civilization, one must appreciate that the extended order resulted not from human design or intention but spontaneously: it arose from unintentionally conforming to certain traditional and largely moral practices, many of which men tend to dislike, whose significance they usually fail to understand, whose validity they cannot prove, and which have nonetheless fairly rapidly spread by means of an evolutionary selection, the comparative increase of population and wealth, of those groups that happened to follow them. The unwitting, reluctant, even painful adoption of these practices kept these groups together, increased their access to valuable information of all sorts, and enabled them to be ‘fruitful, and multiply, and replenish the earth, and subdue it’...This process is perhaps the least appreciated facet of human evolution.

In the same book, he also writes:

The main point of my argument is...that the conflict between, on one hand, advocates of the spontaneous extended human order created by a competitive market, and on the other hand those who demand a deliberate arrangement of human interaction by central authority based on collective command over available resources is due to a factual error by the latter about how knowledge of these resources is and can be generated and utilized.

Hayek is presenting a false dichotomy here. There are plenty of people on the left who don’t want a “central authority” to have collective control over all available resources. Likewise, there are plenty on the right who don’t want everything to be controlled by a “competitive market” entirely divorced from state management and regulation. He’s off to a bad start already, but let’s put this point aside for now and we’ll come back to it later.

Hayek brought up many examples of social orders that he considered to be spontaneous, such as language and money. He thought of language as a spontaneous order because it isn’t controlled by any individual. No single person created the English language. In this view, language develops spontaneously from the complex and unpredictable interactions of individual people. There’s no central bureaucracy that creates new words or tells people the rules of grammar. Likewise, Hayek thought of the monetary domain as a set of signals about supply and demand. For him, prices and profits are objective pieces of information about the real domain, about things like productivity and efficiency. Prices are the signals that a competitive market sends out about the state of the economy. Based on those signals, people and businesses can then make decisions about how to save, invest, or do just about anything else. The economist Milton Friedman later brought up the example of a pencil in one of his speeches, arguing that a pencil is the spontaneous product of thousands of people around the world, most of whom don’t know each other and may not even like each other.

For people like Hayek and Friedman, central economic planning always leads to disaster because it does a poor job of efficiently allocating scarce resources. That’s where spontaneous order comes in to save the day. Only the market economy and its creative dynamism can allow people to make the choices that produce greater wealth and happiness in the long run. Friedman and the Chicago Boys apparently believed in this vision so much that they were even willing to support brutal dictatorships in order to make it happen, because nothing says limited government more than a dictatorial regime imposing market orthodoxy on an unwilling population.

In his 1945 paper “The Use of Knowledge in Society,” Hayek defended the price system for producing a spontaneous order based on the organization of localized bits of knowledge. He begins by addressing what he considers to be the central problem of economics, which is how individuals should utilize incomplete knowledge to best serve their particular ends in life:

The peculiar character of the problem of a rational economic order is determined precisely by the fact that the knowledge of the circumstances of which we must make use never exists in concentrated or integrated form but solely as the dispersed bits of incomplete and frequently contradictory knowledge which all the separate individuals possess. The economic problem of society is thus not merely a problem of how to allocate “given” resources...if “given” is taken to mean given to a single mind which deliberately solves the problem set by these “data.” It is rather a problem of how to secure the best use of resources known to any of the members of society, for ends whose relative importance only these individuals know. Or, to put it briefly, it is a problem of the utilization of knowledge which is not given to anyone in its totality.

Hayek argues that knowing about general rules and principles isn’t the only thing that matters for organizing an economy. It’s also important for individuals to know about specific and unique circumstances. He writes that “it is almost heresy to suggest that scientific knowledge is not the sum of all knowledge. But a little reflection will show that there is beyond question a body of very important but unorganized knowledge which cannot possibly be called scientific in the sense of knowledge of general rules: the knowledge of the particular circumstances of time and place.” He continues with the argument:

It is with respect to this that practically every individual has some advantage over all others because he possesses unique information of which beneficial use might be made, but of which use can be made only if the decisions depending on it are left to him or are made with his active cooperation. We need to remember only how much we have to learn in any occupation after we have completed our theoretical training, how big a part of our working life we spend learning particular jobs, and how valuable an asset in all walks of life is knowledge of people, of local conditions, and of special circumstances.

Hayek then explains that the kind of economic regime which follows from the conclusion above is a system that leaves economic decisions up to the people who are actually familiar with these local conditions and circumstances, as opposed to a central board of planners that tries to integrate all available knowledge and issues orders after doing so. Hayek argues that someone doesn't need to know everything about global economic conditions in order to function and run a business. He just needs to know how local conditions are affected and respond accordingly. Hayek puts it like this:

There is hardly anything that happens anywhere in the world that might not have an effect on the decision he ought to make. But he need not know of these events as such, nor of all their effects. It does not matter for him why at the particular moment more screws of one size than of another are wanted, why paper bags are more readily available than canvas bags, or why skilled labor, or particular machine tools, have for the moment become more difficult to obtain. All that is significant for him is how much more or less difficult to procure they have become compared with other things with which he is also concerned, or how much more or less urgently wanted are the alternative things he produces or uses. It is always a question of the relative importance of the particular things with which he is concerned, and the causes which alter their relative importance are of no interest to him beyond the effect on those concrete things of his own environment.

In an important passage that's worth quoting extensively, Hayek explains how the price system works in practice and why it’s allegedly useful:

Fundamentally, in a system in which the knowledge of the relevant facts is dispersed among many people, prices can act to coordinate the separate actions of different people in the same way as subjective values help the individual to coordinate the parts of his plan. It is worth contemplating for a moment a very simple and commonplace instance of the action of the price system to see what precisely it accomplishes. Assume that somewhere in the world a new opportunity for the use of some raw material, say, tin, has arisen, or that one of the sources of supply of tin has been eliminated. It does not matter for our purpose...which of these two causes has made tin more scarce. All that the users of tin need to know is that some of the tin they used to consume is now more profitably employed elsewhere and that, in consequence, they must economize tin. There is no need for the great majority of them even to know where the more urgent need has arisen, or in favor of what other needs they ought to husband the supply. If only some of them know directly of the new demand, and switch resources over to it, and if the people who are aware of the new gap thus created in turn fill it from still other sources, the effect will rapidly spread throughout the whole economic system and influence not only all the uses of tin but also those of its substitutes and the substitutes of these substitutes, the supply of all the things made of tin, and their substitutes, and so on; and all this without the great majority of those instrumental in bringing about these substitutions knowing anything at all about the original cause of these changes. The whole acts as one market, not because any of its members survey the whole field, but because their limited individual fields of vision sufficiently overlap so that through many intermediaries the relevant information is communicated to all. The mere fact that there is one price for any commodity...brings about the solution which...might have been arrived at by one single mind possessing all the information which is in fact dispersed among all the people involved in the process.

Hayek concludes by summarizing how the price system disseminates knowledge and information:

We must look at the price system as such a mechanism for communicating information if we want to understand its real function...The most significant fact about this system is the economy of knowledge with which it operates, or how little the individual participants need to know in order to be able to take the right action. In abbreviated form, by a kind of symbol, only the most essential information is passed on and passed on only to those concerned. It is more than a metaphor to describe the price system as a kind of machinery for registering change, or a system of telecommunications which enables individual producers to watch merely the movement of a few pointers, as an engineer might watch the hands of a few dials, in order to adjust their activities to changes of which they may never know more than is reflected in the price movement.

Assessing the Libertarian Arguments

Now let’s analyze Hayek’s arguments. There are many possible refutations here, but the easiest is to point out examples of complex civilizations that efficiently allocated resources without using prices at all. Andean civilizations in South America, such as the Tiwanaku and Inca, developed complex states and empires without the corresponding rise of a large financial class. The state controlled the distribution of resources, handing out food and equipment as necessary, and people usually paid taxes to the government in the form of labor, by doing some kind of work. Anthropological data shows that these systems thrived for centuries and they appear to have worked very efficiently, in the sense that they consistently avoided extreme resource shortages.

Leaving ancient history aside, one can also point out that markets under modern capitalism have routinely produced oligopolies and monopolies, creating many inefficiencies and externalities along the way. In other words, capitalism itself has a tendency to centralize economic planning in the hands of a few powerful corporations, which then control the distribution of resources for other individuals and corporations. Contemporary examples would include the likes of Amazon and Walmart, both of which establish prices through central planning for millions, or perhaps billions, of different commodities.

Hayek was wrong to view prices as innocent markers of supply and demand, as impartial signals of information about the state of the economy. Prices function more like symbolic quantifiers of social power, as mediated by class struggles, monopolies and oligopolies, and institutional rivalries. Prices are imposed and administered by those who have economic power. Capitalists price their commodities to beat out the profit rates of their competitors, to seize control over new markets against established rivals, and to extract profits from their hard-working labor force. Capitalists are not that interested in distributed networks of knowledge; anyone who knows anything about Google knows that its business model is to accumulate and centralize as much data as possible, which it then markets and sells to salivating advertisers. Capitalists are interested in controlling the social distribution and utilization of economic resources. More specifically, they are interested in augmenting their power by trying to organize society on their own terms, and that process includes pressuring governments and workers to accept their demands through a wide array of threats and coercive actions.

This fact brings up the cardinal sin of libertarianism: it has the wrong dialectics. The relevant social struggle that allocates economic resources isn’t between the state and the market. It’s between the dominant classes and those they oppress. As Marx and Engels famously wrote in The Communist Manifesto, “the executive of the modern state is nothing but a committee for managing the common affairs of the whole bourgeoisie.” The state does not exist independently of society. It’s a vehicle for social control and organization, and different factions in society have always squabbled over who gets to drive it.

It's crucial to emphasize this last point because it underscores why libertarianism is hopelessly flawed at the descriptive level, as an attempt to describe how the world actually works. It's a simple fact of modern history that the state and the accumulation process under capitalism are profoundly intertwined. The state does not merely “protect” private property; it can also actively create it. In the 1930s, at the height of the Great Depression, the United States government banned companies from purchasing their own stock, judging this kind of activity to be a form of stock manipulation. But in 1982, after the collapse of the New Deal coalition allowed Reagan to obtain power, the government abandoned the lessons of the past and started to allow a restricted version of stock buybacks. The predictable result was that companies started pouring money into buying their own stocks, sending valuations sharply higher with little regard for actual performance or “economic fundamentals.” In the 1990s, the Clinton administration issued new rules about CEO salaries that wound up incentivizing companies to pay their executives through lucrative stock packages. Through these and other actions, the state encouraged massive wealth redistribution toward capitalists and away from workers. Once the apologists of capital took over the state, there was little doubt about who would benefit. As I write in my book, The Physics of Capitalism, the modern bureaucratic state is the linchpin of the capitalist economy:

The state provides critical top-down constraints on economic activity, and thus exerts enormous amounts of influence over the cycles of production and distribution. The concept of a “free market” is largely an abstraction because virtually all governments have a strong impact on the dynamics of market activity. Governments decide what counts and doesn’t count as property; they also enforce those property rights. Governments have considerable influence over monetary supply and the availability of credit; central banks have saved the global financial system from collapse at least twice in the past twenty years. Governments define the rules governing market operations. Governments can even create new global markets for domestic companies through warfare and other forms of strategic competition, like sanctions, embargoes, and blockades. Trade and commerce cannot be decoupled from state power. Likewise, the exercise of state power cannot be decoupled from the class dynamics that constrain the distribution of labor and wealth. The state does not act in a vacuum; its actions are shaped or influenced by various kinds of social and class struggles. The state is a thunderous battleground among competing economic classes and social groups. Economics, especially in the modern world, cannot be understood separately from the collective actions of the state.

Let’s further explore this argument by considering a favorite libertarian bogeyman: eminent domain, the process of government seizing private property for public use or public benefit. Starting in 1999, the city of Toledo in Ohio waged a multi-year campaign to clear out homes and businesses so that auto giant Chrysler could expand its manufacturing plant. Toledo ultimately kicked out 83 homeowners and 16 companies, including small auto businesses, while giving Chrysler massive tax breaks and other benefits totaling over $280 million, all in the name of saving jobs. A libertarian might look at this situation and blame the city government for screwing over hard-working Americans. But the real villain here is Chrysler, because it exploited its size and economic power to pressure the city government into handing over a huge amount of land. And not only that, Chrysler did it without paying a dime. Instead of buying up these homes and businesses with its own money, Chrysler simply socialized the costs and risks of the project by turning to the city government. Toledo took out a $28 million loan from the federal government to cover the expenses associated with relocating the homeowners.

That’s how things work in America. Dominant capitalists like Chrysler routinely use the threat of eliminating local jobs to make county and municipal governments do anything they want. Consider the long-running symbiotic relationship between General Motors and Detroit. In 1981, Detroit used eminent domain to forcibly displace about 4,000 people and many businesses in a working-class area, transferring the land to GM so that the auto company could build a new factory. GM promised thousands of jobs, about half of which did not actually materialize. The aggressive use of eminent domain by Detroit, largely at the behest of Big Auto, was one of the factors that contributed to the city’s social and economic decline.

One can further indict the entire auto industry for pushing the Federal Highway Act through Congress in 1956. The interstate highway system that resulted from this legislation absolutely decimated poor communities throughout the country, and that was actually one of its key objectives: to uproot and destroy black neighborhoods. Roughly half a million households and over a million people were displaced through eminent domain proceedings associated with the interstate highway system. Communities lost churches, parks, and critical businesses that anchored economic life. In the first half of the twentieth century, Miami’s community of Overtown was a vibrant cultural mecca known as the “Harlem of the South.” But the expansion of I-95 after the passage of the 1956 bill led to the destruction of 87 acres of housing and commercial property. Only about 8,000 people out of an original 40,000 remained in Overtown after the highway’s expansion, according to Richard Rothstein in his book The Color of Law.

That’s the fundamental point here. Eminent domain is not about government overreach; it’s about class struggle. It’s about big corporations and their political allies uprooting and displacing small businesses and homeowners that don’t have the power to fight back. It’s about the rich grabbing and destroying poor communities whenever they need more land. The problem isn’t the government; the problem is class domination and the insidious ways it works through government.

And then there’s Hayek’s example with tin. As we’ve seen, libertarians fondly believe that exchange markets function as a kind of “invisible hand” that efficiently orders and organizes society. Part of what goes into this fantasy is the idea that price changes reflect supply and demand changes, that the nominal domain is a giant mirror of the real domain of physical production and supply. But as Hayek himself wrote in one of the passages above, libertarians and neoclassical economists don’t seem to care much about what’s causing the changes in supply and demand. This is a tremendous flaw in their worldview, because contrary to Hayek’s gibberish, the underlying causes of market activity are pretty much the only things that matter for business planning.

Let’s go back to the tin example. If you knew that the last available tin reserves on the planet were about to be depleted, you wouldn’t simply “economize” tin, you would completely replace it with something else. That knowledge would profoundly affect your business strategy going forward. Libertarians might say that my counterexample is unrealistic, so let’s talk about some actual history then. For decades during the Cold War, tin prices were heavily influenced by a global cartel involving multiple countries on different continents. Founded in 1953, the International Tin Council artificially inflated tin prices beyond what they would’ve normally been with a more decentralized market. Financed by both producer and consumer member states, the ITC would buy up large quantities of tin in global metal markets to control the available supply and to ensure overall price stability. However, tin prices did rise a good amount in the 1970s, largely because individual producer nations, especially Malaysia and other Asian countries, started violating the terms of the cartel. In the 1950s and 1960s, tin prices were relatively stable as most member nations followed the cartel’s policies. But from 1970 to 1980, the price of tin virtually tripled, going from roughly $5,000 per metric ton to $15,000 per metric ton. The cartel broke down in the 1980s after a sharp fall in demand for tin made it virtually impossible to prop up tin prices. The ITC simply didn’t have enough money to continue buying tin in the necessary quantities to guard its inflated price.

Now let’s get back to Hayek and his example. Suppose you were an entrepreneur in the metals industry who during the 1960s and 1970s expected to see tin prices drop, both because you weren’t aware that the ITC even existed and because you thought that the price increase was caused by temporary supply disruptions or bottlenecks. If this was your sincere belief, then you probably wouldn’t have been very judicious in the consumption of tin, because you’d expect to buy a lot more of it quite soon once prices dropped. Instead of conserving tin because of an expected price increase, you would’ve used it like crazy thinking that it was going to be cheap again in a short time. But if you were someone who was paying attention and knew that the price of tin in global markets was largely rigged by a multinational cartel, you would’ve made very different business choices, like saving tin as much as possible to lower production costs, or maybe seeking an alternative material. The point is simple: what really matters is the cause of the changes in supply and demand, not the changes themselves. It’s only by knowing these causes that firms and businesses can properly plan for the future.

What this example also highlights is Hayek’s flawed assumption about the supposedly widespread dispersal of economic knowledge. In reality, the world is full of informational asymmetries; some people and businesses simply know much more than others, and they will use that relative advantage to try and eliminate their competitors. What’s more, they’ll go to great lengths to preserve these knowledge hierarchies and asymmetries, from seeking intellectual property protections like patents to having people sign non-disclosure agreements. The goal for a high-flying capitalist is not to disperse and diffuse knowledge to different people, but to concentrate and limit knowledge as much as possible. Making important information exclusive and difficult to access is part of what allows the ruling classes to rule. It’s what gives rise to things like insider trading and stock market speculation.

Economic knowledge is not dispersed; it’s socially stratified and exploited by elite groups to reinforce their relative advantage over marginal groups. It’s precisely these social and institutional hierarchies that mediate and constrain the levels of supply and demand that prevail in the economy. Class power is what structures the nominal domain and market activity. The price system isn’t there to solve any fundamental economic problem; it’s a symbolic expression of social power, engineered and manipulated by the wealthy in whatever ways they can get away with for the purpose of even greater accumulation, as was painfully evident in the aftermath of the pandemic. It’s exactly the kind of sentiment that Brian Niccol, the CEO of Chipotle, expressed to investors in a Q4 2021 earnings call:

We’re pretty fortunate with the pricing power we have...so we have more room to take the price as we need to. We see no resistance to date with the levels that we’re currently at.

During the coronavirus pandemic, economists were busy parroting corporate propaganda about rising prices being a reflection of higher production costs. But given that net post-tax corporate profits in 2021 were at their highest level in recorded history, it’s far more reasonable to conclude that corporations used their pricing power to jack up prices as much as they could, in the absence of serious resistance from governments and consumers. Data from the Bureau of Economic Analysis showed that corporate profits in 2021 went up 35%. That’s why executives at the steel producer Steel Dynamics, which saw 2021 profits increase by 809%, said during an earnings call that the company was “not materially affected by inflation.” According to them, higher prices “exceeded” the increase in supply chain costs. The Canadian food company Nutrien, a big producer of fertilizers, saw 2021 profits shoot up by $1.2 billion on “higher selling prices that more than offset higher raw material costs and lower sales volume,” according to an earnings call. Notice that last part: despite higher supply costs and lower demand for its products, Nutrien nevertheless scored a huge profit purely on the basis of higher prices. Chevron told investors that its profit spike of 240%, driven by soaring gas prices, was part of the “best two quarters the company has ever seen,” leading the company to give a dividend increase and to promise that it would keep production low in order to inflate prices even further.

There were many stories during the pandemic about limited housing supply, but hardly any mention of the fact that some of that limited supply was caused by new homebuilders like Lennar deliberately building very few homes, even though they had more than enough capacity to deliver more supply to the market. Lennar CEO Jon Jaffe told investors in an earnings call that “we could sell another 1,000 homes in the quarter if we wanted to without too much effort. It just doesn’t make sense to do that.” Diamondback Energy CEO Travis Stice promised to put “returns and shareholders first” by keeping production low. Stice added that “no wants to see that shareholder return program put at risk with volume growth.” The head of research for the bank Barclays stated that “the longer inflation lasts and the more widespread it is, the more air cover it gives companies to raise prices.”

For company after company, the story was always the same: commit as much price gouging as possible, then turn around and blame inflation, abetted all along by a plutocratic corporate media environment that offers no serious pushback. The pandemic experience underscored that prices are symbolic social constructs embedded in the wider social and political relations that collectively manage the distribution of economic resources.

There are many other problems with the arguments from Hayek and the libertarians. First, it’s virtually impossible to decouple human actions from human plans and designs. People take consequential actions based on complex intentions and motivations. For example, when CEOs and other executives make decisions about corporate investments, they are usually doing so in the context of a pre-existing strategic blueprint. And very crucially, they’re engaged in top-down planning. Corporations generally don’t make decisions by letting workers participate at the board meetings, nor do they allow workers to vote on the best course of action! There’s no sense of democracy in the capitalist workplace. There’s no direct democracy because workers don’t vote on major corporate plans, nor do they participate in the decision-making process. And there’s no representative democracy because workers don’t vote for their corporate leaders. In too many cases, there’s very little accountability either. Plenty of executives that drove their companies into the ground walked away with lucrative bonuses and financial packages while their workers lost good-paying jobs and pension funds. The multinational corporation under modern capitalism is the perfect example of a constructed order, instead of a spontaneous one.

Second, Hayek’s spontaneous order is a useless conceptual abstraction that looks at human beings like mindless bacteria swimming around in a petri dish. Humans have no concrete desires, no contingent motivations. They just unintentionally conform to prior traditions and moral standards. We’re all basically glorified robots obeying our ancient algorithms. This is a common problem with libertarianism more broadly. There’s often a refusal to look at how the world actually works, and then to make policy recommendations on that basis, while hypothetical examples and scenarios are elevated to the position of reality. Economic praxis is later inferred from those intellectual fantasies. Praxeology has almost become a religious fetish for the modern libertarian.

Most importantly, Hayek largely ignores the role of differential power and social hierarchy in human affairs, apart from making ideologically naïve and reflexive critiques of any government he considers to be socialist or communist. He has no descriptive conception of power and domination within a market economy itself. Because he’s lost in the weeds about competition and spontaneous orders, he does not appreciate how markets are fundamentally shaped and constructed by those who have the legal, political, military, and institutional power to do so.

Markets don’t just emerge spontaneously; they are organized and constructed by elite groups that have acquired extensive social power. In 1938, the United States government created Fannie Mae and the entire secondary mortgage market, forever changing the country’s real estate industry and giving more Americans than ever before the opportunity to become homeowners. In the late 1940s, the British mining company De Beers launched a massive advertising campaign meant to convince people that diamonds represented eternal love. With the legendary catchphrase, “diamonds are forever,” De Beers almost single-handedly initiated the cultural expectation that men should buy diamond rings for their engagements, even suggesting how much of their salary men should spend on the rings. Diamond sales for De Beers in the United States went from $20 million in 1940 to over $2 billion by 1980. In the 1970s, the French water company Perrier launched its own advertising blitz in the United States, aiming to convince middle class Americans to buy bottled water, which at the time was considered a silly and unnecessary luxury by most people. But Perrier’s gamble worked, more people started seeing bottled water as a sign of elegant sophistication, and sales of bottled water skyrocketed. For a completely different case, consider that global markets for military weapons are also tightly controlled and carefully constructed. The United States would never sell F-35s or other advanced weaponry to a strategic competitor like China.

The era of European colonialism and imperialism provides numerous other examples of how markets are created and manipulated by those who have power. One of the more notorious cases was the British effort to deindustrialize India in the 18th century by sabotaging the Indian textile industry, which produced a substantial portion of all the world’s clothing. Parliament imposed steep tariffs on Indian imports to boost domestic textile manufacturers, and then encouraged the East India Company to prey on weak Indian states and to shut down textile factories that could compete with British clothing being exported to India. The Opium Wars against China in the 19th century are another example (see Figure 1 below). Since the 18th century, China had refused to open itself up for British imports, so Britain encouraged the production of opium in India and then sold that opium illegally to the Chinese through various unofficial channels and black markets. When the Chinese government started cracking down on the clandestine shipments, Britain responded by launching a major naval assault that ultimately forced the Chinese to accept British goods flowing through Hong Kong and other trade ports without any tariffs.

Cartels and monopolies, from Nokyo in Japan to Standard Oil in America, have also engineered and dominated markets in various ways. One could go on and on, but the central point is clear: markets don’t simply develop because people “want stuff” and producers respond to those desires via the price system. Our individual preferences are not exclusively autonomous; they are shaped and molded by those who have social power, by those who have the capital to spend on mobilizing public opinion and cultural attitudes, by those who have the weapons and technology to subdue or eliminate weaker opponents. Markets alone do not create order; they only work because of a pre-existing order that brings them into life.

Hayek also seems oblivious to the chaotic and non-linear instabilities produced by his “competitive markets.” After Otto Heinz failed to corner the market on stocks for the United Copper Company in 1907, his brother’s bank was tainted by association and its customers withdrew their deposits en masse. The localized financial hysteria eventually spread further, and pretty soon people throughout the country started a run on the banks, leading to the Panic of 1907. This panic followed on the heels of a recession triggered by the San Francisco earthquake in 1906, which caused huge imbalances in the Western financial system as gold and cash rushed to the Bay Area for rebuilding and settling insurance claims.

Heavily deregulated markets have produced chaos and anarchy worldwide, and especially in the United States, where money laundering, predatory lending, patent trolling, pyramid schemes, tax evasion, pharmaceutical price gouging, and monopolies that act like fiefdoms, among other failures, are sapping the country’s strength and literally killing its people in large numbers. In other words, “free” markets don’t always produce an ordered society; very often they just produce death and destruction. Prominent examples include the financial panics of the 19th century, the post-Soviet collapse of the Russian economy in the 1990s, and the financial crisis in 2008 that led to the Great Recession.

Conclusion: Language and Pencils

As a parting thought, let’s revisit some of the examples mentioned above and see whether they’re spontaneous in the way that libertarians think. Take language first. Why do 30% of the words in the English language come from French? Was it because of spontaneous cultural diffusion? Did English merchants doing business in France adopt local words to speed up communication and make more money? Did English tourists pick up French words during their travels? Of course not. The simple reason why English became so French is because of the Norman invasion in 1066. The Norman nobility that took over the country spoke French, and so French became the language of law, administration, and the royal court. It was an act of violent conquest that so profoundly transformed English, not spontaneous interactions among rational individuals.

One can extend this thesis much further. Why did Spanish become so widely spoken throughout the Western Hemisphere? Was it because of trade and cultural exchanges, or was it because the Spanish ruthlessly conquered indigenous civilizations and forced them to adopt a European culture? Why is Arabic spoken over so much of the Middle East and North Africa? Could it have something to do with the Islamic Conquests in the 7th century, which were spearheaded by Arab armies and leaders? As a historical matter, it’s almost inconceivable to explain how languages change and spread without considering the history of warfare, geopolitical rivalries, and so many other factors related to social power dynamics.

Now let’s move on to the market for pencils, one of Friedman’s most famous examples. The writing material in pencil is graphite, a naturally occurring form of carbon. Graphite is composed of layered sheets containing carbon atoms in hexagonal arrangements. Although the interactions within a given sheet are very strong, the interactions between the sheets are much weaker. When the sheets are pressed against another material, the pressure makes them slide off layer by layer. That’s basically how a pencil is able leave such crisp and clear marks on paper. In the 16th century, the English discovered a massive deposit containing a very rare, high-quality form of graphite. They called it plumbago, Latin for “lead ore,” thus beginning the erroneous cultural association between pencils and lead. England held a monopoly on pencil production over the next century, allowing it to charge ridiculously high prices in European markets.

In the late seventeenth century, German manufacturers made several innovations in graphite and pencil production that collectively undermined England’s monopoly. Germany emerged as a pencil powerhouse in the eighteenth century. In 1761, the cabinet maker Kaspar Faber founded the pencil company A. W. Faber Company, which was renamed to Faber-Castell in 1900 after the marriage of a Faber heiress with a count from the House of Castell, a German noble family. Faber-Castell would go on to become the most iconic pencil company in history. It has introduced numerous innovations over the years, including the creation of the six-sided pencil, which was meant to prevent pencils from rolling off tables. German and English dominance over the pencil industry proved to be a thorn for the French during the Revolutionary and Napoleonic Wars, when France was cut off from importing foreign pencils. But because necessity is the mother of invention, the French also created new methods of pencil production to overcome dwindling foreign supplies.

From the middle of the nineteenth century until the middle of the twentieth, the majority of the world’s graphite came from mines in Sri Lanka and Madagascar, both colonial territories controlled by the British and the French, respectively. These European powers made sure to squeeze out all the profits they could from the exploited miners toiling away in dirty conditions for hardly any pay. The British even seized vast tracts of land in Sri Lanka, robbing poor peasants along the way and forcing them to work in tea plantations and graphite mines.

When World War I broke out, the British and the French stopped selling graphite to the Germans, revealing the monopolized control that the Allies had over the global graphite trade. In the latter half of the twentieth century, globalization fundamentally changed the nature of pencil production. Faber-Castell has shifted most of its production overseas, precisely because labor costs are cheaper. The company operates over a dozen factories outside of Germany. And the majority of the world’s pencils are now produced in China, where corporations exploit cheap labor costs and large-scale industrial capabilities to strike huge profits. Contrary to Friedman’s assertions, the production of pencils has always been carefully planned, organized, controlled, and dominated by imperialist powers and large corporations. It was never a spontaneous process between cooperative businesses and individuals.

If that's the case, why do libertarians see spontaneity and distributed networks everywhere? Because it's an ideological crutch. It's a way of excusing and justifying the status quo under capitalism, which is thoroughly corrupt and rigged in every significant way by dominant corporations. In a society that desperately needs revolutionary change, libertarianism can only offer reflexive excuses for the rich and powerful.

The Political Economy of Pharmaceuticals in the United States

The pharmaceutical industry in the United States is one of the most ruthless and corrupt in the entire world, aided and abetted by a feeble government that refuses to rein it in. In many countries around the world, it’s typical for governments to negotiate with drug companies so that drug prices stay low and affordable. But in the dystopian system that operates in the United States, the dominant capitalists who control the pharmaceutical industry won’t allow that to happen. What makes the situation even more infuriating is that so many essential drugs, which make billions in profits for Big Pharma, can only be developed because of critical government funding and research from the National Institutes of Health (NIH) and other bodies. We have a system where public money is subsidizing private profit, and thousands of people in America die every year as a result. Because it operates in such a lax regulatory environment, the American pharmaceutical industry has managed to score sky-high profit margins relative to other major industries.

For some concrete examples, consider the drug sofosbuvir, also known as Sovaldi under its brand name. It’s a highly effective treatment for hepatitis C, an infectious disease that harms the liver. The NIH provided $62 million for the fundamental research that led to the drug’s development. The drug company Gilead Sciences then bought sofosbuvir from its developers for a whopping $11 billion. Gilead priced the drug in the United States at $1,000 per pill, compared to just $4 a pill in India.

Perhaps the most notorious example of this casual corruption is insulin, which roughly 30 million diabetics in America need to survive. Insulin is a critical hormone that allows cells to absorb the energy contained in food. In Type I diabetes, which affects roughly 5% of diabetics, the body has a difficult time producing insulin because the immune system attacks insulin-creating cells, so patients need to inject some more into their bloodstream with an insulin shot. In Type II diabetes, the body can produce insulin just fine, but cells have a hard time interacting with it and they grow resistant to its effects. Patients with Type II also take injections of insulin.

This important hormone was discovered in 1923 by Frederick Banting, James Collip, and Charles Best, working under John Macleod at the University of Toronto. Banting himself refused to put his name on the patent because he reasoned that doctors shouldn’t profit from life-saving drugs. Collip and Best ended up selling the insulin patent to the University of Toronto for a single dollar. They wanted the drug to be cheap and widely available. But fast-forward a century and one can see how insulin has become emblematic of pharmaceutical price-gouging. From 2010 to 2020, the cost of insulin tripled in the United States. The average price per month in 2016 shot up to $450. Cartel behavior is rampant in the pharmaceutical industry. Two major insulin variants, Lantus and Levimir, experienced price hikes at roughly the same time on 13 separate occasions from 2009 to 2015. A 2019 study that administered a survey to people going to the Yale Diabetes Center found that about 25% of them reported rationing insulin, and cost was a major reason why they did it.

To understand why pharmaceutical companies operate this way, it's useful to understand their basic business model. Drug companies are interested in scoring profits and delivering "shareholder value," a euphemism for pumping up the stock price as high as it can go. Perhaps the most important profit mechanism to a drug company is the patent, because that’s what allows them to restrict access to medicine, to make their products exclusive, and to therefore charge high prices. When patents on existing drugs are close to expiring, Big Pharma companies invest a lot of money on developing minor variants of those older drugs so that they can extend patent protections and maintain market dominance. For the most part, these new variants are hardly better than their predecessors, but a medical marvel is not the goal. As long as some manipulative study can show a marginal improvement over previous versions, the companies will get both the patents and the profits they so desperately crave from the government agencies that are beholden to this corrupt system.

This is exactly what’s happened with insulin. Most of the insulin medications that American consumers buy nowadays are newer variants, allegedly better than the older versions of insulin. But studies show that the differences are often trivial or insignificant. As a result, a 2017 paper concluded that "older insulins have been successively replaced with newer, incrementally improved products covered by numerous additional patents." Over 90% of Type 2 patients with private insurance are prescribed the latest and most expensive insulin variants, even though these don’t produce much better health outcomes.

The Role of the State in Fostering Innovation

Pharma executives know that a huge chunk of their bottom line depends on robust government funding, since it's precisely the government that largely funds the risky fundamental research that the industry doesn't want to touch for fear of failure. Using patent information on drugs approved between 1988 and 2005, a 2011 paper from Bhaven Sampat and Frank Lichtenberg made the following conclusion about the role of state funding:

Overall, we find that direct government funding is more important in the development of “priority-review” drugs, sometimes described as the most innovative new drugs, than it is for “standard-review” drugs. Government funding has played an indirect role...in almost half of the drugs approved and in almost two-thirds of priority-review drugs.

The researchers take time to emphasize the importance of state-funded research in the context of HIV drugs:

The nineteen HIV/AIDS drugs we studied were exceptional in terms of all our indicators of direct or indirect government influence...Nearly a third of these drugs had a public-sector patent, and close to 95 percent cited government-funded research.

A study published in 2018 by Ekaterina Galkina Cleary and her colleagues also found that federal funding played a critical role in drug development. They examined "the contribution of NIH funding to published research associated with 210 new molecular entities (NMEs) approved by the Food and Drug Administration from 2010–2016." The research group described some of its main results as follows:

NIH funding contributed to every one of the NMEs approved from 2010–2016 and was focused primarily on the drug targets rather than on the NMEs themselves. There were 84 first-in-class products approved in this interval, associated with [over] $64 billion of NIH-funded projects.

As in so many other economic sectors, the state is usually the one that carries the burden of funding fundamental research for new drugs, since capitalists are unlikely to pour vast sums of money into something that may not yield any concrete financial benefits. But despite this reality of state-backed innovation, Big Pharma is allowed to price gouge consumers as if public interests aren't even an afterthought.

Price Gouging as a Systematic Strategy

Price gouging is a core profit-boosting strategy of the pharmaceutical industry in the United States, all facilitated by a corrupt government that refuses to do anything about it. Pharmaceutical companies like to spread propaganda about pouring money into research for innovative drugs, but the reality is that their priorities are entirely about profits and shareholders. After the 2017 tax law passed by Republicans slashed corporate tax rates, the pharmaceutical industry showered its shareholders with money. Nine drug companies alone spent a total of $50 billion on stock buybacks and dividend programs. Four companies accounted for 80% of this total; Pfizer, AbbVie, Merck, and Amgen each spent $10 billion for their shareholders.

The pharmaceutical industry commits price gouging simply because it can, because there’s no effective mechanism of state or social control to stop them. Let’s consider a few concrete cases. The company Mylan maintained a virtual monopoly on sales of EpiPens for years after first acquiring the device in 2007. EpiPens are injection devices that administer the hormone adrenaline, also known as epinephrine; they are used by people, often children, who suffer from severe allergies or asthma attacks. In 2009, two EpiPens had a $100 wholesale price for pharmacies. By July 2013, Mylan had jacked up the price to $265; in May 2015, the price reached $461, and by May 2016 Mylan raised it again to $609. The price hikes enraged parents throughout the country, since so many kids depend on the device to prevent serious health issues, and even to save their lives.

Another notorious case emerged in 2015, when the company Turing Pharmaceuticals (later renamed Vyera) purchased the marketing rights for Daraprim, the brand name for a generic drug known as pyrimethamine. The drug is used for a wide variety of purposes, such as treating parasitic infections like malaria and toxoplasmosis and helping patients with weak immune systems suffering from AIDS or cancer. After Turing took control of Daraprim, its CEO and founder, Martin Shkreli, immediately hiked the price from $13.50 per tablet to $750 per tablet, an increase of over 5,400%. The move drew widespread scorn and outrage, especially since many people had to take the drug for long periods of time. Shkreli defended the move by resorting to the usual nonsense about high pharmaceutical research costs. In reality, the pharmaceutical industry spends the vast majority of its R&D funds on useless and ineffective variants of pre-existing drugs. Shkreli later offered hospitals half off for Daraprim, which still amounted to a price hike of 2,500%. Not that the reduction mattered much anyway; patients generally take the medicine at home, after they leave the hospital, meaning that they and their insurance companies still had to foot the bill for $750 a tablet.

Shkreli also tried to defend the price hike by claiming that insurance would cover it anyway, apparently unaware that businesses in America do something called "passing on the costs to consumers." Shkreli was later arrested and imprisoned for an unrelated issue: committing securities fraud with the hedge funds he operated before he entered the pharmaceutical industry. In 2022, a federal judge ruled that Shkreli and his former pharmaceutical company must return $64.6 million in profits that they made from the obscene price hike on Daraprim, and two years later the Supreme Court refused to hear Shkreli's appeal on the matter.

Shkreli’s outrageous price hike was in no way unique or very unusual (see Figure 1 below). After buying the drugs Isuprel and Nitropress in 2015, Valeant Pharmaceuticals immediately proceeded to jack up their prices by 525% and 212%, respectively. The drugs are used to treat heart-related conditions and high blood pressure. Like Turing’s speculation with Daraprim, Valeant’s brazen decision sparked widespread outrage in the media, but Valeant’s stock price soared to record highs by the summer of 2015 as investors cheered the prospect of higher profits. In 2016, outgoing Valeant CEO Michael Pearson testified at a Senate hearing that his company had been "too aggressive" with the price hikes. Valeant was later mired in a number of legal controversies, as several of its high-ranking executives were charged with fraud and the company’s stock price plummeted. By 2017, the stock had fallen over 90% from its record highs. The company eventually changed its name to Bausch Health in 2018 as a way of trying to distance itself from these prior scandals. But despite the name change, Bausch continued the nefarious strategy of hiking prices for drugs used by vulnerable populations that had few other options.

Indeed, many pharmaceutical companies target buying opportunities for critical drugs that have no cheap generic alternatives on the market, then spike the price right away to exploit their market power, since patients have no other choice but to keep buying these drugs. These price hikes have nothing to do with supply demand. It’s not like orders for Daraprim spiked by 5,000% when Shkreli decided to send its price through the stratosphere. These corrupt corporations are committing ruthless price gouging, pure and simple. They’re exploiting monopoly power over particular drug markets to artificially jack up prices as a way of boosting profits and market capitalizations. Their selfish decisions are all about capital accumulation for their tycoons and large shareholders, not about serving the market.

The Manipulation of Research Costs

Although pharmaceutical companies claim that R&D costs represent a substantial share of their revenues, there are many good reasons to be skeptical of these claims. In the 2000s, the pharmaceutical industry claimed that roughly 17% to 19% of its revenues went into R&D. More recently, an average as high as 25% has been cited by some sources. But as scholars like Donald Light and Rebecca Warburton have pointed out, the majority of this funding is heading towards low-quality and low-efficacy replacement drugs designed to extend patent protections. Funding for basic or fundamental research in new drugs only makes up roughly 1-2% of all industry revenues. In a 2014 article, Donald Light points out that many new prescription drugs cause more harm than benefits. He writes the following:

Few people know...that systematic reviews of hospital charts found that even properly prescribed drugs...cause about 1.9 million hospitalizations a year. Another 840,000 hospitalized patients given drugs have serious adverse reactions for a total of 2.74 million. About 128,000 people die from drugs prescribed to them...A policy review done at...the Center for Ethics at Harvard University concluded that prescription drugs are tied with stroke as the 4th leading cause of the death in the United States. The European Commission estimates that adverse reactions from prescription drugs cause 200,000 deaths; so together, about 328,000 patients in the U.S. and Europe die from prescription drugs each year. The FDA does not acknowledge these facts and instead gathers a small fraction of the cases.

Light elaborates by explaining that very few new drugs represent genuine innovations:

About 170 million Americans take a prescription drug, and many benefit from them. For some, drugs keep them alive...However, independent reviews over the past 35 years have found that only 11 to 15 percent of newly approved drugs have significant clinical advantages over existing, better-known drugs. While these contribute to the large medicine chest of effective drugs developed over the decades, the 85 to 89 percent with little or no clinical advantage flood the market. Of the additional $70 billion spent on drugs since 2000 in the U.S. (and another $70 billion abroad), about four-fifths has been spent on purchasing these minor new variations rather than on the really innovative drugs. In a recent decade, independent reviewers concluded that only 8 percent of 946 new products were clinically superior, down from 11 to 15 percent in previous decades...Only 2 were breakthroughs and another 13 represented a real therapeutic advance.

Part of the problem is that much of the biomedical research which supports the adoption of these new drugs is full of flaws and loopholes that are ignored and marginalized. It's well-known, for example, that p-hacking is a major problem in the social sciences, especially for biomedical research conducted by smaller companies. P-hacking is the practice of manipulating and distorting datasets or experimental procedures so that they yield a p-value of under 5%, which is the widely accepted alpha level in the social sciences, otherwise known as the point of "statistical significance." Getting results under this alpha level is critical to having research published in peer-reviewed journals, so many research teams will search for ways to make that happen if the initial results look disappointing. For example, a company may initially design a study to last for a year, but then may call it quits after 7 months because they've obtained the statistically significant results they wanted. For a similar example, a drug might be initially intended to work for men and women, but then the company digs through the data and finds statistically significant results for men over 60, so the whole marketing strategy shifts and all of a sudden it's a drug meant for older men!

Light emphasizes the importance of what he calls the "trial-journal pipeline," in which research teams work with marketing departments to feed the FDA with biased studies. He points out that "commercially funded clinical trials are at least 2.5 times more likely to favor the sponsor's drug than non-commercially funded trials." In a 2013 paper, Marc Rodwin made the same basic observation, explaining that "scholarly studies have revealed that drug firms design trials that skew the result and that they distort the evidence by selective reporting or biased interpretation." The central point is crystal clear: the pharmaceutical industry has flooded the world with numerous drugs that are largely ineffective or harmful, yet preposterously claims that it deserves sky-high profit margins because it's producing so much innovation!

To better understand the claims of the pharmaceutical industry on research costs, it’s worth looking at the work of Joseph DiMasi, a professor at Tufts University who has become widely known for publishing severely flawed statistics on average R&D costs for new drugs. First of all, we should note that the Tufts Center for the Study of Drug Development where DiMasi worked is heavily funded by the pharmaceutical industry. In 2001, for example, roughly 65% of the funding received by the Center came from pharmaceutical companies. That rate has varied over the years, but it’s always been true that drug companies have strongly supported the work of the Center, thus clearly posing a conflict of interest.

In a 2003 paper, DiMasi claimed that the average R&D cost for a new drug was roughly $800 million, a huge value. Thirteen years later, he updated that already ridiculous figure to an even more ridiculous $2.5 billion, an astonishing sum. DiMasi obtained these numbers from surveys that he sent out to various pharmaceutical companies. In both 2003 and 2016, ten of these companies responded to him, although he doesn’t disclose which ones. In fact, there are many important details about his data and his sources that are never publicized, making it impossible to actually replicate his work. But what we do know is enough to fully invalidate his conclusions.

The central problem is that his samples are extremely biased because they only include high-priority and innovative drugs that receive lots of research funding. It would be like trying to determine the average price of new cars and only looking at Teslas; you’re obviously going to get a sharply inflated number. In reality, the vast majority of new drugs that enter the market are actually generics, biologics, and variants of pre-existing drugs. The development costs for these types of drugs are generally low, or least much lower than for priority review drugs.

Another problem with DiMasi’s numbers is that he adds an estimated opportunity cost to the initial average cost he determines from his data. In the 2003 study, for example, his initial estimated average was $400 million, which he then inflated to $800 million after the opportunity cost calculation. The basic idea behind this adjustment is that a pharmaceutical company could’ve used the R&D money for the drug in other ways, such as marketing for existing drugs or expanding production capacity. Since the company would’ve made money through these other investments, DiMasi then infers that it therefore incurred an additional “cost” when it developed its drug. The main problem with this argument is that, for many pharmaceutical companies, blockbuster drugs comprise a substantial share of their total revenues. It’s not uncommon for a single drug to make up half of all annual sales. If these companies don’t produce new blockbusters, they’ll basically go bankrupt because rival companies will outcompete them in the market. The real opportunity cost of R&D funding is the firm’s survival, and it’s impossible to quantify that in any meaningful sense, so no one should bother trying.

In addition, the notion of opportunity cost is highly ambiguous. No one can possibly know the outcome of a hypothetical scenario, because by definition it’s hypothetical. Sure, maybe a drug company could’ve taken that money it spent on R&D and used it effectively to boost profits, but it could’ve also done something stupid with that money that would’ve caused the firm’s total collapse. There’s no way to know. Opportunity cost can never be measured, even in principle; it can only be guessed. In addition to these arguments, others have pointed out that government funding is absolutely critical to the development of new revolutionary drugs, but DiMasi’s studies say absolutely nothing about that. In short, nobody really knows the average R&D cost for new drugs, because determining that information would require the pharmaceutical industry to open up its books in a radical way. The simple truth is that the pharmaceutical industry loves to exaggerate research and development costs because it wants to justify sky-high drug prices across the board, even for what should be cheap generics.

The Road Ahead

We’re stuck in a situation where a corrupt industry is making huge profits and, apart from the production of vaccines and generics, delivering pathetic results in many cases. The federal government is afraid of upsetting Big Pharma’s crony capitalists by bringing down the hammer on their predatory industry. Instead of regulating the pharmaceutical industry, the government helps it make lots of money, regardless of all those who die along the way.

We need a new economic and political model that views drugs as public goods. The government needs to have greater pricing power so that a few corrupt capitalists aren't setting prices based on whichever direction the wind is flowing that day. The patent system needs to be thoroughly reformed, as I have argued elsewhere, so that ineffective variants of pre-existing drugs don't receive patent protections and market exclusivity. The government also needs to impose greater requirements and regulations on how the pharmaceutical industry reinvests its profits, since so much of that money now is being squandered on stock buybacks and dividend payments for shareholders. And there are a million other things that need to change as well. The pharmaceutical industry is too important for public health to be left in a putrid state of anarcho-capitalism. We need more public oversight and control. Our lives are at stake, and promoting major change in this corrupt industry should be a national priority.

Patents are not innocent symbols of technological innovation. In the neoliberal era, they have become critical elements of capital accumulation, allowing corporations to dominate markets by erecting barriers and excluding potential competitors. In the United States, patents and trademarks are given by the United States Patent and Trademark Office (USPTO), which is an agency of the Commerce Department. USPTO is a unique federal organization in the sense that it’s financed by fees collected on patent applications, not by general revenue funds from the government. This may sound like a good thing, but it’s actually a huge problem because it incentivizes the approval of ridiculous patents that have no technological value of any sort. Let's dive right in.

Patents and the Legal System

USPTO generally approves over 50% of all patent applications for any given year. In 2020, to pick a recent example, it received roughly 597,000 applications on “utility patents,” or claimed inventions, and approved 352,000 of them, for an approval rate of 59%. Patent approval rates have generally been high for long periods of time (see Figure 1). If it didn’t approve patents at such high rates, fewer people would apply for them and revenues for USPTO would plummet.

The vagueness of the law certainly doesn’t help the situation. We are told that anyone who "invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent." For a patent to be legally valid, it must meet at least two standards: it has to be novel or unique, and it can’t be an obvious change to existing technology. In other words, the patent must be for some new product that represents a genuine technological innovation. But it’s not easy to know how to apply these criteria at a practical level, so companies frequently engage in expensive legal fights to find out which patents are valid and which ones are not.

Patent lawsuits usually start out in federal district courts, but the major action happens at the next level. The United States Court of Appeals for the Federal Circuit is the court responsible for all appeals related to patent litigation. It has a notorious pro-patent bias. Its reign of terror over the patent system began in 1982, when Congress changed the law to give the Federal Circuit a virtual monopoly over all patent cases, as opposed to allowing the general court system to handle patent disputes, which is how the process worked before. The idea behind this move was to cultivate a legal nucleus of expertise regarding patent issues. Things didn’t exactly work out that way.

What happened instead is that the Federal Circuit became a reliable stooge for patent lawyers everywhere. From 2004 to 2014, the Supreme Court overruled the Federal Circuit 10 separate times. Nine out of these ten Federal Circuit decisions that were eventually struck down favored patent holders. The most famous patent case to land before the Supreme Court in this period was undoubtedly Alice Corp. v. CLS Bank International, decided in 2014.

Alice was a company that held four software patents related to financial exchanges for escrow accounts. The software was designed to reduce settlement risk in financial transactions. CLS Bank, a financial firm, had developed its own independent software for managing financial transactions, but Alice argued that it had gotten there first. Alice kept threatening CLS Bank for apparently infringing these patents, which prompted CLS Bank to sue Alice in order to stop the harassment. A district court initially ruled that the patents were invalid, but the Federal Circuit reversed that ruling in a 2-1 decision. However, a wider judicial bench from the Federal Circuit failed to find consensus and the case went to the Supreme Court.

The main issue in the case was whether someone could patent an abstract idea by implementing it through generic computer methods, like reducing settlement risk in financial transactions by using a simple algorithm or making people happier throughout the day by sending them text messages. In a unanimous decision, the Supreme Court rejected Alice’s patent claims and concurred with CLS Bank. The Supreme Court held that people cannot claim patents over ideas implemented on typical computers. The Alice case instantly became a landmark decision, making it much tougher to enforce claims on software patents.

In 2013, the economist Shawn Miller published a paper analyzing the level of innovation for existing patents. Miller wanted to find out how many low-quality patents could be lurking in the system, and his findings suggested that many patents did not meet legal requirements on innovation. He looked at 980 patents and discovered that a shocking 37% of them "were at least partly invalidated." About 30% were completely invalidated. Based on these results, he estimated that roughly 28% of all patents would be declared "partially invalid" if they went to court. He explained his findings as follows:

While the main theoretical benefit of patent protection is increased innovation, some assert a swamped US Patent and Trademark Office has granted an inefficiently large number of patents with negligible innovation value. This [article] tests this argument’s plausibility and determines the characteristics of patents without innovation by analyzing 980 litigated patents subject to anticipation or obviousness decisions between 2000 and 2010...This [article] estimates a surprising 28 percent of all patents would be found at least partially invalid if litigated. Software, business method, and licensing firm-owned patents possess significantly higher innovation-based invalidity rates than other types of patents.

Low-Quality Products and Patent Trolling

This paper highlighted one of the fundamental problems with the patent system: its overwhelming tendency to approve patents for frivolous products that have absolutely zero technological value or future potential. In 1999, inventor Alan Hirsch was actually given a patent for a method that induced or enhanced "penile erection through the delivery of odorants for inhalation" (see US Patent 5,885,614). The odorant contained, among other things, "a mixture of lavender and pumpkin pie" and a "mixture of doughnut and black licorice." That’s right: USPTO gave a patent to someone who came up with a BS "cure" for impotence. In 2004, someone received a patent for a "support apparatus" meant to help someone rest his head "on a bathroom commode or urinal" (see US Patent 6,681,419). In 2003, three inventors got a patent for a "beerbrella," a removable mini-umbrella that attaches to a beer to keep it cool under the summer heat (see US Patent 6,637,447). Other examples include a machine that spins a woman around until she gives birth, patented in 1965, and a "bread refreshing method" for toast, patented in 2000. In 2012, Apple actually received a patent on a rounded rectangular design used for the iPhone. A few years later, USPTO thought it was a good idea to give Amazon a patent for a flying warehouse.

The proliferation of absurd or low-quality patents has allowed corrupt companies, known as patent trolls, to launch frivolous lawsuits against anyone who they suspect might be violating their patents, although big companies aren’t innocent in this drama either. Patent trolls are companies that own lots of patents, but they don’t actually produce anything themselves. Their business model is entirely based on suing other companies, who are often forced to settle for huge sums of money to avoid even more expensive legal fees.

Consider the company Shipping and Transit LLC, previously known as Arrivalstar. It held 34 patents related to GPS vehicle tracking, none of which represented any kind of technological innovation. The company’s patents originated with Martin Kelly Jones, a failed businessman who realized he could make more money as a predatory leech on the patent system. Over the course of its decrepit existence, Shipping and Transit filed over 500 lawsuits alleging patent infringement and obtained hundreds of payments from victims that didn’t want to get bogged down in expensive legal fights. This troll has sued transportation companies, municipal transport agencies, and even businesses that simply sent an email to customers providing them with a tracking number for their delivery.

Despite filing so many lawsuits, a court had never ruled on the merits of its patent claims. When some of the companies it sued decided to go to court instead of settling, Shipping and Transit instantly pulled its lawsuits to avoid a judgment on merit. Judges began to notice this strategy of legal dodging and ordered the troll to pay attorney’s fees, which quickly sent the company into terminal shock. Shipping and Transit filed for bankruptcy in 2018, valuing its useless patent portfolio at $1, which was still too high.

Another notorious example of a patent troll is Acacia Research Corporation. The company buys up patents from willing inventors and then ruthlessly sues anyone that might even be remotely violating them. It’s especially active in the software space and has sued Apple multiple times, in addition to other tech companies. As of 2010, Acacia had filed 337 patent lawsuits in 18 years, which collectively generated over $350 million in revenues.

According to a 2012 study by James Bessen and Michael Meurer, lawsuits from patent trolls, or “Non-Practicing Entities,” imposed $29 billion of direct legal costs on targeted companies in 2011 alone. The indirect costs to the economy are vastly greater, largely because these legal squabbles affect the research and development pathways that companies choose to pursue. Indeed, many small startups and businesses have suffered enormously because of legal and financial pressure from patent trolls.

Take Capstone Photography, a small business that was targeted by a patent troll in 2013. Michael Skelps, the general manager of the company, described what happened next:

Capstone spent 10 months on litigation and $100,000 in legal fees, equivalent to a year’s worth of profit for my small business, to defend ourselves. The financial toll was devastating, and included laying off 60 percent of my workforce. All our resources were diverted to a battle we should have never had to fight in the first place, instead of to job creation and further innovation in my field.

Echoing this experience, Bessen and Meurer summed up their results like this: "we find little evidence that [patent trolls] promote invention overall." In 2011, about 2,100 companies faced nearly 6,000 legal cases regarding patents. The number of patent infringement cases that year was six times higher than in the 1980s. The numbers since 2011 have only gotten worse, with trolls filing more cases and targeting more companies.

The problems are especially severe for software patents, as James Bessen explained in a 2014 article. Software patents and business method patents, which are usually just software patents by another name, are very likely to be involved in litigation. Citing data from US Courts, a government source, Bessen claimed that around 75% of patent legal defendants have been sued over software patents. And once they’re in court, things often don’t go well for software patent holders. As Bessen points out, "software and business method patents are much more likely to make invalid claims. One reason is that when patent applications have fuzzy boundaries, it is difficult for patent examiners to tell whether they are different from previously existing technology." Bessen cites the sample from Miller’s study, which found that 38% of all software patents and a stunning 53% of all business method patents had at least one invalid claim, both higher than the overall rate of invalid patents.

Patent trolls like to claim that they’re defending helpless inventors against big corporations. But their legal strategies suggest otherwise. In a 2013 paper, the legal scholar Colleen Chien explained that patent trolls like to go after smaller businesses instead of bigger ones. She put it this way:

Although large companies tend to dominate patent headlines, most unique defendants to troll suits are small. Companies with less than $100M annual revenue represent at least 66% of unique defendants and the majority of them make much less than that: at least 55% of unique defendants in [patent troll] suits make under $10M per year. Suing small companies appears to distinguish [patent trolls] from operating companies, who sued companies with less than $10M of annual revenue only 16% of the time, based on unique defendants.

The rationale for this strategy of targeting the small fish is that startups don’t have big legal departments to defend themselves. It’s much easier to get them to settle out of court and move on to the next victim than it is to bring Google or Microsoft to the negotiating table. And given the huge costs of patent litigation, it’s easy to understand why small companies want to settle.

The American Intellectual Property Lawyer’s Association (AIPLA) estimated that litigation costs can run anywhere from $250,000 in cases with patents worth less than $1 million to some $2.4 million in cases with patents worth over $25 million. AIPLA claims that higher-value patents can impose over $4 million in legal costs. These are huge sums of money that many small businesses just can’t afford, thus giving patent trolls a big advantage in their campaign of legal harassment. Furthermore, many startups don’t bother patenting their methods and inventions right away because a patent application can be expensive, especially if lawyers are involved. Some applications with lawyers can run over $10,000, and sometimes much more.

Patents and Power in the Struggle for Accumulation

Patent trolls are not the only ones engaging in strategic harassment and frivolous lawsuits. Big companies do it too. In the 1980s, Gary Reback worked as an attorney for Sun Microsystems, a new computer company that would go on to create the Java programming language, among other achievements. A team of IBM lawyers showed up out of the blue one day and accused SM of violating seven patents held by IBM. Reback’s colleagues patiently demolished the accusations one by one at the whiteboard. After an awkward silence, the lead IBM lawyer said that “maybe you don’t infringe these seven patents. But we have 10,000 U.S. patents. Do you really want us to go back to Armonk [IBM headquarters in New York] and find seven patents you do infringe? Or do you want to make this easy and just pay us $20 million?” After some negotiations, SM agreed to pay IBM just to stop the harassment.

Amazon is another example. In 1999, Amazon sued Barnes & Noble for violating its 1-Click patent. Amazon had received the patent for developing a simple method that allowed customers to purchase products online with just one click, after saving and storing their payment information on an Amazon server. Although the companies settled their dispute in 2002, Amazon should have never gotten such an absurd patent in the first place. Saving credit card information and letting people buy something with one click is a painfully obvious extension of pre-existing software technology. And other jurisdictions have also shared this assessment. In 2011, the European Patent Office rejected Amazon’s request for a patent on its 1-Click method precisely because it did not represent a genuine technological innovation. Amazon’s American patent expired in 2017, but the damage was done by then.

Perhaps the most notorious recent struggle has been the international legal war waged between Apple and Samsung over their smartphones. Apple fired the first salvo by accusing Samsung of copying several style and design features of the iPhone. Samsung counter-sued in various jurisdictions, alleging that Apple was actually the one infringing on Samsung patents. The two companies eventually had pending legal cases everywhere from South Korea to the United States. In 2018, they reached an undisclosed settlement that brought their legal squabbles to an end. Many of the patents in question were about simple designs and stylistic features; these patents should have been denied from the outset.

Patents are a critical tool for blocking competition and reinforcing predatory pricing strategies, as the pharmaceutical industry knows all too well. The pharma giant AbbVie once had a stranglehold on the rheumatoid arthritis market through Humira, a blockbuster drug that delivered over $100 billion in revenues to AbbVie from 2016 to 2022. From 2003 to 2023, AbbVie increased the price of Humira on roughly 30 separate occasions. By 2023, the drug's price had gone to over $80,000, a stunning 60% rise since 2016. The dominant reason why AbbVie made so much money from Humira is because of patent-centered industrial sabotage, as it constantly sought and received frivolous patents designed to extend the drug's market exclusivity, all while threatening lower-cost competitors with massive, crippling lawsuits. In total, AbbVie had filed 311 patents related to Humira as of 2023, with 165 of them being approved. In 2014, to pick one example, AbbVie seriously tried to patent a specific dosage containing 40 milligrams of Humira, arguing that this exact dosage could treat a unique form of arthritis. When the application was approved, it added 11 years of additional patent protection after 2016. But the good times didn't last, as Amgen and other major companies launched several lawsuits against AbbVie, accusing it of holding invalid patents. The two giants reached a settlement that allowed Amgen to start selling a competitor to Humira in 2023, after which the price of Humira finally began to drop, as did AbbVie's sales (see Figure 2).

A landmark 2021 study showed that pharmaceutical companies regularly acquire other companies and drugs exclusively for the purpose of eliminating cheaper alternatives that can compete with their existing products. The authors of the study called such purchases "killer acquisitions." These are, in effect, business decisions intended to sabotage innovation and concentrate capital accumulation in a few monopolies. The authors write:

Using pharmaceutical industry data, we show that acquired drug projects are less likely to be developed when they overlap with the acquirer’s existing product portfolio, especially when the acquirer’s market power is large because of weak competition or distant patent expiration. Conservative estimates indicate that 5.3% to 7.4% of acquisitions in our sample are killer acquisitions. These acquisitions disproportionately occur just below thresholds for antitrust scrutiny.

The examples go on like this for industry after industry. The fundamental problem is that because patents have turned into a tool for financial speculation and capital accumulation, we as a society have forgotten about their central purpose: to produce innovation. Patents have become a rent-seeking virus, an industrial-scale cudgel used for bludgeoning competitors and reinforcing market domination.

The Way Forward

Reforming this corrupt system will require many major changes. First, it's obvious that USPTO grants too many ridiculous patents that offer no technological progress and few economic benefits in large part because of its funding stream. An easy reform is therefore to fund USPTO from the government’s general revenue funds instead of application fees. On the legal side, patent law should go back to being administered by a broader distribution of courts. And patent trolls should receive harsher punishments for launching frivolous cases. The government could even set up an agency or a regulatory board that has the power to manage and resolve certain patent disputes, especially those involving small businesses. These are all easy and simple bureaucratic changes that require no radical transformations. But we can't stop there.

We need a fundamental reappraisal of what "technological innovation" means so that we only give patents for products that represent genuine innovations. It's pathetic that anyone has to say this, but it shouldn't be possible in the United States of America to patent a rectangle or other basic shapes. Design patents should be restricted for things like complex machinery and devices, not simple items that a child can imagine while dreaming. We also need to shift the focus and purpose of patent law, from the current model of boosting capital accumulation for major corporations to a model that prioritizes widespread access with the goal of making the benefits of innovation broadly available to the public at low costs. Following this path is especially important for research and innovation funded by government money, which is so critical for the pharmaceutical industry and many other sectors. Patents can reward private innovation, but they also need to reward the public for setting up the financial and social infrastructure that allows innovations to happen in the first place. The goal should be to replace the current system, in which patents serve as vehicles of financial speculation, with another more dynamic system where patents are increasingly seen as public goods meant to provide public benefits.

The reality is very different, and far more complex. Getting it right on housing costs and housing supply implicates the totality of the social order, and specifically a society's regime of political economy. The United States provided enormous housing supply in the 2000s during the Bush years, but then it all unraveled with the foreclosure crisis in the Great Recession, pulverizing the dream of affordable home ownership for millions of Americans. China is the global master of building mass housing, putting up more housing capacity than any other nation and erecting entire ghost cities where few people live. And yet that didn't prevent the emergence of a major real estate bubble that has upended much of the Chinese economy. We have the resources and the knowledge to establish a stable and affordable housing regime, but doing so will require carefully considering and balancing the different political and economic forces that have a strong causal impact on housing costs.

America's affordability crisis, in housing and other areas, is not just an issue of supply constraints, but a broader challenge that involves the fundamental structure of our political economy and the way that structure then shapes the management of supply and demand.

The Centrality of Political Economy in Housing Supply Dynamics

China and Eastern Europe have the highest rates of home ownership in the world, in some countries exceeding 90 percent. How come? History clearly shows that the most reliable path towards affordable housing is the expansion of public housing. Socialist governments constructed massive public developments that all but eliminated homelessness. After the collapse of the communist governments in Eastern Europe, most people obtained these homes from the government for pennies on the dollar, often paying just a small registration fee and acquiring full legal ownership as a result. And the market for these communist-era apartment homes remains quite strong. The Czech Republic is a great example, as the panelák building complexes from communist times still housed, as of 2005, roughly one-third of all Czechs, an astonishing 3.5 million people (see Figure 1). It’s a glaring reminder that private capital often struggles to produce affordable housing because doing so would be unprofitable. Only a strong public housing program coupled to effective regulations can deliver on the promise of universal housing at affordable rates.

Let's take a closer look at what happened in China and Eastern Europe. World War II destroyed or damaged many Eastern European cities and communities, especially since the Eastern Front was the biggest and deadliest sector of the entire war. As communist governments were installed after the Soviet victory, they launched a large-scale building campaign to provide housing for their devastated countries, using pre-fabricated housing components to construct things cheaply and quickly. The modularization of this housing boom created massive efficiency and productivity gains, allowing governments to build millions of units in a rapid amount of time. That historic push led to a wave of new apartment towers and city blocks, providing mass public housing to millions of people. The apartment blocs have often been stereotyped as ugly and unappealing in the West, reflecting the bourgeois sensibilities of those making the criticism. But for hundreds of millions of people, they were and continue to be a vital lifeline for cheap and stable housing, automatically eliminating one of the great stresses of life.

The land and homes were usually state property, with people making a modest monthly rent payment that was heavily subsidized by the state. A similar system still operates in China, where the government owns the land and individuals lease it for a set period of time, usually 70 years for residential purposes. Chinese people have private property in the sense that they own the houses or apartments built on the land, but they do not own the land itself.

There are a few interesting facts worth highlighting from the historical experience of China and Eastern Europe. First, these housing booms were directed and carried out by the state. Socialists don't hire private consultants to waste time and make generic recommendations. They mobilize and deploy resources to accomplish big things. Second, the state provided all the money necessary to finish the projects on demand. No construction company had to spend time wondering about its interest costs. In China, state-owned banks provide loans to private developers with artificially ultra-low interest rates, precisely to keep costs down and to encourage more development. Third, restricting customization through modular manufacturing and modular design strategies led to phenomenal economies of scale, allowing entire mini-cities to go up in months or a couple of years, a breakneck pace unheard of in much of the Western world, where the obsession with luxury aesthetics and customized designs often overrides practical issues around functionality and affordability.

Figure 1: Panelak housing in Prague, Czech Republic. Millions of Czech people live in these apartment units constructed during the communist era, and the market for them is still robust. Panelak developments were constructed from modularized panels, which facilitated rapid rates of construction.

Housing and the Complex Dimensions of American Political Economy

Things work very differently in the United States, where private capital dominates housing politics. This domination leads to ideologically rigid and inflexible proposals that are completely inadequate. For example, supply-side liberalism is now in intellectual vogue with Abundance, the new book from Ezra Klein and Derek Thompson. Before I turn to criticism, I want to say some positive things. Their vision does represent a somewhat meaningful improvement over the current technocratic managerialism that pervades the Democratic Party, so they should be commended for that. They offer many useful insights for unshackling certain parts of the bureaucratic state. On the other hand, their proposed solutions on housing are too narrow and would do virtually nothing to address America's fundamental issues around affordability.

In their book, Klein and Thompson complain that too many local, state, and federal regulations are hampering our ability to build things quickly, including housing. For just one example, they go hard after the California Environmental Quality Act, legislation passed in 1970 that mandates environmental reviews for many construction and development projects in California. But beyond bromides about how regulation is restricting new construction, they have no comprehensive theory about the role of political economy in providing industrial-scale supply, for housing, energy, or anything else.

Let's examine how these different dimensions of political economy affect housing supply and demand in the United States. A good place to begin is with money and financing, the most critical barriers to home ownership or affordable housing for most people. The housing bubble in the Great Recession, which produced millions of foreclosures, is a landmark example of what happens when housing supply and demand are driven by ruthless financial speculation. Of course, the housing bubble couldn't have been possible without a wave of new home construction, fueled by an environment of low interest rates and lax regulations (see Figure 2). But the situation was aggravated and turbocharged by corrupt and predatory lending to borrowers from various subprime lenders and Wall Street firms looking to make a quick buck by skirting every rule in the book. The wave of financial speculation in housing nearly brought down the entire economy, and only direct government bailouts and interventions prevented an otherwise inevitable collapse.

Figure 2: There was a boom in new single-family housing (blue line) leading up to the Great Recession, even though nothing really changed about local zoning regulations. The chart underscores the futility of hyper-obsessing over zoning laws, as developers will find plenty of reasons to build under the right financial conditions. Source: Lance Lambert at ResiClub Analytics.

For a concrete example, consider Ameriquest, the largest subprime lender in the country before the Great Recession. Ameriquest became notorious for fabricating loan applications and doctoring information in order to make potential borrowers qualify for subprime loans. In 2006, it accepted a $325 million settlement with states nationwide for its predatory behavior. At least 240,000 of Ameriquest’s roughly 750,000 customers from 1999 to 2005 were entitled to compensation from the company under the deal. Ameriquest was a viper’s den of criminality. Former employees later reported that people in various company branches would joke about “taking the loan to the art department,” meaning they would take it to the copy room so that W2 statements and bank statements could be altered and falsified. The copy rooms were decked out with scissors, tapes, and special markers for blanking out written text, among other things someone might need to commit mortgage fraud. Other methods of fraud included having borrowers sign a document that guaranteed them a fixed-rate loan, then literally ignoring or tossing that document away after the closing. Ameriquest was not alone in its brazen illegal behavior. In a very real sense, the housing bubble was a story of fraud and crime more than anything else.

Finance pervades so many other issues in the real estate industry. For example, private builders won't build much if interest rates are high, which is exactly what happened in 2022 and 2023 as interest rates soared. Real estate developers usually pay for large projects by taking out loans from private banks, meaning that higher interest rates will raise production costs, especially for large developments. Unsurprisingly, developers often counter this threat to the bottom line by jacking up rental prices and housing costs.

Developers also won't build much if labor and material costs are too high. They won't build much if they're facing spiking land prices from hyperlocalized upzoning. If you jack up their production costs, they will hesitate to build in large quantities. Capitalists don't want to lose money. In many cases where upzoning is spatially restricted and ultralocalized, developers just bring in more expensive luxury units to cover for the extra costs from land speculation, thus sabotaging the fundamental goal of affordability, as I explain in greater detail in the next section. In addition, there's growing concentration in the home building industry, with a few dominant builders increasingly controlling the supply in many major metro areas. Under these circumstances, there are fewer and fewer incentives to compete, and it's well-known that, in every single quarter, many large builders actively avoid selling and building thousands of homes that they're easily capable of churning out otherwise.

We haven't even scratched the surface of the problems. Another major issue is the delegation of power and authority over zoning and the broader contours of real estate development to local governments. Why do local authorities in America have so much power over zoning and housing in the first place? Americans even pay their real estate taxes to local and county governments. To cut a long story short, it's because of federalism and the history of American political development, which collectively have enshrined a strong private property rights regime. These kinds of local barriers to massive social change have existed for many societies throughout history, even if they weren't federal republics. It wasn't until the French Revolution that the French finally overcame eminent domain issues for the construction of massive canals and irrigation projects in Provence, which until then had been successfully resisted by local towns and landowners. It's easy to say that we should have large-scale upzoning, but we're not going to get it until there's more centralized control over the economy from the state.

A big problem with the position of many supply-siders, who think of upzoning as some kind of eternal panacea, is that land prices tend to rise rapidly after the expansion of high-density housing. That’s because capitalists and investors are expecting big returns from the forthcoming high-density construction, so they’ll do just about anything to grab the necessary land and whatever else is on it. That mad rush to seize valuable land then contributes to higher rental prices and housing costs, as capitalists try to pass on higher acquisition costs to consumers.

Here's another big issue. How can regular people compete in many real estate markets where something like 25% of the sales are going to investors, from small groups to big Wall Street firms? If you put a substantial chunk of a country's housing supply up for speculation, you shouldn't be shocked to know that people will have a hard time affording homes. Real estate investors, from big Wall Street firms to smaller investment groups, play a major role in the rental and housing markets, squeezing out competitors and driving up prices and rents to boost profits. A 2021 report by the Pew Research Center described the ownership composition of the rental market as follows:

The Census Bureau counted nearly 20 million rental properties, with 48.2 million individual units, in its 2018 Rental Housing Finance Survey, the most recent one conducted. Individual investors owned nearly 14.3 million of those properties (71.6%), comprising almost 19.9 million units (41.2%). For-profit businesses of various sorts owned 3.7 million properties, or 18.8%, but their holdings totaled 21.7 million units, or 45% of the total. Entities such as housing cooperative organizations and nonprofits owned smaller shares of the total. Businesses own larger shares of units because individuals, while far more numerous, tend to own one or two properties at most, while businesses’ holdings are larger. In fact, 72.5% of single-unit rental properties are owned by individuals, while 69.5% of properties with 25 or more units are owned by for-profit businesses. Most rental properties are owned by individuals, but only a small share of individuals own rental property, according to IRS income-tax data. In 2018, 6.7% of individual tax filers (about 10.3 million) reported owning rental properties. Those filers reported owning 1.72 properties on average.

When so much housing supply is controlled by profit-hungry corporations, it's not surprising to see strong upward pressure on rental costs. For example, rent prices tend to soar when private equity firms take over existing apartment units and buildings.

Many economists and analysts discount the importance of large institutional investors in the housing market because those investors own a small share of all single-family homes. They’ll point out that Wall Street real estate firms purchase “only” about 2% of single-family homes in the United States. This statistic is routinely regurgitated as a way of excusing Wall Street’s predatory behavior. But obsessing over this number is a complete distraction from the more fundamental and structural problem: that most of the damage in the housing sector is being done by upper class investors, not Wall Street tycoons. According to a comprehensive Harvard report, investors made up a stunning 28% of all single-family home sales in the first quarter of 2022, up from 19% in the first quarter of 2021 and way above the 16% share they had from 2017 to 2019. The authors of the report stated that investors had a huge impact in certain markets:

Investors focused on markets with rapid home price appreciation, accounting for especially high shares of sales in Atlanta (41 percent), San Jose (38 percent), Phoenix (36 percent), and Los Angeles (34 percent) at the end of 2021.

These are staggering numbers, and they’re a fundamental reason why so many middle-class families in America cannot afford to purchase a home. Although it’s inaccurate to blame large Wall Street firms exclusively for America’s housing affordability crisis, it’s absolutely fair to point out that real estate investors in general are indeed pricing out regular families in numerous real estate markets across the United States. And although some of these investors operate small companies with limited financial resources, most real estate investors find themselves easily in the upper tiers of the American economic ladder, if not much higher.

Nevertheless, we shouldn’t be too lenient on the big Wall Street players. That’s because these apex investors really do their damage among affordable, inexpensive single-family homes that require lots of repairs. They don’t typically bid on more expensive properties, leaving those to the professional classes, like doctors, scientists, and programmers. Because big-time investors scoop up these otherwise affordable homes with all-cash offers and other sweeteners, they’re pricing out individuals in the middle class who may be trying to enter the housing market for the first time. In 2021, the real estate company Invitation Homes, which heavily invests in single-family housing, spent roughly $2 billion purchasing 4,802 homes. Large corporations like Invitation can often buy properties before they’re listed for public viewing, can use algorithms to determine which homes they should target, and can present all-cash offers that are hard to turn down. It’s absolutely not an even playing field; most American families cannot compete under these conditions.

Yet another critical issue for housing affordability is the distribution of wealth and income. Over the past few decades, we've had a corrupt neoliberal economic regime that has funneled wealth and income to the ultra-rich while leaving everyone else in the dirt. A typical teacher in San Francisco makes around $80,000 a year, so how can she be expected to afford a $1 million home? But imagine a world where a typical teacher in America made $250,000 a year. Suddenly price tags of $1 million or more look reasonable for a middle class couple, where middle class here is being redefined so that the couple in total makes around $400,000 or $500,000 a year. Wealth and income redistribution therefore must be essential components in resolving the affordability crisis, but they rarely come up in the discourse of the corporate media on the housing issue, which is dominated by upzoning propaganda to the near-exclusion of any other considerations.

And there are so many other dimensions to this problem, not all of which I'm going to cover in this post. Consider rent control, one of the tools used by various municipal governments around the world to try and control housing costs. Rent control can come in many different variations, some being more ambitious than others. It’s definitely not the best instrument for achieving the goal of housing affordability, and almost always doesn't work just by itself. But it’s nowhere near as bad as often portrayed by conservative economists, and it can be more effective when implemented in conjunction with other comprehensive strategies.

For example, Vienna is the most prominent major city in the world where strong public housing programs coupled with effective rent control regulations have produced relatively affordable rental prices for most middle-class families. In 2015, Vienna had roughly 890,000 apartments used as a primary residence, with 160,000 being occupied by owners and about 693,000, or 78% of the total, occupied by tenants (data from Statistics Austria, Microcensus 2015). Among rental apartments, about 401,000 are social housing units and 292,000 are privately owned. That means 58% of the city’s entire rental stock is social housing. Crucially, only 66,000 housing units were exempt from rent control laws, about 7.4% of the total housing stock. This figure implies that the vast majority of Vienna’s housing, over 90%, is either covered by rent control regulations or falls under social housing.

The notion that higher supply by itself will fix America's housing crisis is sheer fantasy. Even if we could suspend the laws of physics and make an infinity of new homes, it would guarantee nothing about affordability if the market is poisoned and destabilized by ruthless financial speculation. Contrary to the proposed policies of many housing advocates, bringing more supply to the market is not a panacea for sky-high rental and housing costs. China is the posterchild for why that argument doesn’t work: China suffers from huge levels of housing oversupply and yet still experienced outrageous increases in housing costs from 2000 to 2020, in violation of what the simplistic story about supply and demand would suggest. A 2016 paper from a team of economists explored the housing dynamics of the Chinese real estate market in the early 21st century. The researchers wrote the following:

Consistent with casual observations made by many commentators, our price indices confirm enormous housing price appreciation across China in 2003–2013. In first-tier cities, which include the four most populated and most economically important metropolitan areas in China (Beijing, Shanghai, Guangzhou, and Shenzhen) housing prices had an average annual real growth rate of 13.1% during this decade. Our sample also covers 31 second-tier cities, which are autonomous municipalities, provincial capitals, or vital industrial/commercial centers, and 85 other third-tier cities, which are important cities in their respective regions. Housing prices in second-tier cities had an average annual real growth rate of 10.5%; third-tier cities had an average annual real growth rate of 7.9%. These growth rates easily surpass the housing price appreciation during the US housing bubble in the first decade of the twenty-first century and are comparable to that during the Japanese housing bubble in the 1980s.

A major reason why this happened, though certainly not the only one, is that real estate in China turned into a massive asset bubble, a speculative wave in which people rushed to buy multiple homes as investment properties even though they were already homeowners. The Chinese government has tried to restrain this speculative boom, but not always successfully, and the real estate market in China remains an uncertain mess to this very day. Bringing more supply is critical to creating housing affordability, but only in the context of a different political economy that weeds out speculation from the real estate sector and provides greater financial stability for workers and families.

Recent Evidence on the Effects of Upzoning

There have been many recent studies that have analyzed the impact of upzoning policies on housing supply and housing costs, and in this next section I'll focus on a few of the prominent ones. In general, the research shows that upzoning does have a causal impact on boosting housing supply, but its impact on housing costs is usually so muted that it has no practical effect on the central issue of affordability.

In 2023, researchers at the Urban Institute published a major study that analyzed the effects of upzoning on housing supply and affordability in the United States. Upzoning is the process of changing local zoning regulations to allow for the construction of more high-density housing, like multi-family units. The Urban team built an enormous panel dataset for 1,136 cities across the United States covering the period from 2000 to 2019. The research group concluded that upzoning does expand housing supply, but largely for luxury and high-end rental units, not for lower-end units. They wrote the following:

We find that reforms that loosen restrictions are associated with a statistically significant 0.8% increase in housing supply within three to nine years of reform passage, accounting for new and existing stock. This increase occurs predominantly for units at the higher end of the rent price distribution; we find no statistically significant evidence that additional lower-cost units became available or moderated in cost in the years following reforms.

The researchers basically found that upzoning favors the construction of high-end rental units that are only affordable for people with much higher incomes than the median level. They also found that upzoning policies produced slightly lower median monthly rents compared to other jurisdictions that implemented downsizing policies. In other words, upzoning had a small effect, but overall it wasn't much, and certainly nothing comparable to the necessary scale for resolving our housing challenges.

Another big study from 2020 looked at upzoning policies in Chicago and concluded that, contrary to common wisdom, they actually had a negative impact on housing affordability. In 2013 and 2015, Chicago upzoned several large parcels of land. The economist Yonah Freemark examined the consequences of these policy changes and detected "significant, robust increases in values for transactions on parcels that received a boost in allowed building size." Freemark also concluded that "the short-term, local-level impacts of upzoning are higher property prices but no additional new housing construction." Here is yet another example of where localized upzoning policies, in the absence of other major constraints, actually backfired and producing higher housing costs.

A paper from 2023 studied upzoning policies in the Australian city of Brisbane over a 20-year period and concluded that "higher rates of new housing supply are robustly related to higher prices." The study looked at almost 26,000 sites across 19 major activity centers and observed that "activity centers with the highest new housing supply saw faster price growth, not lower growth." This central result still held even after controlling for zoned capacity and other factors. The researchers summarized their argument in the following passages:

It may be the case that planning controls are not a binding constraint on housing supply in Brisbane over our study period. But such an explanation still challenges the supply side argument in general. Brisbane dwelling prices increased by more than Sydney prices from 2001-2016, and over the whole period increased by more than prices in San Francisco (250% versus 235%); both cities that are routinely argued to be supply-constrained...These results suggest that changing planning systems to facilitate...density by increasing zoned capacity is unlikely to result in noticeable housing price effects...The regression results coupled with the overall descriptive data supports the idea that zoned capacity alone is not a major determinant of the rate of new dwelling supply, nor dwelling price. Only 22% of the sites with zoned capacity in 1996 had any redevelopment in the subsequent twenty year period. Despite all the development that did occur, zoned capacity in every activity center continued to grow over the 20-year study period, and in total zoned capacity was twice as high at the end of the study period than the start. Meanwhile, the increase in dwellings in these activity centres was in line with the city average, with many below it.

Many studies do find that upzoning helps to boost housing supply, but it’s prices that really matter, and there’s no solid evidence at all that upzoning brings down housing costs in any appreciable sense. In 2016, the city of Auckland in New Zealand enacted sweeping changes to its zoning regulations, upzoning roughly 75% of its residential lands. A major 2023 study later revealed that the upzoning program substantially boosted housing supply, but this particular study did not evaluate the effects of the policy on rental prices and housing costs in general. In the following year, the same research group did publish a study analyzing costs, and claimed that Auckland’s upzoning had produced rents that were roughly 28% lower than the counterfactual of no zoning reform. These results are relatively robust and have been validated by other research teams.

Although this study out of New Zealand did show that large-scale upzoning can help slow down the rise in rental costs, it’s also clear that this policy shift didn’t solve Auckland’s fundamental affordability problem at all, because housing costs continued to rise in Auckland long after 2016, just at a slower rate than they otherwise would have. It’s worth emphasizing that the authors end the paper by writing that "our findings...suggest that Auckland’s reform is not a panacea, and that additional policy tools and further reforms may be necessary to bring housing costs down to affordable levels."

There are several important caveats to mention about this Auckland study. First, their dataset stopped in 2022, which means they were unable to capture the enormous 7.8% surge in average weekly rent in Auckland that started in the following year, going from $579 in 2023 to $624 in 2024. This percent increase was higher than the annual percent rise in New Zealand as a whole (6.9%) for the first time since the upzoning was enacted in 2016. And second, Auckland’s annual average increase in weekly rent from 2017 to 2020, about 3.4%, was much higher than the rate of general inflation in New Zealand over the same period, which was under 2%. These facts collectively underscore the point that although Auckland’s reforms may have produced some positive impacts, and there's no question that they helped improve the situation slightly, they are woefully inadequate by themselves at addressing the broader affordability crisis plaguing the city.

A Bold and Simple Vision for the Future

What we really need to fix our crisis of affordability in the United States is an entirely new political and economic regime around housing. Here are a few ideas that, if implemented collectively, would do exactly that:

1. We need an expansive and ambitious program of public housing financed by the federal government, coupled with local economic development so that new neighborhoods are surrounded by parks, shopping centers, entertainment, museums, and other cultural and economic amenities.

2. We need a new regime of public financing that provides cheap borrowing options for developers, so that they're not discouraged from building because of high interest rates.

3. We need to nationalize, or collectivize through local cooperatives, some of the major developers so that the government and the people have a more direct impact on the pace of housing construction.

4. We need to heavily restructure the nature of real estate investment, either by imposing regional-specific annual caps on investment properties that can be purchased or through other regulations that make more of the existing housing supply available for regular people.

5. We need less wealth concentration at the top so that more regular people can afford to buy homes in competitive markets. That means we should promote unionization, higher taxes for the rich, and other policies designed to boost income levels for the working and middle classes.

6. We need to give the federal government broader powers to regulate zoning regulations at local levels, so that short-sighted local homeowners don't sabotage development because they're worried about their property values.

7. Finally, but perhaps most importantly, we need a total revolution in the meaning of housing, with the adoption of modularized design and manufacturing strategies that can produce rapid economies of scale and deliver major results in a short period of time. That means prioritizing things like functionality, affordability, and walkability over aesthetics and luxurious amenities.

If we're actually serious about solving this terrible social problem, then that's what it will take. To think otherwise is to live in a fantasy land. The current economic regime is broken and needs to be thoroughly revamped. We can have affordable housing, but it will take a different kind of society to get us there.

Trade Dynamics in War and Peace

Trade dynamics have always been driven by grand strategy during wartime. In 1806, Napoleon launched the Continental System against Britain, hoping to choke off European trade with the British. Enforcing the continental blockade proved to be a monumental effort that the French could never master, so violations among subject nations were common. Russia's repeated violations were a major reason for Napoleon's decision to launch a disastrous invasion in 1812 (see Figure 1). During World War I, the Royal Navy enforced a massive naval blockade against Germany, cutting it off from international trade and contributing to major shortages of consumer goods in the German economy. But wars are not just important because they constrain trade during a phase of intense violence. They also structure post-conflict trading regimes, as the dominant victors rewrite the rules and institutions of a regional or global order. To the victor go the spoils of trade.

Beyond war, imperial powers tend to embrace open trade in their ascendant phase and then turn against it on their way down. Britain embraced free trade in the middle of the 19th century when it dominated global manufacturing, then increasingly turned against it later on as competitors like Germany and the United States caught up. Economic historians like John Nye have pointed out that Britain's relationship with free trade in the 19th century was more complicated than what conventional wisdom holds. Britain mostly removed or reduced tariffs on manufactured goods, an area where it had massive comparative advantages and thus faced no real competition from anyone else in the world. But it still kept many tariffs on consumption goods, like rum, sugar, and most importantly French wine, which faced huge entry barriers to the British market. These wine tariffs were an important reason why British people drank far more beer than wine for several centuries. Only the Cobden-Chevalier Treaty in 1860 eliminated the major tariffs that both sides had slapped on each other, but Britain continued to impose tariffs on many goods and items throughout the 19th century, and its average tariff rates exceeded those imposed by France for most of the century.

The United States also embraced free trade after World War II when it dominated global industry, as much of Eurasia lay in ruins from the war and needed lots of time to rebuild. American manufacturing dominated the global economy, so it was easy for American policymakers to support free trade, since it practically meant that other countries would open up their markets to cheap American goods that faced little competition. But as countries like Japan and Germany reindustrialized and went up the value chain, producing increasingly advanced industrial products and vehicles, American companies faced greater competition and lost market share, even domestically in the case of the auto industry, when Japanese cars flooded the American market as cheaper and more fuel-efficient alternatives. Frustrated by its deteriorating position, the United States adopted a more belligerent trade strategy. In the 1980s, the United States pressured Japan to accept the Plaza Accords, designed to strengthen the yen so that Japanese exports would no longer be as competitive as before. The exact consequences of the Plaza Accords on the Japanese economy are still debated, but they signaled a new hostile direction from Washington, which had previously allowed its key puppets some economic leeway as a way of building up a strong bulwark to the Soviet Union in Eurasia.

The New Cold War and the Decline of American Hegemony

With the fall of the Soviet Union by 1991, the United States was on top of the world, or at least so it seemed. Once Japan faded as a competitor in the 1990s, Washington supported another round of trade expansion, allowing China to join the World Trade Organization in 2001. That opened up the United States to a flood of cheap Chinese goods (the "China Shock") and prompted an offshoring wave from American companies, though rising automation was also a critical factor in the gradual loss of American manufacturing jobs. Few American policymakers, however, anticipated the real scope of China's meteoric rise, and this sudden emergence on the global stage quickly burst Washington's sense of superiority.

The United States then shifted towards a path of preserving its strategic hegemony, a process that started with Obama's Pivot to Asia and continued with Trump and Biden. Obama was putting tariffs on Chinese tires as early as 2009. Over the past decade in particular, the American ruling classes have come to believe that they need to stop or slow down China's rise at nearly all costs and by almost any means. We are now in the beginning stages of a New Cold War, this time between China and the United States. It's precisely in the context of this wider geostrategic rivalry that America's turn against free trade starts to make more sense, as US elites now believe that the current global trading system is benefiting China more than anyone else, and thus US trade policy has shifted towards the active and deliberate sabotage of the current international system. It's the decline of American hegemony that's responsible for the breakdown of the global trade order.

Washington's desperate gambit is unlikely to work this time around. China has quickly become the greatest industrial powerhouse in world history. China consumes 30% of the world's energy, produces 55% of the world's steel, nearly 60% of the world's aluminum, one-third of the world's motor vehicles, one-third of global manufacturing output, 51% of the world's coal, over 60% of all electric vehicles, over 80% of the world's solar panels, and the list goes on and on. And because China's domestic market is now enormous, a rapidly increasing share of Chinese manufacturing output is consumed domestically. In addition, China's domination of intermediate industrial parts and goods is so comprehensive that no matter where in the world industrial manufacturers shift their final production of consumer goods, they will still need to import many of the critical components necessary for final manufacturing from China.

Furthermore, China's state capacity is enormous. The Chinese state can easily create hundreds of billions of dollars of domestic demand through monetization via lending from state-owned banks, since the financial system in China is controlled by the government. The world's four largest banks by assets are all Chinese and all owned by the Chinese government. China can easily wash away any lost corporate revenues from this trade war. On top of all this, China is not in Washington's strategic orbit. It's not an American satellite or puppet regime, like most of the countries in Western Europe. That implies it cannot be bullied into submission, which is the fate that Japan suffered in the 1980s. By contrast, the United States is internally divided. Even many Senate Republicans don't agree with Trump’s tariff strategy. There are also deep internal divisions within the Trump administration, not to mention that the country is headed towards multiple political and constitutional crises.

Under these circumstances, a new trade war is unlikely to reverse America's decline. If anything, it will rapidly accelerate it, but what follows the current order is anyone's guess. China is the world's manufacturing superpower, but will it follow industrial might with military might? Will it resort to gunboat diplomacy to expand its trade networks? Will it pressure and coerce trading partners into establishing a new international order? And how will the United States respond? Let us hope that Jan Coen's dictum foreshadows merely a trade war, and not a real one.

Burke’s General Argument

Burke argued that the revolutionaries were enforcing an abstract idea of liberty that was divorced from the practical developments of human life. Burke believed in promoting liberty, but only a kind of liberty that grew naturally over time and satisfied the needs of individual people. He posited that society should be guided by the inherited wisdom from past generations, dismissing the notion that liberty could be imposed on people by a rational government. For Burke, the French revolutionaries had no real knowledge of the challenges and complications of governing a big society. According to him, governance was a skill that could only be acquired through long and hard experience.

Figure 1: Capture of the Tuileries Palace by Jean Duplessis-Bertaux, depicting one of the most famous episodes of the French Revolution, when National Guard troops stormed the Tuileries Palace as a response to growing anger against the monarchy in the middle of a foreign invasion.

Burke distrusted the capacities of human reason, holding that reason alone could never be an adequate substitute for social customs and traditions. In this regard, he oriented his politics towards the past, like most conservatives still do. In one part of the essay, Burke laments that “the age of chivalry is gone” and that “the age of logical tricksters, economists, and calculators has taken over, and the glory of Europe is extinguished forever.” He also warns against experimenting too much with society, writing that “a spirit of innovation is generally the result of a selfish temperament and limited views.”

Instead of using abstract reasoning, Burke wanted those in charge to govern through prejudice, by which he meant the innate wisdom that we develop through experience. Rather than making revolutionary breaks with traditional institutions, Burke thought that people should only strive to change society by gradual and incremental steps. He deplored what he saw as the mob mentality of the “swinish multitude,” his infamous reference comparing common people to pigs. Burke believed, in effect, that major social and political decisions should be carried out by a land-owning aristocracy. He was extremely distrustful of the street protests and violent outbursts rocking France at the time. For him, allowing democracy into society could only mean allowing chaos and violence to spread everywhere. With this fundamental skepticism towards democratic agitation, he laid the groundwork for nearly all subsequent conservative thought in the Western world.

The Principles and Problems of Inheritance

Burke focuses his attention on the principle of inheritance, a kind of hereditary succession which he claims has allowed the English people to develop numerous liberties and freedoms over time without resorting to revolutionary upheaval. He writes the following:

You will observe that from Magna Carta to the Declaration of Right it has been the uniform policy of our constitution to claim and assert our liberties as an entailed inheritance that we have from our forefathers and are to transmit to our posterity, as an estate specially belonging to the people of this kingdom, with no reference to any other more general or prior right...We have an inheritable crown, an inheritable peerage, and a House of Commons and a people inheriting privileges, franchises, and liberties from a long line of ancestors.

Burke did not tell his readers that King John only signed the Magna Carta in 1215 because he had no other choice. John and his European allies had completely bungled a disastrous war against the French in the previous year, leading to widespread anger among the English aristocracy. On the verge of revolt, the wealthy barons offered John their continued loyalty if he would agree to sign a document that drastically reduced his power and control over the English economy. Although John went ahead and signed the Magna Carta, the barons accused him of breaking its terms and a full-blown war erupted for control of the English monarchy. The basic point is simple: the Magna Carta was not the product of some legitimate “inheritance.” It was an opportunistic power grab by a corrupt aristocracy furious over an idiotic king that had squandered their money on a losing war.

This example underscores how Burke vastly underestimates the power of radical breaks with the past to produce long-lasting changes that eventually come to be seen as “traditional” features of society. For another example, consider the dominance of Parliament in British politics. Was this development purely the result of a gradual inheritance from the Anglo-Saxons? You’d be highly naïve to think so. Parliament tamed the monarchy because it won a brutal civil war in the 1640s that culminated with Charles I on the chopping block (see below). It was precisely because the parliamentary forces triumphed in the English Civil War that future British monarchs never got any crazy ideas about the extent of their actual power. The English Civil War was Britain’s French Revolution. It was a radical rupture with the past that changed the future course of history.

Figure 2: Depiction of execution of Charles I in 1649, unknown artist.

This fact is totally ignored in Burke’s conservative reading of English history, and many English conservatives continue to think like that to this very day. They imagine the history of their country as being characterized by a series of gradual reforms that culminated in the present moment. Nothing could be further from the truth. Like virtually all other societies, modern Britain is largely the product of violent struggles. Of course, it should be emphasized that conservatives do not oppose all social and political changes. As Burke himself clarifies, “the idea of inheritance provides a sure principle of conservation and a sure principle of transmission, without at all excluding a principle of improvement.” In other words, society should only change in the context of existing institutions, traditions, and hierarchies. People should avoid making sudden changes that could threaten the existing order and the dominant social hierarchies within that order. That’s a core principle of conservatism as a worldview.

And yet therein lies the challenge. For the principle of inheritance to have any meaning, it requires political and economic stability. Otherwise, it becomes difficult to seriously claim that social order is inherited instead of imposed by those who the power to do so. Nothing encapsulates this problematic nature of Burke’s argument better than the following passage, probably the central passage of the entire text and the distillation of everything that’s wrong with conservatism, where he talks about the stability of the inheritance principle despite times of uncertainty and chaos:

Some time after the conquest great questions arose upon the legal principles of hereditary descent. It became a matter of doubt, whether the heir per capita or the heir per stripes was to succeed; but whether the heir per capita gave way when the heirdom per stripes took place, or the Catholic heir when the Protestant was preferred, the inheritable principle survived with a sort of immortality through all transmigrations...This is the spirit of our constitution, not only in its settled course, but in all its revolutions. Whoever came in, or however he came in, whether he obtained the crown by law, or by force, the hereditary succession was either continued or adopted.

There are so many fundamental flaws with this passage. An immediate issue is that it implies that someone can “inherit” the throne even if he comes to power through violent conflict, slashing and burning his enemies along the way! All the king apparently needs to do after his bloody rise is proclaim, “everything is good now that I’m in power, just remember that my kids and their kids will inherit the throne.”

Burke’s reasoning makes the principle of inheritance a justification for the same kind of chaos and anarchy that he claimed to despise in the French Revolution. Indeed, look at English history before modern times: it was dominated by massive succession wars and conflicts, many of which lasted decades, like the Wars of the Roses. That’s because English society became a mad scramble to seize ultimate power at all costs, given the enormous privileges granted to the monarch and his or her descendants. Once so much power is concentrated on a particular faction or family, just about anyone in the aristocracy would try to seize such a lucky position, which incentivized armed conflicts on an almost ritualistic cycle. But if a transition of power, wealth, privileges, or social liberties happened because of deadly force, it’s silly to pretend that the principle of inheritance actually applied during or after the transition. In such scenarios, it’s obvious that the principle of inheritance is serving as an ideological cover for the new regime to justify its rise to power, to make its domination seem normal and acceptable for the rest of society.

The Possibilities of Radical Political Change

Conservatives generally believe that the later phases of the Revolution, especially the Terror under the Jacobins, vindicated Burke’s skepticism and hostility towards radical government. But they did no such thing. It's true that the Terror produced many tragic outcomes and left behind thousands of dead innocent people. It's also true that many of its leaders fell in successive waves of persecutions. At the same time, it was during this most radical phase of the Revolution that French armies gained the upper hand and started winning the war. The great victory at Fleurus came in June 1794 when the Jacobins ruled the nation, not in 1792 when Louis XVI was warming his behind on the expensive sofas at the Tuileries. The Jacobins turned the war around before Napoleon delivered the coup de grâce with his invasion of Italy in 1796.

When the Jacobins first came to power in 1793, France was in total chaos internally and facing invasions from all sides externally. In a single year, they restored internal order and began defeating the European powers on all fronts. It was also under the most radical phase of the Revolution that the French abolished slavery in the colonies, in February 1794. The radicals also gave bread to the poor and protected them from the capricious swings of the market. In many ways, the radical phase of the Revolution is what saved the Revolution.

It’s heart-warming to see a conservative like Burke throwing his arms wide open for empiricism by supporting the collective experience of the past, but holding humanity hostage to the traditions of past generations makes social policy extremely static and inflexible, incapable neither of adapting to rapidly shifting circumstances nor of solving major problems. Change is the natural order of things. The most successful societies are those that can adapt and change quickly, sometimes even radically, to ensure their stability in the long run. Indeed, some circumstances demand that civilizations implement nothing less than a radical political program if they hope to survive. Shutting off the possibility to this kind of rapid social transformation would be downright suicidal. Societies ruled by corrupt elites interested in boosting their wealth and power cannot adequately respond to popular demands and concerns. Popular anger inevitably rises and society reaches a critical state of instability, like a pressure cooker about to burst. Conservative societies are sclerotic; they aim for order and stability, but produce chaos and instability instead, precisely because they cannot make the necessary changes to resolve the fundamental problems and tensions confronting them.

Nor should we be fooled by Burke’s moralizing against chaos and violence. Conservatives are more than happy to unleash violence on the poor and the weak if it means protecting the economic interests of the dominant classes. When their precious wealth is at stake, the rich will do whatever they can to defend the status quo. They have no problem obliterating anyone who challenges the system. The rule of the upper classes does not rest on wisdom or experience. It rests almost exclusively on power. Those who have power dictate terms to those who don’t, unless the latter decide to challenge and confront the roadblocks they face from the ruling classes. Burke’s real complaint is not about violence in some abstract sense; he’s specifically concerned about violence that comes from the masses below, violence that threatens the stale kabuki theater of a few corrupt plutocrats. Burke wants to prevent the reorganization of society, the redistribution of wealth, the reshuffling of power. He wants to maintain ancient economic hierarchies, not because they are useful, but because they serve the interests of the rich. He wants hierarchy for the sake of hierarchy. Because if history is clear about anything, it’s that rigid social and political hierarchies don’t produce order and stability, they produce eternal chaos and violence.