(1) Gatekeepers of Law: Inside the Westlaw and LexisNexis Duopoly

Westlaw and LexisNexis are duopolies that charge lawyers hundreds of dollars per hour for access to legal information that is technically public. The moat is less in the data, and amore in the organization methods: headnotes, key numbers, citations, and search algorithms that make millions of cases navigable. A lawyer could theoretically find almost any case for free on government websites like PACER, but the time costs and nuisance of searching for those cases, and quickly skimming them, alongside the pass through economics of legal costs, makes a subscription worthwhile. The nature of this duopoly kind of reminds me of Bloomberg terminals. The author argues that AI legal research is the first real threat to this duopoly in decades, as LLMs could potentially replace the organizational layer (although hallucinations are still a big problem… don’t get your legal advice from Claude). Great article, it’s worth a read.

(2) Age of Invention: How the Dutch Did It Better

One underrated part of European history is the rapid financial development in the Netherlands. The Dutch financial miracle was driven by many factors: (1) low customs duties offset by higher consumption taxes; (2) thirty living; (3) religious toleration; (4) the use of law-merchant, which resolved commercial disputes more quickly than civil courts; (5) transferrable bills of exchange, which created a liquid credit market around working capital (this is super underrated); (6) banking (again, the role of banks as the clearinghouse for bills of exchange is so underrated); and (7) public registries of land ownership and sales (which prevented title disputes and reduced risks for mortgages). The article covers these, and many more, but the common theme behind these factors were lower credit market frictions, which cut interest rates and increased investment.

(3) The Secret to Good Questions

“Questions are a crude way to measure career progression. As you get more senior, you spend less time doing the work and more time asking about it. By the time you reach the boardroom, your principal output is questions. […] What makes a good question obviously depends on the context. The “five whys”, a succession of “why” questions that are meant to uncover the root cause of a problem, originated in Toyota and has spread from there. But what works well as an investigative technique in a specific environment is not going to help in forging relationships. (“Why were you absent yesterday?” “I was at a funeral.” “Why?” “What do you mean, why?”).”

(4) The Left’s Common-Sense Problem on Crime

Some US progressives believe that prisons should be abolished, stealing is okay when it targets greedy corporations, and killing CEOs can be “an effective act of raising political consciousness.” These beliefs are about tribal signaling, not well-reasoned policy views that account for second-order impacts. The Economist phrases the issue perfectly: “What many of these ideas have in common, besides moral idiocy, is that they ignore incentives and knock-on effects. Yes, America should lock up fewer non-violent offenders. But if criminals can’t be locked up, what deters them from mugging you? Yes, some people struggle to afford groceries. But theft is hardly a solution to the affordability crisis. When some people steal, the prices rise for everyone else. [The] left’s worst ideas about economics [also] gloss over incentives. Rent controls deter construction; rules making it hard to fire workers discourage firms from hiring them in the first place; and so on.”

(5) SharkNinja Built a $6 Billion Empire Selling Gizmos Nobody Needs

SharkNinja may be the first appliance company built around the “reveal video.” Their mops use black-water tanks so buyers can see the obscene amount of dirt; their “FacialPro Glow” collects whatever it sucks from your pores in a lighted cartridge; their ice cream and pizza machines turn slop dinner into slop content. Founder Mark Rosenzweig got his start selling vacuums on Home Shopping Network and replicated the live entertainment on social media, which worked because, in some sense, TikTok is just a larger infomercial channel. The Ninja Slushi sold out ten times and built a 170,000-person waitlist, while a video of its TurboBlade fan received more than 75 million views. One of its executives admits, “There’s nothing we make that you actually need.” But after doing this for 25 years, they’ve clearly found a repeatable formula with making viral appliances.

(6) How Robocallers Outwitted the Government and Wrecked the Do Not Call List

Congress passed the Telephone Consumer Protection Act (TCPA) in 1991 and created the Do Not Call (DNC) Registry to stop telemarketers. It worked against legitimate companies. It did nothing against scammers, in part because the penalties are civil, not criminal: the government can fine violators but has no practical ability to collect from the people who simply ignore the judgments (or who are judgment proof overseas). Meanwhile, VOIP made it trivial to acquire thousands of phone lines cheaply. Autodialers and robo voices scaled the operation, and then spoofing made caller ID useless. The FCC introduced STIR/SHAKEN to authenticate caller identity, but enforcement is patchy. This is a good lesson in legal realism… following the law is basically optional if the violators have no assets to seize and no fear of prosecution.

(7) The Fugitive Prince

In September 1940, Romania’s King Carol II’s armored train hurtled toward the western border carrying his side piece, three poodles, and 30 truckloads of the country’s gold. He escaped into exile in Portugal. Decades later, Paul Lambrino (one of his grandsons) spent his entire adult life trying to reclaim the throne and his “stolen” inheritance. In the 2000s, the Israeli billionaire Beny Steinmetz financed “Project Prince,” a scheme to recover royal assets worth >$1 billion through Romanian restitution courts. They won the transfer of a valuable plot of land, the Baneasa Royal Farm, allegedly by bribing a state official on a yacht in Monte Carlo. Romania’s anti-corruption directorate indicted the entire group in 2015. Paul fled across Europe, and Steinmetz was later arrested in Greece. But then, the twist: courts in Greece, Cyprus, Malta, and France all rejected Romania’s extradition requests, and Interpol deleted the arrest warrants, concluding that the prosecution involved “political elements” and secret-service wiretapping. It’s a wild story.

(8) Working Title

Why is title insurance so expensive? Title insurance has a loss ratio of 5%, so for every dollar of premium collected, 5 cents goes to paying actual claims. Auto insurance runs over 70%, and homeowners insurance sits near 60%. Title insurance exists because the real estate lawyer lobbies embedded this in the standard closing process decades ago. Buyers pay for it, have zero choice in the matter, and never shop around because their attorney handles the selection (and often are required to use a particular insurer in their mortgage documents). Every participant in the closing process benefits except the homeowner. It’s one of the most broken parts of the insurance industry, and it is embedded because of state regulations, default behavior, and opacity.

(9) Should We Do More to Fight Lookism?

We build institutions against racism, sexism, and ageism, but one of the largest inequalities of all goes unnamed. Attractive people earn more, find partners more easily, and even win in court more often, while the unattractive are judged more harshly and rated lower at everything. Some philosophers call this “lookism” and argue that the taboos with ugliness is itself part of the injustice: “[Lookism] does not generate the same awareness as gender or race-based discrimination partially for the reason that there is a different kind of discomfort associated with being identified as ugly.” People proudly identify by race or gender, but nobody wants to identify as ugly, so there is no National Association for Unattractive People, no HR training, and no protected group of ugly people. I wonder what reparations would look like for ugly people.

(10) How Chris Hohn Built the World’s Most Profitable Hedge Fund

Chris Hohn, 59, is the closest thing Britain has to Warren Buffett. His Children’s Investment Fund (TCI) is the fifth most profitable hedge fund of all time, running roughly $77bn in just 15 positions with about half a dozen analysts (Rishi Sunak was once one of them). His filter is pricing power: monopolies and duopolies that can push through inflation-busting price increases, which is why 31% of the portfolio is infrastructure: two Canadian railways, Spain’s Ferrovial, France’s Vinci. His biggest position is GE Aerospace, on the thesis that there hasn’t been a fundamentally new aircraft engine in over 50 years. Very few businesses have durable moats; rail competes with trucks, toll roads with side streets, and almost everything competes with “just don’t.” A portion of TCI’s fees goes straight to his foundation, one of the UK’s largest; he gave away over £1bn last year, drives a Toyota Prius, and wears a cheap plastic watch.

Miscellaneous

Why Berlin’s airport is struggling. Hemingway’s Hills Like White Elephants. Killing mosquitoes with miniature drones. Betting market for weddings. Some development lessons from the failed Chinese SEZ Shantou. The Economist’s favorite new topic is air conditioning in Europe. Creative way to get an info edge. Growing importance of the Panama Canal. Chinese retaliation over CK Hutchison losing the Panama ports. Leopold could have used a bit of situational awareness. Combined review of three Napoleon biographies. Origins of tourism in St. Barth. The national beer of Laos.

(1) Revenge of the Realtors: The Procompetitive Case for Consolidating Multiple Listing Services

Multiple Listing Services are the invisible infrastructure of US housing. The hundreds of balkanized local property exchanges emerged in the 1880s when real estate brokers met up to swap listings and never evolved into a national system (unlike, for example, Rightmove in the UK). The near-universal view is that local MLSs (mostly organized as cooperatives owned by realtor associations) are anticompetitive cartels and should be eliminated, but this article takes a contrarian view: it would be good to consolidate MLSs. Zillow, Redfin, and Opendoor all depend on MLS data feeds to exist, so mega-MLSs could standardize data and cut transaction costs. The consolidation wave for MLSs, which started two decades ago, is partly a defensive move to build footprints large enough to bargain with the national portals. I disagree with the argument, but it’s still worth reading for the history of a very unusual dual-sided marketplaces plus cooperative business model with local economies of scale. If anyone figures out how to disrupt the MLS, they will have a trillion dollar business on their hands.

(2) The Incredible Story of the Cartel Olympics

In February 2023, Mauricio Morales, a former Olympic runner who worked at UN refugee camps, was kidnapped off a Mexico City street by members of La Union Tepito. After days of torture, he was brought before a cartel boss named Don Paco, who noticed the Olympic rings tattoo on his wrist. Don Paco explained that Mexico’s biggest cartels organize a secret annual tournament: soccer, flag football, and boxing, where rival bosses gather to place bets and conduct business without shooting each other. Mau was told to coach and play on La Union’s flag football team. They told him: “If you win, you live. If you lose, you die.” Months of training in a prison facility followed. The story sounds like fiction, but it was verified with UN contacts and human-rights investigators, and other sources. Crazy times south of the border.

(3) Notes on Panama

Matt Lakeman is a great travel writer, and this is his old article on a journey through Panama. Panama is essentially two countries: the Canal Zone and everything else. The Canal generates about 6% of Panama’s GDP, but it shapes the entire economy through the logistics, banking, and clustered financial services. One of the weirdest on-the-ground impacts of this financial services clustering is the skyline of Panama City, comprising dozens of huge residential skyscrapers that are (1) mostly empty, (2) stuck in the middle of slums, and (3) mostly meant for South American money laundering. He also covers the history of Panama’s military (abolished after Noriega was ousted by the US), various stories about the Darien Gap, and his impressions of the vibe of a country that exists primarily because of a single piece of infrastructure. For more history on Panama, I’ll plug a strong recommendation for David McCullough’s The Path Between the Seas.

(4) How Did TVs Get So Cheap?

A 50-inch TV that cost $1,100 at Best Buy in 2001 now costs under $200, not inflation-adjusted but in raw nominal dollars. So how did TVs get so cheap? The answer is almost entirely about LCD manufacturing and the scaling of “mother glass,” the enormous sheets of ultra-clear glass onto which thin-film transistors are deposited via photolithography. Generation 1 mother glass was 12 by 16 inches; Generation 10.5 is 116 by 133 inches, nearly 100 times the size. Because equipment costs scale more slowly than glass area, going from Gen 4 to Gen 8 cut equipment costs per unit of LCD panel area by 80%. The result is a manufactured good that has fallen in price per area-pixel by over 90% in the last two decades. The story combines almost all of the standard methods of improving manufacturing efficiency: geometric scaling, process learning, yield improvement, and massive capital investment cycles driven by Korean and Chinese panel makers. It’s probably the best case study of manufactured deflation in the modern world.

(5) Liechtenstein: The Magic Princedom

Liechtenstein has about 38,000 people, is 15 miles long, has no airport or railway station, has dissolved its army, and has more registered companies than citizens. The country was literally purchased in 1712 by the Austrian Liechtenstein family, who needed a principality to secure a seat in the Holy Roman Empire. 300 years later, the same family still rules. Prince Hans-Adam II rebuilt the family fortune after WWII confiscations, turning a micro-state into one of the world’s wealthiest country per capita. The 2003 constitution gave the prince veto power over parliament but also handed citizens the right to abolish the monarchy by referendum (they voted to keep it). For more information, check out this fun video about Liechtenstein.

(6) Dismissing Europe as a museum is glib; think of it as a theme park

“From Barcelona to Bruges by way of Florence, vast swathes of Europe’s once-idyllic cities are a procession of foreigners shopping for souvenirs. Hilltop villages in Provence or Tuscany seem to exist chiefly as backdrops for linen-clad types sipping spritzes. […] In summer Europeans are not so much residents of their home cities as extras in someone else’s social-media feed. […] These days, it is better to think of Europe as a theme park [than a museum;] museums preserve history [while] theme parks offer [visitors a] feeling of being part of it. [But] any European tempted to be flattered should remember that being a lifestyle superpower is not nearly the same as being the real thing. While America flogs artificial-intelligence models and China monopolizes critical materials, only [Europe’s] cramped peninsula seems to be able to satisfy the global middle class wanderlust. Who needs big tech when you have the Eiffel Tower and the Parthenon? […] Wary of anything that smells like economic success, Europe is busy undermining itself as a destination [with too much success, and] tourists are getting [the truly] authentic local experience: sitting next to other tourists, and doing their utmost to pretend they are not there.”

(7) Kim Kardashian, Ryan Reynolds, and the Age of the Celebrity Brand

Consumer brands used to dominate using shelf space and TV advertising. Now the distribution runs through celebrities who own the brands. Before, Michael Jordan launched Air Jordan with Nike; now Kim Kardashian owns SKIMS (doing $1 billion in annual sales) and Ryan Reynolds sold Aviation Gin to Diageo for $610 million before cashing out to Mint Mobile to T-Mobile for $1.35 billion. So what changed? The shift happened because social media gave celebrities direct access to consumers without needing to pay for TV ads or retail store placement, plus DTC shipping logistics and Chinese manufacturing that enabled drop-shopping. I do wonder if these brands are actually enduring, since the marketing is built entirely around one person. Not all celebrity brands work either: “Wrestler Hulk Hogan’s line of frozen cheeseburgers were a flop mostly because they were awful.”

(8) Samurai City

Tokugawa Japan maintained the peace for nearly 300 years with a tightly designed, and highly unusual, system of social and physical control: “Between 1600 and 1868, Japan was dominated by the Tokugawa family. […] The apex of [their political] system was the city of Edo (today’s Tokyo), at times the largest city in the world, and one of the strangest urban structures in history. [….] Tokugawa Edo stands as a monument to the power of rent-seekers who produced little and demanded immense resources as a condition of civil peace. […] These forces [resulted in] a city with tight restrictions on built density and land use, acute housing shortages, and a huge share of privileged state stipendiaries [within the samurai class]. Edo is a particularly striking example of these phenomena. But it is far from alone.”

(9) The $11 Billion Casino-Style Economy Built on Players Who Can Never Cash Out

Social casino games are slot machines where people win fake money. They get around gambling laws because the players never cash out. The App Store is, in effect, the world’s largest casino: if it were a single gambling venue, it would generate more revenue than the entire Las Vegas Strip. The $11 billion social casino economy is driven by whales: 1% of gamers account for about 64% of game billings in the App Store. Apple is one of the biggest winners in this… mobile gaming is half of the App Store and an estimated 10-15% of Apple’s earnings. “Apple makes more from games than from selling laptops.” Kind of weird that the iPhone is mostly a gaming console.

(10) Herding the Fro-Yo Sheep

“This summer, every trendy dessert joint in NYC has a mile-long line of transplants and tourists. One New Yorker is protesting in his own way—by “baa”-ing at them. Fifty people were standing in a line on 7th Avenue […] waiting to purchase cups of frozen yogurt from the new West Village location of Myka, a Madrid-based dessert chain. Walking alongside the line, he shook his head in disbelief. “Baa!” he shouted at the idling fro-yo pilgrims. “Baa! Baa!” […] Anyone who waits in a line for fro-yo must be a transplant who needs to move back to Ohio, […] [ Some people said “This is a microcosm of everything wrong in current culture” and “In Spanish we say ‘Pendejos.’ ” [But it is this] aspect of the line phenomenon—that the behavior is driven by algorithms and the influencer economy—that gets under [his] skin. […] The shepherd of @sheepofnyc may be young, but he is still wise enough to understand that on-line life, like online life, is not really living.”

Miscellaneous

Apparently the new name for recruiters is talent engineers. Tech land is full of zombie unicorns. The most successful founders are retardmaxxing aura blasters. Forget about coding, people are vibe lawyering. 1 in 7 US college students have the literacy levels of 5th graders. Ken Griffin buys all 125 units in Miami condo building. Updates on US-Cuba negotiations. Good advice for managers. The old masters paintings are back in vogue. Declining productivity of agriculture in Africa. Why soccer involves luck more than other sports. Trump is hollowing out the DOJ. The future of work with ChatGPT Voice. Cool website with the national inventory of dams. Betting market for weddings.

(1) Netflix and the Hollywood Endgame

Ben Thompson explains how Netflix won the streaming wars. Traditional Hollywood operated in horizontal layers: studios made content, distributors moved it, and exhibitors showed it. Netflix collapsed that structure into a vertical stack. Disney+, HBO Max, and Peacock bet that content libraries would drive subscriptions. Netflix bet on distribution, and distribution won. He claims that Netflix had a “data advantage” that compounded over time, and that the studios couldn’t replicate this by starting from content alone. I’m actually sort of skeptical of this explanation, since it seems like he’s pigeonholing Netflix into aggregation theory. The more nuanced explanation probably involves (1) Netflix’s early underpriced streaming rights contracts (it basically functioned as content arbitrage to “create” content), and (2) a classic innovator’s dilemma for the studios of not cannibalizing cable TV and movie theaters with streaming. Eventually they had to bite the bullet, but by then, Netflix had a massive subscriber lead.

(2) 100 Lessons From Building A 100-Person Org

Some highlights: (1) recruiting is a long game; (2) the company brand compounds; (3) never hire out of desperation; (4) everyone wants to work remote, but very few people are disciplined and self-motivated enough to make it work; (5) do a post-mortem after every hiring mistake; (6) try not to attribute hiring mistakes to “they sucked” (why did you hire someone who sucked?); (7) “management sucks, be suspicious of people who want to become managers;” (8) the manager doesn’t have to do the IC job, but they should be able to do the job; (9) it is harder to run a 20-person org than a 100-person org because systems and communications break between 20-40 people; (10) “hiring is the highest-leverage thing anyone at the company can do”; and more…. Hiring lessons are easy to understand, but hard to internalize, and best learned by experience.

(3) Steam Networks

Most people don’t know that Manhattan still runs on steam heating. Con Edison operates the largest commercial steam system in the world: 105 miles of pipe delivering steam to over 1,700 buildings for heating, cooling, and sterilization. The “district heating” system dates to the 1880s and peaked in the mid-20th century when steam was the default heating method for most large buildings. It has classic utility-like economics: central steam production achieves efficiencies that individual building boilers cannot, and network density makes the system more efficient as more buildings connect. It’s a good reminder about switching costs and path dependence in designing cities, and some cool history about a now-forgotten kind of local utility (for others, check out the history of town gas and gong farmers).

(4) Could a Handwritten Letter Help You Buy a House? Or Convince You to Sell?

“What is it about handwritten letters that have such power? I’m not the first person to have written a letter in an attempt to convince someone to sell them their house, because it often works. […] For those clients who feel emotional about the sale of their house, who have lived there for a very long time, who are downsizing or finding the idea of moving on difficult, they want to feel the new owner has fallen in love with their house. They want to like who they are selling to. […] For a buyer, writing a personal letter can also be a way to claw back a sense of control in what can be a disempowering and transactional process. […] If you don’t ask, you don’t get and, really, what have you got to lose?”

(5) How Will OpenAI Compete?

Benedict Evans lays out four key vulnerabilities in OpenAI’s position. (1) No durable competitive lead: only five percent of ChatGPT users pay, average usage sits at three prompts per day, and the company admits to a “capability gap” between what models can do and what people actually do with them. (2) No distribution advantage: Google, Apple, and Meta already own the surfaces where billions spend their time, and chatbots all look the same. (3) Staggering capex without the cash flows to fund it: the big four cloud companies spent $400 billion on infrastructure last year; OpenAI tries to match this through fundraising and other people’s balance sheets. (4) OpenAI is starting from the technology and working backwards toward a product; Steve Jobs said to start with the customer experience and work backwards to the tech, but OpenAI did the opposite.

(6) How Corporations Use the Texas Two-Step to Avoid Asbestos Lawsuits

Companies have faced billions in asbestos liabilities since the 1980s. A decade ago, some clever lawyers found a workaround: the Texas Two-Step. Step one: reincorporate in Texas, which has a unique “divisive merger” provision that lets a company split in two without triggering fraudulent transfer rules. One entity keeps the operating business; the other keeps the liabilities. Step two: move the liability entity to North Carolina, which is unusually lenient about allowing these cases into bankruptcy court. The goal is not to eliminate the liabilities entirely but to cap them through a Chapter 11 trust, so the bankruptcy becomes a negotiating play for a global settlement rather than an escape hatch from all of the liabilities. J&J, Georgia-Pacific, and Trane have all used the maneuver. One lesson is that this trick exploits a gap between state-level corporations law and federal bankruptcy law (which allows the bankruptcy stay and global settlement method). The final verdict on the Texas Two-Step is still unclear, but for now, the lawyers are dancing.

(7) Why Probability Probably Doesn’t Exist (But It Is Useful To Act Like It Does)

Ignoring the fact that saying “probably” is (probably?) a statement about probability… this article from Nature is a good reminder about Knightian uncertainty—some risk is just unquantifiable. There is a temptation to calculate probabilities for everything, but the real world is messier than coin flips or casinos games; we don’t have enough information to know most probabilities. Even if we think we know the probability, what is the probability that we’re wrong about our estimate of the probability? In one sense, it’s a pointless and nihilistic academic argument. But in another sense, it’s a good reminder that we are probably overconfident in our estimates of probabilities.

(8) The Decline of Outside Magazine

“He laid out a vision of Outside, Inc., as a tech-media empire, “an Amazon Prime for the active-life-style participant,” with subscription numbers akin to those of Disney+ or Netflix. Rather than depending on the volatile advertising market, the company would get recurring revenue [from] a membership program, Outside+, which would provide unlimited access to the magazines’ articles and other perks. Thurston has bought up non-media brands, including Gaia, a mapping app; FinisherPix, a photography service; Inntopia, a travel-booking software; and Trailforks, a trail database. (Last year, Thurston also reacquired MapMyFitness, his first company, from Under Armour.) He estimated that the worldwide audience interested in healthy, active life styles was at least a billion people, and argued that the industry was recession-proof.”

(9) Why Is It So Hard to Build an Airport?

Aviation drives 8% of global GDP, yet the U.S. has built more commercial nuclear reactors in the past 25 years (two) than major commercial airports (none). The average large U.S. airport is 82 (!) years old. Why don’t we build more airports? The core problem is noise. Jet engines in the 1950s were a huge nuisance (much worse than earlier piston engines), and longer airport runway configurations exposed vastly more people to that noise. Citizens filed billions in lawsuits. Courts ruled that noise constituted a “taking” that required compensation, and some airports had to buy and demolish entire neighborhoods via eminent domain to fix the issue. Technology has helped (turbofans replaced turbojets, for example), but rising traffic has offset those gains, and we are still stuck in a gridlock of it being politically impossible to build airports close enough people to be useful, while hemming in existing airports as once-far-away cities expanded to be near them.

(10) Why Are There So Many Unfinished Buildings in Africa?

Short answer is weak banks, poor land titling, and grabby relatives. Half-made buildings are everywhere in African cities. The reasons stack up: (1) Developers cannot get loans without collateral, and mortgages are rare, so households build incrementally as they earn money; (2) many Africans “save in concrete,” tying up their capital in buildings rather than in liquid savings using bank accounts, which creates a vicious circle that limits bank lending (since they don’t have banking relationships); (3) visible construction leads to targets on the backs of the homeowners as being wealthy (because family members ask them for money); and (4) weak property rights encourage people to pour concrete to strengthen their claims to land. The fix is straightforward in theory (better banking, property rights, and functional tax systems) and impossible in practice. I had never really thought of concrete or bricks as a store of value before.

(0) Miscellaneous

US colonization project in the Pacific islands. The buying cemeteries PE guy. Pending nationalization of Thames Water. Collapse of Spirit Airlines. Capitalism at its finest in franchising. Television’s “killer app” was sports. Importance of modern drywall. German grocery chain Aldi expands in the US. Very cool gift for anyone who likes airplanes. Weird story of a CIA gold heist. Relationships and deep value investing.

(1) Conquest Of The Incas

Excellent long-form article about Pizarro’s conquest of the Incas. For those that forgot world history class, Spanish conquistador Francisco Pizarro infamously conquered the Inca Empire at Cajamarca in 1532, which was the most lopsided battle in history… his 168 men defeated the 100,000+ strong Incan army without losing a single life (!). This article covers the wider history and context of his conquest: Atahualpa’s room full of silver that was paid as ransom, the internecine native politics in the march on Cuzco, the puppet rule of Manco Inca, Pizarro’s lessons from Cortez’s earlier conquest of the Aztec Empire (and why Cortez is the greatest of the evil conquistadores), development of the encomienda system, and Incan perspectives on the conquest. This was the best non-book summary of the Incan conquest that I have ever read (for the best book, see The History of the Conquest of Peru). Worth a read if you’re into that sort of thing.

(2) How to Argue for Libertarianism

See also: Problems with Libertarianism. The logical chain starts from the posited ethical goal of “greatest good for the greatest number of people” (for a definition of “good,” see Parfit). To create “goods,” actions must be taken by people, and those actions require coordination. But effective coordination requires information, which is decentralized (in the minds of people). Markets enable coordination by revealing information through prices (the price mechanism). More informed decisions are good, and decisions made by decentralized participants (using prices) are more informed than centralized decisions made with centrally gathered info. The problem with centralized allocation: “Actors in private markets pay most of the costs of their actions and receive most of the benefits. Actors in political markets rarely do either. In private markets, market failure is the exception. In political markets, it is the rule.” In some sense, the core of libertarianism philosophy is that markets are the best guide to decision-making.

(3) Amazon’s Durability

Amazon recently launched Amazon Supply Chain Services (ASCS), which allows third parties to use Amazon’s logistics network of ocean freight, trucking, last-mile bicycle delivery, and more. This announcement is a continuation of Amazon’s externalization of internal services: first with FBA (for on-Amazon 3P sellers to use their 1P logistics network), then AWS (for datacenters), and now logistics for anyone (not just Amazon 3P sellers). “Amazon, more than any other company, operates with decade-long timeframes, consistently making real-world investments at massive scale that (1) convert their marginal costs into capital costs and (2) gain leverage on those capital costs by selling them to other businesses.” The article focuses on Amazon’s AI orchestration layer, which is called Bedrock, as their next platform-as-a-service bet, which contextualizes the various decisions that Amazon has made (such as designing their own datacenter chips, not competing on building frontier models, etc.), which all make sense with a long-term view of the market. Other long term bets: drones; satellites; robotics; and much more.

(4) Electrocloud!

The once-legendary Sam Insull built the modern electrical utility with two insights on economies of scale: (1) centralized power generation is more efficient than distributed generators, and (2) utilization can be increased by combining the varied load factors of industrial daytime demand with residential evening use. AWS did the same for servers a century later by spreading demand across millions of customers with different usage patterns. Both electrification and cloud computing changed industrial behavior from discrete to continuous, which reduced uncertainty in unit economics costs (by turning fixed costs into variable costs) and spurred an increase in startups (first factories, then software). This article draws some parallels between electric utilities and datacenters, including the 1920s bubble and Insull’s disastrous end. Everyone was ultimately right about electricity demand, but wrong about the stocks. Maybe this time is different?

(5) You Can’t Tell People Anything

“I’m convinced that a new idea or a new plan or a new technique is never understood when you just explain it. People will often think they understand, and they’ll say they understand, but then their actions show that it just ain’t so. […] Without some kind of direct experience to use as their touchstone, people just don’t have the context that gives them a place in their own minds to put the things that you are telling them. […] What you have to do is find a way to give them the experience, to put them in the situation. Sometimes this can only happen by making real the thing that you’re describing. […] I expect none of you to really get what I’m talking about, because this principle also applies to itself. But I fully expect I’ll get the occasional email saying “Oh! so that’s what you meant.” or “Why didn’t you tell me that?” I did, but you can’t tell people anything.”

(6) Crumbl Cookies Has Private Equity Backing And Unhappy Franchisees

The story of the hottest new food franchise concept, Crumbl Cookies. Crumbl grew from a single Utah store in 2017 to over 1,100 locations in eight years. The founders had zero baking or franchising experience, but they knew how to sell in the age of TikTok. Their approach was to copy the hype cycle of “sneaker drops” with weekly releases of limited edition cookies (I guess people get excited about flavors like the Almost Everything Bagel Cookie?). Underneath their tasty videos are some unsavory franchise economics. Buildout costs for each new location are >$1.3 million, average box-level net profits are $251,706, and median box-level net profits are just $77,359 (a brutal ROIC). Meanwhile, the average store revenue fell 37% from 2022 to 2023, while Crumbl flexed pricing and raised its royalty rate to 8%. Franchisors are ultimately in the business of selling businesses, so the cookies won’t be the only thing to crumble.

(7) The Dilemma That Brought Down Kodak

George Eastman was a bank clerk who hated hauling around camera equipment. He developed dry plates in his mother's kitchen, and, in 1888, introduced the first Kodak camera. At its peak, Kodak controlled 90% of the US film market and 80% of camera sales, and was the most recognizable name in consumer technology. But in 1975, the digital camera arrived. Digital cameras threatened their razor-and-blade model with the sale of proprietary film, processing, and prints (because digital cannibalized them). So Kodak held off from undercutting its core products. Meanwhile, Sony, Canon, and Nikon invested heavily into digital cameras, overtook the market, and pushed Kodak into bankruptcy (in 2012). The story is both a canonical example of the “innovator’s dilemma” and a reminder of the evanescence of even the most durable businesses.

(8) Owning Distribution or Fulfillment in Consumer-Facing Last-Mile Logistics

Reflections on where the value will accrue in last-mile delivery… “It’s hard to imagine Uber having much pricing power if [ride-sharing] is a frontend for Waymo and meal delivery [is] a frontend for Waymo. It is also possible to imagine a setup where there are different AV and drone delivery companies with inconsistent adoption in different markets, and that all of these businesses have enough technical and regulatory risk to keep them busy and artificially slow their expansion. [Uber] may have locally weak pricing power in cities that shift to [AV fleets] under a single brand faster, but [Uber] would still have the national brand. When someone lands in an unfamiliar city, are they going to Google to find an AV company operating locally to get them from the airport to their hotel [or will they just check Uber]?”

(9) The "Chicken of Tomorrow" Contest and the Story of "Big Chicken"

Before the 1950s, chickens were rarely consumed for meat (they were used mostly for eggs). Everything changed with the USDA's "Chicken of Tomorrow" contest, which evaluated over 2,000 chicken varieties on 18 criteria, ranging from body structure to feed conversion efficiency, to find the ideal chicken. The winning breeds were, with small variations, the chickens that we still eat today. The complex cross-breeding in this competition produced ideal "genetic IP" for chickens that is now locked up by a few big hatcheries in the US, effectively turning chicken genetics into proprietary IP (which is still monetized with every chicken to this day). The entire modern poultry industry traces back to this single government competition. Kind of crazy that this obscure 1950s USDA contest is the reason why we are all eating Chick-fil-A today.

(10) What Makes Art Great?

The title is sort of misleading. This is really about why AI is bad at writing, and what that reveals about what makes writing (and art) good. The core argument: great art is not predictable or obvious, but surprising. "This property of constantly destabilizing the reader is, I think, also a general property of great works of art. They are constantly breaking their own forms, subverting them, playing with the reader in a way that requires us to rise to their level. They are extremely strange." Think of Moby-Dick’s sixteen pages of whale quotations from various books right off the bat. Writing can surprise in many ways: word choice, grammar, sentence structure, plot, form. There are also subtle rhythms like rhyming: assonance, patterns that run through the text, layers alluding to other layers (for the best example of this, see Godel, Escher, Bach). AI writing sounds stilted because it optimizes for the most probable next token, which is by definition the least surprising one. Great writing does the opposite. To "charge" language with meaning is to be surprising, to have layers of patterns, and to produce something with unique depth. LLMs are prediction machines, and art is about being unpredictable.

Miscellaneous

50% of US men under age 50 have a sports gambling account. Reflections on Norway. The secret engineering of landfills. Tyler Cowen interviews Craig Newmark. A life of chivalry in the Middle Ages. The expansion of Chobani yogurt. M&A in the elevator industry. eBay gets takeover offer from Ryan Cohen. Price matters more than moats. Thoughtful corporate culture at Clay. How modeling created Miami Beach. Military automation with Maven. NYC set to simplify the environmental review process for buildings. Amusing and critical review of Rene Girard’s philosophy. Very technical explanation of Codex computer use features. Productivity and personal identity.

(1) Sam Altman May Control Our Future—Can He Be Trusted?

The scariest sentence in the English language is “Ronan Farrow wrote an article about you.” Farrow broke famous investigative stories on Harvey Weinstein, Britney Spears, and now… Sam Altman (nightmare blunt rotation). The article alleges that Altman is dishonest and covers endless stories of him “not being consistently candid” with investors, employees, and the public. My honest impression of the article is that it is a biased hit job, and I almost, sort of, feel bad for Altman. The article indirectly seems to show that the allegations are flimsy, the journalist is biased, and the accusers are kind of hypocritical: Ilya complains that Sam cannot be trusted and secretly starts a boardroom coup; Amodei says that OpenAI is not “non-profit enough” and launches a for-profit competitor; the list goes on. “Flexibility with the truth” is pretty common in venture-backed startups; the story makes Altman seem more honest than most. Credit where it is due… Altman is the only founder to admit that his startup is not-for-profit.

(2) History of Taylor Corporation

Glen Taylor is the richest man in Minnesota. He is mostly known for his public-facing investments like the Timberwolves, Lynx, and the Star Tribune, but most of his wealth comes from his private company Taylor Corporation. Taylor Corporation is a printing company that started out printing wedding napkins, expanding over several decades with 200+ acquisitions of specialty printing companies in stationary, graphic design, birthday cards, and more (with random diversification into community banks). Taylor keeps a super low profile: “Only those who live in the Mankato area or happen to run their own printing companies have heard of Taylor Corporation—maybe. They operate under a variety of names, many of them acquisitions, like Web Graphics Midwest, Litho Tech, Ad Graphics, LabelWorks, and Precision Press—not exactly the stuff that grabs the attention of Forbes.” Who knew a wedding napkins rollup could be a billion dollar business?

(3) Crony Beliefs

There are two broad types of beliefs: merit and crony. Merit beliefs help us navigate the world, and have consequences that cause us to update those beliefs—is the bridge safe to drive on, is this software useful, is this burger good, etc. In other words, merit beliefs have a feedback loop from our actions, where being right or wrong changes some outcome in our lives. Crony beliefs are those where being right or wrong is irrelevant for our lives—climate change, Trump, most of politics, etc. The article argues that crony beliefs are mostly about social signaling of our values and identities, serving social cohesion rather than truth. I think the framework is too broad (e.g., are obscure opinions about Napoleon’s battlefield tactics meant to socially signal anything other than autism?). The most useful section is for identifying beliefs are motivated by social signaling instead of truth: (1) abstract and non-specific (e.g., focused on politics more than policies; (2) no discussion of second-order impacts or tradeoffs; (3) conspicuousness; (4) overconfidence; (5) strong emotions (like defensiveness); and (6) reluctance to make a financial bet on outcomes.

(4) The Incredible Double Life of a Spyware Salesman Turned Spy

Communication Control Systems (CCS) was the world’s largest retailer of spy gadgets. Its owner, Ben Jamil, sold real spy equipment—neck ties with hidden cameras; phone scramblers; wire taps (before they become illegal); and more. After initially creating a business on wiretaps (legal when CCS started in the 1950s), he pivoted into espionage defense equipment. But the business operated in an export control gray zone. The FBI eventually arrested him in 1979, then turned him into an informant who used CCS as a federal sting operation to catch criminals. He sold out his clients to buy his freedom. It’s a good read for the espionage thriller anecdotes that sound straight out of James Bond, and for the irony of the spies being spied upon by the spy shop owner.

(5) Rightness is a Prison

“It’s actually really hard to be right about all but the most basic things. […] Given that perpetual wrongness is [our] natural condition, it’s funny that we try so hard to avoid acknowledging that we’re wrong, either to others or to ourselves. [Super-forecasters] aren’t strongly psychologically committed to being right about any particular fact or circumstance. [The] irony is that we have a tendency to put the least effort into questioning our most consequential beliefs — like what our priorities are — even though, logically, we should spend the most time trying to disconfirm them. The more you practice, the quicker you’ll develop the facility of holding your opinions lightly, and eventually you’ll do it by default. This sounds like a small thing, but it’s a major transformation.”

(6) Red Lobster’s Last Gasp

Red Lobster is the largest national seafood chain in the US, with 540 locations, a lobster tank in every front window, and cheddar biscuits that defined casual dining for a generation. The store originally expanded after being acquired by General Mills in 1970, went through a rotating cast of PE owners, landed in Chapter 11, and now is owned by its former creditors at Fortress. Red Lobster’s biggest issue is long-term lease agreements on unprofitable stores, entered during a sale-leaseback under previous PE owners, that blocks them from shedding unprofitable stores, even in bankruptcy. A younger CEO was brought in to revive the struggling chain, and the article covers his efforts with bringing back hush puppies, $20 unlimited shrimp, posting Tik Tok videos, etc. But the uphill battle gets harder with every year of deferred capex. Red Lobster is “cooked.”

(7) AI Startup Has Helped Reverse Thousands of Denied Insurance Claims

Claimable is a startup that helps patients to appeal their denied insurance claims. US healthcare companies deny 850 million insurance claims a year, and fewer than 1% of patients appeal their denials. The company is especially focused on the denial of claims related to pre-authorization, which are easier to appeal, and claims that 75% of its appeals are successful. “Claimable uses a large language model trained on insurance laws, legal precedents and medical literature to spit out an appeal letter. Claimable sends it to the insurer’s appeals department and, sometimes, to top executives, politicians and journalists in an effort to add more pressure. 4,000 people have used the site to file appeals so far, and Claimable [claims] 3 in 4 have gotten denials reversed.” Pretty good example of using AI to automate a byzantine process that millions face each year.

(8) A Very Big Fight Over A Very Small Language

Romansh is spoken by less than one percent of Switzerland’s population, split across five dialects that evolved in isolated mountain valleys. The Swiss government tried to unify them into a single standard version called Rumantsch Grischun, and it triggered a revolt over identity. Schools were forced to teach a language nobody spoke at home, teachers refused to use it, and communities complained about loss of identity. The Swiss government spends roughly $7 million per year to keep Romansh alive, which is surprisingly cheap. Language is core to identity and an underrated barrier to further integration in the EU as well. Languages are a key source of cultural separation.

(9) Did Craigslist Kill Newspapers? Legend Meets Reality.

Craigslist is one of the most underrated companies of the early internet. Launched as an email newsletter for friends in 1995, Craigslist soon became the dominant online classifieds website, replacing the once-lucrative print classifieds for local newspapers (and, alongside TV, is usually blamed for the decline of local newspapers). Craigslist is a good example of how the business model matters more than the operator—Craig Newmark and Jim Buckmaster are explicitly non-profit-maximizing (their former investor eBay actually sued them for rejecting the church of shareholder value). Despite rejecting profit maximization and ignoring competitors, they have achieved an estimated $1 billion (!) in revenue, with 80%+ profit margins, and continue to grow, despite ongoing competition from vertical marketplaces (like Airbnb) and horizontal classifieds (like Facebook Marketplace). Craigslist is a perfect example of Buffett’s ham sandwich test.

(10) How Big Data Created the Modern Dairy Cow

The modern dairy cow exists because farmers stopped trusting their eyes and started measuring butterfat. The Babcock test, praised for having “made dairymen more honest than the Bible ever had,” revealed which cows actually produced valuable milk (more butterfat content). In the early 1900s, US farmers organized cooperatives to measure the butterfat content of each cow’s milk, which turned breeding for higher yields into a science. These cooperatives became Dairy Herd Improvement Associations that pooled data across thousands of herds, eventually coordinated nationwide through the USDA. The rise of artificial insemination spread superior cow genetics nationwide, milk yields exploded, and the rest of history. Big win for dairy data cooperatives.

Miscellaneous

Good primer on private credit. Unstoppable growth at ByteDance. Why Uzbekistan is the land of Chevrolet cars. Gianluigi Aponte passed MSC to his children. What really matters at the end. Prayer is like meditation. A profile of Apple’s incoming CEO John Ternus. How Iran actually moves oil from Kharg Island. Danish Sound dues. The most popular grocery store brands in 1925. The unit economics of PopUp Bagels. History of MSA Safety. The reason so many “family influencers” are Mormons. Moral whiplash of Kanye West. The case against cynicism. The story of crypto exchange Hyperliquid. Investing in Cuban real estate. Traveling in Extremadura. Use of AI in legal research. Maybe you are just a conflicting parliament of sub-agents with diverging priorities.

(1) Some People Can’t See Mental Images. The Consequences Are Profound.

Not everyone has “mental imagery.” This condition, called aphantasia, results in a lack of visual memories—the inability to picture the face of someone, recall the imagery of past experiences, and so on. Mental visualization influence our emotions, so people with aphantasia have weaker emotions because they can’t visually relive past joys or traumas. On the other hand, extremely strong visual memories (called hyperphantasia) are an emotional amplifier. Aphantasia shows the deeper connection between mental visualization, memories, and emotions: “Many could remember very little about their own lives. Even for the events they did remember they could not muster the feeling of what those had been like. They knew some things had made them happy and others had made them sad, but that knowledge was factual, and it did not evoke any emotions in the present. We believe that our selves are made up largely of memories, and that the loss of those memories would be a self-ending catastrophe. But he knew now that there were also thousands of people just like him, who had work and marriages and ideas and thwarted desires and good days and bad days. All they lacked was a past.” They have the eternal sunshine of the spotless mind.

(2) How A Billionaire Owner Brought Turmoil And Trouble To Sotheby’s

The best parts of this article are the mechanics of Sotheby’s auction business. The auctioneer’s job is to dramatize the possibilities of a sale. For major consignors, Sotheby’s waives the seller’s fee and brings in third-party guarantors to backstop the price.. Private sales, where transactions are brokered outside the auction room, have become an increasingly important revenue line. The business model runs on expensive relationships and theatrical generosity, which is exactly where Drahi’s cost-cutting instincts collided with reality. Auction economics reward lavish spending on client relationships, so penny-pinching with a “relationship” business tends to be short-sighted. Unfortunately for Drahi, it sounds like cable-style cost-cutting doesn’t work with cultural assets.

(3) The Brand Age

The latest from Paul Graham: “Brand is what is left when the substantive differences between products disappear. But making substantive differences between products disappear is what technology naturally tends to do.” He covers the story of the Swiss watch industry, which differentiated itself with thinness and the accuracy of mechanical movements until the quartz-based watches became cheaper, thinner, and more accurate. After the “quartz crisis”, Swiss watches shifted into brand marketing, distinguishing themselves as luxury items and selling scarcity as the core value proposition. Watches turned into men’s jewelry, male equivalents of Cartier bracelets. Graham is critical of brands: they shift the focus away from technical quality, cater to the base psychology of customers (wanting to feel superior with status symbols), and they’re “a boring problem.” Not to relate everything to AI, but… if everyone can build the same software, does software become brand-focused? Or is software already brand-focused? How much hate for Microsoft Teams is due to the connotations?

(4) Against Sloppypasta: Don’t Paste Raw LLM Content at People

Until LLMs, writing long documents was “proof of work.” The process of writing, both the time required and the reflection that it produces, was implicit evidence that the author had thought deeply about the problem. LLMs have created the novel issue of “sloppypasta,” where anyone can instantly generate a mass of thoughtful-sounding text in response to anything. I think this will become a real issue in companies… AI writing long memos and emails, people asking AI to summarize them, and everyone outsourcing their thinking to LLMs. The article makes an excellent argument that sloppypasta is bad for both the creators and recipients: creators lose credibility and understanding, while recipients take an asymmetric burden of reading, vetting, and acting on the information. Lots of great quotes in the article, but the best argument for writing yourself (and only using AI for research/assistance) is to reduce personal brain rot: “Use AI to accelerate your work or improve what you send. Don’t use it to replace thinking about what you’re sending.”

(5) Cuba’s Broken Economy Leaves It At Donald Trump’s Mercy

Good profile of Cuba’s economic crisis. The embargo is strangling Cuba’s economy: GDP shrank 7% last year, population has declined over 20% since 2020, and foreign reserves are exhausted. Their main problem is a lack of fuel: the Russian embargo and Venezuelan coup eliminated their access to oil (which powers the electricity grid), resulting in constant blackouts and driving most factories to a halt. The additional pressure on exports, especially tourism and doctors sent to other countries, have cut off remaining sources of FX. Trump and Rubio are not calling for regime change just yet—demands currently focus on liberalizing the private sector (and probably putting Trump’s name on the Havana Libre). What do the locals think? “Not everyone wants a McDonald’s on the Malecón, Havana’s seafront promenade, but they do want an economy that functions.” Capitalism’s final victory will be the Malecón McDonald’s.

(6) Blank Street: Coffee, Counter-Positioning, and Collateral Damage

Blank Street is a venture-backed coffee chain that launched in NYC in 2020. Their model is classic disruption (in the Clayton Christensen sense): counter-positioning against the higher-priced and large footprint Starbucks locations with smaller stores focused on a limited selection of cheaper, fast turnaround, grab-and-go coffees. Blank Street is popular with investors for their unit economics, but their defensibility is an open question: “It would be painful for Starbucks, with its existing footprint and cost base, to compete with Blank Street on price, [but] there are not many barriers to entry preventing another company from replicating its model. [They are] a SoftBank Vision Fund 3 away from having direct competitors. Attractive unit economics are a necessary but insufficient condition for building a sustainable business. Electric scooter rental businesses like Bird and Lime had positive unit economics.” I’d probably bet on Blank Street’s model.

(7) Review of Demons by Fyodor Dostoevsky

For those just getting into Russian literature… Dostoevsky’s timeless mirror of society: “The ideas don’t matter, because at the end of the day they’re pretexts for desires—the desire to dominate, the desire to obliterate the world, the desire to obliterate the self, the desire to negate. […The] desire for status came first and wrapped itself in liberal politics in order to reproduce and advance itself […their] radical politics serve only as a mechanism of self-justification and a lever to pull. It’s not a novel about people, and it’s also not a novel about ideas. It’s actually a novel about desires, motives, urges, and the ways in which we construct stories to make sense of them. But where do desires come from?”

(8) A Love Letter to Novels

Fiction and feelings: “When read at the right time, a novel really can change you, the way the kindness of a stranger might alter the course of your day or the words of a mentor might shift your perception of a problem you thought was impossible to solve. Many people will pick up a book during a dark period in their lives and emerge from their reading experience transformed. For example, the emotions and events depicted in a novel [can] help you realize that you’re not alone, as famously noted by James Baldwin: You think that your pain and your heartbreak are unprecedented in the history of the world, but then you read [a book]. Dostoevsky and Dickens taught me that the things that tormented me most were the very things that connected me with all the people who were alive, or who ever had been alive. […] Timing is crucial. It doesn’t matter how well-written or revered a book might be; you have to read it at the right time for it to truly change you. […] Having a personal library with unread books […] is partly what allows someone to choose the right books at the right time. […] The stories are how we make sense of our lives.”

(9) Get Married Young

“The happiest young women (22-35) today are not footloose and fancy free, they are married moms. The ones least likely to be happy are single and childless. Data from the General Social Survey indicates that 41% of young married moms (22-35) are “very happy” compared to just 14% of their female peers who are single and childless. […] What about young men? […] Young married men (22-35) who are married with children are almost three times as likely to be “very happy” with their lives compared to their peers who are single and childless. 14% of young men who are single and childless are “very happy” compared to 37% of their peers who are married fathers. […] These data suggest young men and women who take their cues from pop culture—shows like “Friends” that celebrate single life in New York City—may be in for a rude surprise.”

(10) Why Your Next Flight Is Likelier to Hit Turbulence

Clear-air turbulence over the North Atlantic has increased 55% since 1979. We don’t know why, we can’t see it on radar, and we can’t predict where it will come next. The article has raises lots of alarms about turbulence, but the combined statistics made me less concerned about the risk: only 1 in 3,000 US flights encounter severe turbulence, nearly all of the 40 annual US deaths from turbulence occur on small private aircraft, and 80% of the remainder are flight attendants (often because they aren’t wearing seatbelts). The odds of dying on a US flight are only 1 in 46 million, and turbulence risk continues to decline because of engineering—sensors measure wind force in real time and auto-adjust the flaps/spoilers/ailerons to counteract turbulence, aggregated real-time data from flights can flag turbulence zones to other planes, and more. But moderate turbulence is still unsettling… so if you’re flying over Colorado or Chile (or any mountain ranges), buckle up!

Miscellaneous

Foreign policy in the Marco Rubio era. Buying a Romanian election with TikTok gifts. Explaining the global success of Pokemon. Reflections on the Reynolds Pamphlet. The expansion of Heathrow Airport. Solar panels in Cuba. How Swig’s dirty soda became a national obsession. AI is accelerating GMO fruits. Cost estimates for datacenters in space. AI could be reducing energy costs by increasing non-peak utilization. Tricolor accounting fraud. Inside the attempted coup in Benin. AI breakthrough in theoretical physics. AI assistance in building military target banks. New prime minister of Nepal. Declining cost of entrepreneurship. The expanding NVIDIA empire. The parsley test.

(1) Something Big is Happening

Most people don’t realize how fast AI is improving: “The models today are unrecognizable from what existed even six months ago. [Let] me make the pace of improvement concrete: [in] 2022, AI couldn’t do basic arithmetic reliably. It would confidently tell you that 7 × 8 = 54. By 2023, it could pass the bar exam. By 2024, it could write working software and explain graduate-level science. By late 2025, some of the best engineers in the world said they had handed over most of their coding work to AI. On February 5th, 2026, new models arrived that made everything before them feel like a different era. If you haven’t tried AI in the last few months, what exists today would be unrecognizable to you. […] Nothing that can be done on a computer is safe in the medium term. If your job happens on a screen (if the core of what you do is reading, writing, analyzing, deciding, communicating through a keyboard) then AI is coming for significant parts of it. The timeline isn’t “someday.” It’s already started.” Based on my own experience, I’d agree.

(2) 10 Years Building Vertical Software: My Perspective on the Selloff

Public market software is selling off due to the fear that AI will eliminate their moats (creating competing software is easier, switching costs are lower because unstructured data is easier to re-structure, etc.). This article covers the various moats of vertical market software (VMS) and analyzes what changes, if anything, with AI. The moats that are lost: muscle memory; business logic in custom workflows (“LLMs turn all of this into a markdown file”); engineering talent (is this really a moat?); and bundling. The moats that remain: proprietary data (a consensus view); regulatory lock-in because “HIPAA doesn’t care about LLMs” (I think this is underrated even in unregulated markets… good luck getting F500 adoption of vibe-coded Salesforce); transaction-level embedding (e.g., payment processors); and network effects. Regulatory lock-in is the most durable moat remaining: “If your moat is regulatory certification, compliance infrastructure, and deep integration within mission-critical workflows, LLMs are barely relevant to your competitive position. […] Healthcare EHR systems with HIPAA compliance and FDA validation. Life sciences platforms with regulatory lock-in. Financial compliance and reporting infrastructure.” Everyone hates regulation until it protects their special interest!

(3) 18 Brumaire Year VIII (November 9, 1799)

Napoleon’s seizure of power in France was a messy affair. This article covers the play-by-play of 18 Brumaire (the revolutionary calendar date for his coup). Napoleon had coordinated with Sieyès and his brother Lucien (the president of the Council of Five Hundred) to fabricate a Jacobin terror plot, justifying the government’s relocation to suburban St. Cloud, where military control was easier than in Paris. When legislators realized that a coup was in progress, they physically attacked Napoleon on the floor and voted to have him executed. Lucien saved the coup at the last moment by sending in Napoleon’s grenadiers to “restore order.” By midnight, the deputies dissolved the Directory and named Napoleon as First Consul. Historians often see Napoleon’s rise as inevitable, but in reality, the outcome was a coin flip between execution and empire.

(4) Flying Under the Radar with Flightradar24

Flightradar24 is a popular app for tracking flights. What began in 2006 as a hobbyist website for plane enthusiasts has grown into a huge, profitable, privately-owned mobile app (recently valued at $1.5 billion). Flightradar24 has a unique data advantage with over 40,000 ADS-B receivers, mostly owned by hobbyists, that share real-time data with them, and which is monetized both as a consumer-facing app and a B2B data-stream. The article has a lot more info about the nuances of their business model—creating content for SEO, using their newsletter to get new transponder volunteers, partnering with easyJet, the impact of Apple enabling subscriptions on iOS (and the tradeoffs vs ad-supported monetization), and more. Not bad for a company that is “just an app.”

(5) The Battle for Minneapolis

Reflections on the ICE showdown in Minnesota: “When you grow up in Minneapolis, then you leave it, two truths become apparent: winters everywhere else seem fake, and no other city’s politics seem as doggedly progressive, not even New York’s.” They’re fighting a very different kind of ice this winter… “Thousands of Twin Cities residents have started observing ICE, from cars or on foot, since the surge began. Others volunteer to drive workers to their jobs, monitor school drop-offs and pickups, or deliver groceries to people too scared to leave their homes.” And some heartwarming, Minnesota nice sentiment: “I don’t care about your documentation status. You’re my neighbor. […] The people they rounded up, they can drive in the snow, they have good coats: they’re Minnesotans.” Quite a messy month in the homeland.

(6) Can Jollibee Beat American Fast Food at Its Own Game?

The global fast food industry is almost entirely American: McDonald’s, KFC, Subway, Starbucks, etc. Remarkably few chains flow the other direction. Jollibee from the Philippines is a rare non-US contender. The menu (sweet spaghetti, peach-mango pies, garlic-seasoned fried chicken) is catered towards Filipino tastes, and its 80 US locations are in cities with Filipino diaspora. While they have an ongoing challenge of adapting to US consumer tastes, their wider portfolio is growing. Outside of the main brand, Jollibee’s parent company operates over 10,000 stores globally with 19 brands, having acquired Smashburger and Coffee Bean & Tea Leaf. But it’s a long road ahead for mainstream awareness: even 500 locations is less than McDonald’s in Ohio alone.

(7) The Innkeeper of Bruges

Bruges was among the most important cities in medieval Europe. The first “modern” financial exchange (focused on FX, commodities, and futures) emerged organically in Bruges among a group of Hanseatic merchants at an inn run by the Van Der Beurse family (whose name is the origin of “bourse” in many languages). The pieces of a modern exchange emerged gradually. First, the inn itself acted as the trusted intermediary guaranteeing payments between parties (centralized clearing and settlement). Second, commercial courts called the Vierschaar were created to resolve disputes (binding arbitration). Third, to handle counterparty risk, a formal “protest system” for registering debts and defaults with notaries functioned as an early credit bureau, creating a shared blacklist that could follow the merchants’ reputations across Europe. Modern finance started in Bruges and spread worldwide.

(8) How The Diamond Industry Lost Its Sparkle

Diamonds aren’t forever after all. Chinese factories can mass-produce 3-carat lab-grown diamonds in 7 days using machines that cost around $280,000 each. A single factory in Zhengzhou can produce over 100,000 carats per month, and about 70% of the world’s lab-grown diamonds for jewelry now originate from Chinese factories (mostly in Henan). The pricing pressure is intense: “a three-carat lab-grown stone now sells for just 7 per cent of the price of a mined equivalent.” In fact, prices are falling so fast that China put a domestic price floor on diamonds at $15/carat, and the cost to polish and transport them to Antwerp is now more than the cost of production: “It costs $56 to polish one carat in China, more than it costs to produce the stone.” The pricing for lab-grown diamonds is at the price floor of $15/carat. Still higher than the resale value.

(9) Is “Anna Karenina” A Love Story?

“Deciding what to think of Anna is one of the central challenges with “Anna Karenina.” […] Anna does bad things [and] she underestimates just how bad the consequences of those things will be. […] Tolstoy doesn’t know exactly how to think about Anna’s role in her own downfall. He believes we make choices, and that our sense of free will is based on something real. But he also has a deep respect for the complexity and power of our circumstances, and he considers our personalities and psychologies to be “circumstances” too. […] She may have made an unwise choice, giving into desires she should have resisted because she underestimated how unyielding the world would be. We will never know what happened, exactly, just as Anna could not know. That is one of the dreadful lessons of Anna’s story: she could not distinguish between what she was choosing to do and what she was driven to do.” Unfortunately, her clarity came too late to avoid the consequences: “The light […] flared up more brightly than ever before, lighting up for her all that had been in darkness, flickered, began to grow dim, and was extinguished forever.” At least Levin had a happy ending.

(10) Dante’s Ninth Circle of Hell

Dante’s Divine Comedy is a medieval poem about sin and salvation. Its most famous section, Inferno, created the modern depiction of hell as a “fiery inferno” with nine descending circles of sins. But Dante’s lowest circle of hell was ice cold, alluding to the cold and calculating nature of the most unforgivable sin: “In the ninth circle, the innermost pit of hell […] entombed in ice, are the traitors, the people whose actions have shattered the foundations of belief and loyalty. […] The ice, impenetrable and unyielding, mirrors the hardened hearts of the traitors, incapable of regret or redemption. In contrast to the fiery images of hell in the earlier circles, which [depicted] a passionate, albeit misguided, pursuit of earthly pleasures, the icy grip of the ninth circle speaks to a profound lack of feeling and corruption. The act of betrayal, as is depicted in the ninth circle, is deeply dehumanizing. Both the betrayer and the betrayed are diminished by the act. […] It’s the place where hope dies and solely despair remains. Betrayal is the final and worst sin.”

(0) Miscellaneous

Paramilitary parallels with DHS/ICE expansion. Unprecedented shakeup in China’s military leadership. Rising importance of ship tracking software. Declining age gaps in relationships. Latest updates on the Congo. Charitable donor advised funds. Ideas for investing in Guyana. The future of spreadsheets with Claude for Excel. The rise of Moltbook. The main character is Elon Musk. Yet another Olympics cheating scandal. Relatable name and relatable content. What secure attachment looks like. Honesty in relationships. FT recommendations for NYC date spots. Very good pun. Call center scams with multi-district litigation. ICE’s clever data collection methods. Financial derivatives and market indices for AI GPUs. Quirky defeasance clause in EA bonds. How to buy a law firm. How to fix your life in one day.

(1) CSC Generation is a Billion-Dollar Home Brand. Why Haven’t You Heard of It?

CSC Generation is a rollup of distressed home retailers. Their strategy (according to this article) is focused on making better use of customer data for marketing, shifting retailers to online sales, and generally improving performance of fallen angel brands. The article covers the story of two failed takeover attempts of public companies, and the broader challenge of the “change-averse old boy’s club culture […] that was a real obstacle to deals getting done.” CSC seems like they’re doing just fine, with over $1 billion in revenue in less than a decade. “I asked [the CEO] what his long-term goal was for CSC, and if he had a model for what he was trying to build. He did, and maybe unsurprisingly it was a company I’d never heard of: Constellation Software.”

(2) A Theory of Haggling

Haggling is a means of price discrimination. Some people are willing to trade some unit of time for some amount of cash savings. This is why haggling is more common in poor countries. The back and forth is both sides trying to estimate the surplus from the transaction and capturing more of that. “You’ll expect more haggling in a context where information is scarce, and one where the return on time spent haggling is high for both sides.” This also shows up indirectly in richer countries: we’re haggling every time we ask a Chipotle employee to add more chicken to our bowls. In B2B SaaS procurement, volume tiers, multiple meetings, and questions about use cases are sizing up a customer’s willingness to pay. Markets are automated haggling at scale.

(3) Making 13x on a Preferred Stock

In 2007, Goldman took private the hotel chain Equity Inns for $2 billion. Equity Inns had another $145 million of preferred stock trading in public markets, which Goldman didn’t acquire. Instead, they paused the preferred dividends, then stopped financial reporting with the SEC (as a private company), to tank the shares and buy them back cheaply. But Goldman forgot a technicality with SEC reporting requirements: any company with more than 300 “registered” shareholders (shares held directly in their name, not a broker’s “street name”) is required to make public filings of their financials. One shareholder bought many of the preferred shares at 8% of par, divided them into over 300 trust accounts, then informed Goldman of its legal filing requirements. That, plus some shareholder lawsuits over contractual breaches in the treatment of the preferred shareholders, resulted in a settlement buyout of the shares that netted a 13x return over this 6 year saga.

(4) Book Review: Family Unfriendly, by Timothy P. Carney

Why have parenting burdens grown, even as the average family size has declined? This book review traces the shift from quantity to “quality” in parenting: parents became chauffeurs, kids became résumé-building projects, and childhood transformed into a scheduled march towards college admission. Parenting has turned into a status competition based on visible investments (organized sports, academic camps, structured activities), with old-fashioned and unstructured parenting being seen as low status. It seems like the issue is particularly bad in the US, where a culture of extracurriculars, combined with car commuting between them, makes having children much more time-consuming than it used to be.

(5) How Did Russia Get So Big?

Russia was built with the commercial logic of a trading empire. This article shows how Moscow repeated the same playbook over centuries: secure the resource, secure the route, secure the port. First they eliminated their main competitor, Novgorod, to dominate northern trade routes from Arkhangelsk. The next step, conquering Kazan in 1552, gave Moscow control of the Volga River, the main trade route for Siberian fur. Traders followed rivers east in the search for sable furs, soldiers came with them, and forts would spring up along the way. This process repeated across the continent until Russia reached the Pacific. When they needed better ocean access, they took Vladivostok from China in 1860. The Trans-Siberian Railway later tied it all together.

(6) History of Standard Oil: The Cleveland Massacre

Standard Oil’s dominance started with the rapid consolidation of all of its competing refiners in Cleveland over a one-month period in early 1872. Standard Oil went from 10% of the market to 90% in one month. How did they acquire their 25 Cleveland competitors in one month? It wasn’t Rockefeller’s charm… it was an offer they couldn’t refuse. The creation of the South Improvement Company, a secret pact with the railroads over freight rebates, gave Standard Oil an immediate, crushing advantage on transportation costs that scared his competitors into either selling today, or going bankrupt tomorrow. In the end, the SIC was shut down by the Ohio legislature before it even started operations, but the threat was enough to consolidate the market at fire-sale prices.

(7) Why Millennials Love Prenups

Prenups aren’t just for billionaires. They’re especially popular with younger generations: 47% of Gen Z wants prenups. The growing popularity reflects, in part, the normalization of divorce. But just as important, courts are more willing to uphold prenups than in the past: the combination of rising divorce frequency and increasing financial independence for women has made judges less likely to toss out prenup contracts on “equitable” (fairness) grounds. The author debates whether she wants a prenup, but financials aside, she concludes: “The conversation alone is a personality test. Are you about to marry a person who wants to be reimbursed for the wallpaper you put in the nursery? [Or] does this person look at you and think: I want to give [you] everything?” Also, a potential quote of the year: “When Taylor Swift and Travis Kelce announced their engagement, one X user joked, “that prenup is about to be longer than any book travis kelce has ever read.”

(8) The Rise and Fall of the “Bathtub Trust” - A Deep Dive into the Sanitary Ware Cartel

The Bathtub Trust was the nickname for a short-lived Gilded Age industrial cartel of enameled iron bathroom fixtures (bathtubs, sinks, toilets, etc.). While less well-known than the great trusts like Standard Oil or US Steel, the Bathtub Trust is notable for a few reasons: (1) the invention of the Arrott Process for mechanized enameling illustrates how inventions that convert variable labor into fixed capital, with economies of scale, drive industrial consolidation; (2) the device of “neutral” third-party patent licensors was a novel workaround for antitrust law (although it didn’t work!), coordinating the behaviors of competitors using licensing terms instead of explicitly anti-competitive contracts; and (3) the use of “royalty and rebate” terms in the patent licenses, to be administered by the neutral intermediary, resulted in a self-enforcing contract that overcame the compliance issues of past trusts. Despite their creativity, the organizers forgot the government doesn’t appreciate creative workarounds of its policy priorities… “it looked and smelled like an old-fashioned price-fixing trust, and the Taft administration was in no mood to tolerate trusts.”

(9) Excel Turns 40: Why We Can’t Quit Microsoft’s Spreadsheet App

“Is Excel secretly the greatest piece of software ever written?” It has been 40 years since the debut (on Mac!) of Microsoft’s version of VisiCalc and Lotus 123. Spreadsheets were the original killer app for the personal computer: ”It transformed the Apple II almost instantly from a niche tool for gamers and teachers into the first mass-market PC.” While the narrative today is that Excel won via bundling with Microsoft Office, an underrated part of its early popularity was its calculation speed: “Excel first distinguished itself with speed[.] Intelligent recalculation […] used an algorithm to update only the cells affected by a given change.” There are 500 million paying users for Excel today, and its dominance is unchanged—the two main serious threats it has faced, cloud-based collaboration in Google Sheets and AI vibe-modeling, have had no visible impact on growth. The Office 365 bundle is still under $200 per year, so pricing has a long runway ahead.

(10) How WhatsApp Took Over the Global Conversation

“WhatsApp delivers a hundred billion messages a day, around the same number as there are stars in the Milky Way. The platform has more than three billion active users each month—pretty much the same number as Facebook, more than YouTube, and roughly twice as many as Apple’s iMessage. If you exclude China, where WhatsApp is banned, the app serves approximately half the human population older than fourteen. Three-quarters of WhatsApp users are on Android devices, with the rest on iPhones—which more or less reflects the global smartphone market. One reason that WhatsApp has traditionally lagged within the U.S. is because of the unusual dominance of iPhones. But that is changing. Last year, the U.S. was one of the app’s fastest-growing markets, with the number of monthly users exceeding 100 million for the first time.”

Miscellaneous

100x return investing in paint rollers. Thoughts on healthcare costs. Introducing agent engineering. Criteria for acquiring mobile apps. Good life advice. Capital markets in Uzbekistan. Reflections after 8 years as CTO. Issues at Novo Nordisk. Bankrupt rollup of port-o-potties. A history of NYC cemeteries. Oil bargain with Venezuela. Dutch ownership of Duivelsberg. Bull case for Plaid in payments. The future of consumer health apps. Profile of the founder of UNIQLO. Clever way to increase arable farmland. The origins of Coca-Cola bottling franchises.

(1) The Fatima Sun Miracle: Much More Than You Wanted To Know

The “Miracle of the Sun” at Fátima is one of the most famous and inexplicable religious miracles of all time. In 1917, at the predicted time and location (based on three shepherd children’s visions of the Virgin Mary), over 50,000 people in Fátima (Portugal) witnessed the sun “dance” and exhibit other strange movements for 10 minutes. The miracle is unusual because (1) it was predicted in advance, (2) it was witnessed by thousands, including many religious skeptics, (3) there was no obvious incentive for fabrication, (4) there is still no good explanation for what they witnessed, (5) it occurred recently (in 1917) and was well-documented by contemporary sources, and (6) it has survived all attempts at debunking (including this article). “The Miracle has a reputation as the final boss of paranormal experiences, the ultimate challenge for would-be religious skeptics and debunkers.” Can you solve the 100+ year old mystery of what they saw in the Cova da Iria?

(2) Where are the Matriarchies?

Nearly all human and primate societies are patriarchal. Why? The consensus among sociologists is role specialization—males specialize in warfare and females specialize in caregiving. But why do gender specializations consistently produce male-dominated hierarchies? It is much deeper than physical strength. Dictators don’t gain power by physically overpowering their enemies in hand-to-hand combat, they gain power with superior coordination of resources, and the most important resource to manage is other people (armies, logistics, political parties, etc.). So the theorized causal chain is strength → warfare → management → control.

(3) Welcome to the Era of Experience

This AI paper is a few months old (ancient history for AI papers), but it makes a compelling case for the fact that the next phase of AI development will require agents that can interact with, and learn from, the real world as a means of collecting more data. The core limitation that constrains further AI development will soon become a lack of novel real-world data: “valuable new insights [like] new theorems, technologies, or scientific breakthroughs, lie beyond the current boundaries of human understanding and cannot be captured by existing human data.” New data will come from “allowing agents to learn continually from their own experience.” This paper covers some ways that agents will be designed to do this more effectively: (1) data streams that carry over from one conversation or experience to the next, so that the agents can continually learn and adapt over time; (2) action and observation in digital environments (MCP is already moving in this direction, although the streaming-focused learning is nascent); (3) integrations with more diverse “reward functions” like health metrics, profits, likes, energy consumption, etc.; and (4) more computationally efficient mechanisms for “thinking” than human language.

(4) Nobody Has A Personality Anymore

The downsides of therapy-speak, self-reflection, and psychoanalysis: “Therapy-speak has taken over our language. […] There are no feelings at all anymore; only [dys-regulated] nervous systems. […] This is the “healthy” way to think that previous generations were so cruelly deprived of. […] I find it strange that we think this is freeing. This brutal knowing. That this self-surveillance is the liberated way to live. [We] underestimate it, this miserable business of understanding ourselves. […] Explain anything long enough and you will find a pathology. We’re thinking about ourselves enough. After a lifetime spent trying to make sense of every experience, [this] generation might realize that the only problem they had, all along, was being human.”

(5) Book Review: Alexander to Actium

What happened to the Greek world in the 300 years between the conquests of Alexander and the fall of Ptolemaic Egypt to Octavian? This book review covers some of the history: “Golden age Greece was a Greece [of] individual city-states constantly in competition: Athens, Sparta, Thebes, Corinth, Argos, and so on. […] It was this quarrelsome, decentralized patchwork that produced the Greek miracle. […] With Alexander’s godlike achievements, though, that old Greek world crumbled. Suddenly, Greek civilization was dominated by massive kingdoms, ruled by kings whose only claim to legitimacy was that they or their fathers had won their lands at the tip of a spear. […] The kingdoms were essentially for-profit estates run for the benefit of mighty god-kings. Alexandria, the crown jewel of the Hellenistic world, home to the Great Lighthouse and Great Library, was effectively a giant consumer good for the Ptolemaic royal family, where all the surplus of Egypt’s countryside had to be shipped, on pain of death.”

(6) The Slow Death of Epic Systems

The headline is overblown, but indirectly explains how Epic could become healthcare’s IBM (in a bad way). Epic Health is the dominant operating system in EHR for the healthcare industry, but it exists more for maximizing billing than patient outcomes. As a result, they don’t use (or save) the torrent of data generated by healthcare (patient vitals, time-series data, etc). Because of AI, this “data exhaust” is now extremely valuable; it is arguably the last untapped frontier for real-world AI training data. The companies that excel at collecting and learning from that data to improve patient outcomes will supplant Epic as the critical layer of healthcare software. “Clinicians will live inside those tools, not inside an interface built for billing. […] There is precedent for this in every major industry: the core orchestration/data layer eventually recedes into the background while workflow and data intelligence move up the stack [while] the EHR becomes middleware.”

(7) Accommodation Nation: Elite Colleges Have an “Extra-Time-On-Tests” Problem

Students at elite schools abuse lax policies for “learning disability” accommodations. The main benefit is extra time for exams. The issue is most prevalent at elite colleges: “At Brown and Harvard, more than 20% of undergrads are registered as disabled. At Amherst, the figure is 34%. […] 38% of Stanford undergrads are registered as having a disability.” One enabling factor is that the definition of disability has widened to include basically anything: “Students with disabilities [aren’t] kids in wheelchairs, [they’re] rich kids getting extra time on tests.” The ADA hates this argument… “Having ADHD or anxiety might make it difficult to focus. But focusing is a skill that the educational system is designed to test.”

(8) The Cocaine Kingpin Living Large in Dubai

Daniel Kinahan, an Irish drug dealer, runs a billion-dollar cocaine empire from the UAE. This story covers how he entered the trade, scaling his operations through a money laundering partnership with Hezbollah, reinvested the proceeds into a boxing empire, and avoided arrest along the way. Dubai sounds like a comfortable place for criminals… “The family has made extensive use of the Dubai hawala network—an untraceable money-transfer system based on trust among individual brokers. […] Daniel and his brother have also bought various expensive properties in the U.A.E., and have incorporated several companies in Dubai. Financial oversight by Emirati authorities was evidently cursory: the brothers were deemed “low risk.”

(9) The Hermès Heist: How An Heir to the Luxury Dynasty was Swindled Out of $15B of Shares

Nicolas Puech inherited a 6% stake in Hermès, currently worth more than €13 billion. His stake was quietly liquidated and stolen over the course of two decades. His trusted Swiss banker, Eric Freymond put Puech’s bearer shares (!) in Swiss bank accounts and secretly sold millions of shares to LVMH for an attempted stealth takeover of Hermès a decade ago, while the remainder was sold over time and the proceeds diverted into accounts controlled by Freymond. Puech alleges fraud, is apparently penniless, and has sued Arnault, LVMH, and others. Freymond took the truth to the grave, and the location of Puech’s stock—impossible to track due to being bear shares—appears to be lost forever.

(10) What Actually Makes You Senior

“There is one core skill that separates senior engineers from everyone else: reducing ambiguity. […] The moment you hand them something fuzzy [is] when you see the difference. When senior engineers do it well, it looks easy. Like nothing was even done. The project just goes smoothly. Fewer surprises, production fires, or emergency meetings. But what actually happened was that someone did a lot of invisible work upfront. […] If you’re wondering whether you’re operating at a senior level, here’s a test: what happens when someone hands you something abstract/fuzzy/complex? Do you wait for someone else to clarify it for you? Do you start coding immediately and hope for the best? Or do you spend time up front making it concrete enough that you and your team can actually execute with confidence?”

(0) Miscellaneous

Winners of the robotaxi race. SaaS concept of Max MRR. Rage bait for the Pura Vida fans. Patrick Drahi takes off the gloves with creditors. Ode to the heart. The reasons that hangovers get worse with age. Economics of the mobile airport lounge. The pop fiction of Oliver Sacks. The history of Domino’s Pizza. Lessons from building Manus. Update on North Korea. The family of Florida sugar barons. A detailed history of the Delaware Chancery Court. Investing in niche sports. Medieval Erfurt latrine disaster.

(1) Data and Defensibility

The classic half-baked software business model is “data → ?? → moat!” Data is neither necessary nor sufficient for a moat (a barrier to competition), but some data moats are real. The two broad groups are (1) data control and (2) data loops. Data control can be unique/proprietary (Nielsen); fragmented (CoStar); movement/networks (Visa); usage (Salesforce); IP/patents (S&P); and regulatory (FICO). Data loops can come from UGC (Facebook); SEO (Booking); gravity/stickiness (Toast); value exchange (Verisk); and trust/standards (Moody’s). There are more. But not all data moats are equal: the best are usage and trust, and the worst (which the author dismisses entirely) are learning loops: “Product optimization loop [moats]: lol, no. A/B testing is not even a shallow ditch.”

(2) The Bitter Lesson is Misunderstood

The “bitter lesson” refers to the thesis in AI scaling that model architecture doesn’t matter as much as throwing more compute and data at the problem. The massive AI capex boom in the past couple years is based on the first half of this lesson (scaling compute), but “the god of Compute we worship is actually yoked to an even greater one — the god of Data.” But getting new data may become prohibitively expensive: “we have basically ingested the entire Internet, and there is no second Internet. For language models, there’s no obvious way to scale the data.” Two possible workarounds: (1) architects who design clever ways to squeeze more intelligence out of each existing unit of data, and (2) alchemists who can conjure up new data. But data will soon be the hard constraint: “The AI companies that survive 2026 will be the ones who figured out data scarcity in 2025.” I would put it like this… the Copernican Revolution didn’t happen by thinking harder.

(3) Building Bending Spoons

Another recent interview with Luca Ferrari, the CEO of Italian app acquirer Bending Spoons. I was most surprised how, even more than acquiring companies, he focuses on acquiring talent: “for a long time on Slack, [his] label was recruiter.” The interview has a lot more on the hiring, compensation, and culture that seems to be his central focus. Luca also throws the gauntlet to would-be competitors: “[Even] trying to build Bending Spoons—if you understand what you’re doing, which is a prerequisite to even have a chance—[is] dauntingly painful. Many years from the ground up, cultivating the little garden. […] There’s no shortcut. We’re just not a model that, when people see it, they want to replicate. A lot of people have known about [us] for years. I’ve talked to investors, and I’ve seen nobody try, because they understand it’s just too painful.” :)

(4) Berkshire Hathaway: Letter from Warren Buffett (November 2025)

After more than 60 years, this could be Buffett’s last letter to shareholders. He’s 95 years old and reflecting on his long and lucky life: “Father Time […] finds me more interesting as I age. And he is undefeated; for him, everyone ends up on his score card as “wins.” [To] my surprise, I generally feel good. Though I move slowly and read with increasing difficulty, I am at the office five days a week where I work with wonderful people.” It’s a touching letter, filled with nostalgia over the many great business leaders and friends that he has worked with over the years, the charitable pursuits of his children, and his love of America (especially the Midwest!): “Remember to thank America for maximizing your opportunities.” The end of an era, the daedalic hall has fallen.

(5) “The Largest Funder of Al-Shabaab is the Minnesota Taxpayer”

Brutal headline. Minnesota is currently awash in welfare fraud with housing (HSS), school meals for children (Feeding Our Future), Medicaid-funded autism programs, and more. Much of the fraud is from groups in Minnesota’s Somali community (with evidence that at least some of that money has flowed back to the terrorist group Al-Shabaab). Of course, this has naturally turned into an issue of being “racist versus not racist,” but immigration politics aside, it does seem like a lot of welfare fraud is going on: “Autism claims to Medicaid in Minnesota skyrocketed from $3M in 2018 [to] $399M in 2023.” The Donald also weighed in… going to be a rough election cycle for the DFL.

(6) How the Custom Ringtone Industry Paved the Way for the App Store—Then Vanished

Remember ringtones? They used to be huge: “In 2002, 30% of all SMS traffic were requests for downloadable ringtones.” But the billion-dollar industry was silenced overnight: “From the 2007 peak of $1.1 billion in global ringtone sales, the industry shrunk by 97% over the [next] decade.” Some explanations for what happened: (1) the shift to texting; (2) fading novelty (caused by the Crazy Frog ringtones?); and (3) changing social norms with silenced phones (at least outside of the NYC subway). My own pet theory is more obscure: Apple made it difficult to install custom ringtones on iPhones. It’s much easier on Android, and Android is dominant with ringtones.

(7) Inside Cursor

Inside the corporate culture of AI startup Cursor: “My best evidence that the prize is the mission is that during my fall at Cursor, I overheard zero chatter from employees about getting rich. At Stripe and Figma (and most other startups), this was a favorite lunch table topic among the first few hundred employees at a deca-corn. Yet at Cursor, as the valuation goes up and up, I haven’t heard a peep about the second homes people will buy, the great-great-grandchildren that will be put through college, or the time they’ll take off traversing the world. If people have dollar signs in their eyes, they’re not talking about it much. And I think it’s because the thing most of them would do if they could retire tomorrow would be whatever they’re doing now at Cursor.”

(8) Ford and the Birth of the Model T

The early success of Ford was built on more than just the assembly line: (1) the use of very tight tolerance for machined parts (1/64 of an inch); (2) interchangeable parts; (3) casting the engine block as a single piece; (4) shipping cars as parts to assembly plants to reduce shipping costs; (5) removing color/style options to increase standardization (the Model T originally did have some options other than black); (6) creating special purpose machining tools for each part of the line; and (7) only a decade later, in 1913, the creation of the worker-level assembly line itself. Economies of scale led to lower prices, which led to more volume, which led to more economies of scale with process improvements: “By 1913, comparable cars to the Model T cost nearly twice as much as the Model T did. And the price continued to fall. By 1916, the cost had dropped to just $360—a two-thirds reduction in just six years.”

(9) PopUp Is Wall Street’s Bet a National Chain Can Actually Make Good Bagels

Seems like VC funds are getting into brick-and-mortar retail. Blank Street, 7th Street Burger… and now PopUp Bagels. PopUp Bagels is a bagel chain with 16 locations, mostly in NYC, that launched during COVID. Unlike mass-market models with burgers, chicken, pizza, and cookies, bagels are still mostly hyper-local (with the exception of a few chains like Einstein Bros). Their model seems to be keeping a light footprint to improve payback and make buildouts affordable for franchisees: “PopUps are barebones, averaging 700-1,200 square feet with limited seating, making [for] speedy takeout operations requiring only a few workers. Each [store] costs roughly $300,000 to $900,000 to open.” But the franchising route raises some questions about the store economics (why franchise unless the store-level ROIC is low)? The biggest question is probably the same one as everyone asked about Shake Shack years ago: will this work outside of NYC?

(10) On The Vital Question by Nick Lane

A book review of Nick Lane’s The Vital Question on the origins of life. Lane’s theory proposes two energy breakthroughs. First, life emerged from alkaline hydrothermal vents where alkaline fluid meeting acidic ocean created natural proton gradients. The mineral wall of the sea vents acted as the first cell membrane, and the natural proton gradient drove the chemical reactions. Since every cell today powers itself using this exact gradient process (chemiosmosis), life likely emerged where this energy could be efficiently taken from the environment. Second, eukaryotic complexity in particular arose when early cells swallowed a symbiotic bacteria known today as… mitochondria. Mitochondria are famously the powerhouse of the cell: they gave the parent cells hundreds of internal powerhouses instead of one external membrane-bound energy system like prokaryotes, dramatically expanding their energy budget, and this energy efficiency is why only eukaryotes evolved complexity. It’s definitely easier to read this review while using ChatGPT.

(0) Miscellaneous

Zohran reintroduces himself. Somalia’s Pirate Stock Exchange. Best spots to swim in the Bosphorus. Colombia’s golden age of cocaine. The bear case on the AI boom. Dan Gilbert’s revival of Detroit. Profile of Fabrice Grinda. Who is OpenAI’s auditor? Yet another deep dive on Bending Spoons. Taking the slow train across Mongolia. Using airport slots as loan collateral. Does the code matter? The company as a machine for doing stuff. Jeff Yass on prediction markets. Fighting hospital over-billing using AI. Teamshares is going public via SPAC. How Airbus took off.

(1) Mystery of the Giant Shipping Line Linked to Macron’s Chief of Staff

The headline is about Macron, but the real story is about MSC. It’s very hard to find out anything about MSC… The Mediterranean Shipping Company is easily the most obscure company, relative to its importance, in the world. MSC is the world’s largest maritime shipping company (far bigger than Maersk), and the soon-to-be-largest port operator in the world. MSC is 100% owned by one person, founder Gianluigi Aponte, who likely ranks in the top 10 richest people in the world. Aponte started MSC with a single ship in 1970, and scaled through counter-cyclical purchases of secondhand ships, frugal operations on tertiary routes, acquisitions of port terminal chokepoints, and more. This article insinuates a certain shadiness about MSC, particularly its extreme secrecy and their control of the gang-infested port of Gioia Tauro in Calabria (Italy), but the rumors are, well, unproven. If you want to go down a crazy rabbit hole, it’s worth reading about MSC.

(2) Will Boom Successfully Build a Supersonic Airliner?

Betteridge’s law of headlines strikes again. The answer is “no” for lots of reasons: (1) it is very hard and expensive to develop a commercial aircraft (the Concorde cost $19 billion to build in 2025 dollars); (2) building a custom jet engine to go inside of that aircraft is even harder, and will cost billions more (most current jet engines cost $5+ billion to develop, and caused many of their companies to go bankrupt in the process); (3) a company building both the new aircraft and new engine for that aircraft has never been done in the jet airliner era; and (4) even if Boom succeeds in doing the physically and financially impossible, the economics of supersonic travel are unlikely to be profitable. “My impression of Boom is of a company that has struggled a lot (and spent a lot) to get where it is. This is understandable (building a supersonic aircraft is hard!), but [these] struggles have taken place before Boom has gotten to the hardest part: building the airliner, the engine, and raising the billions of dollars required.”

(3) Brand Man: The HJ Heinz Story

This article explains how HJ Heinz built one of America’s most enduring consumer brands. His innovations: (1) automation (he pioneered assembly lines before Ford); (2) creative marketing tactics (distinctive clear bottles, pickle-shaped trucks, billboards, and the famous “57 varieties” slogan); and (3) creative distribution (going direct to consumer through salesmen, rather than wholesalers, like other food manufacturers). Following the path of fellow Pittsburgh capitalist Andrew Carnegie, Heinz also optimized asset turnover over margin expansion to take market share: “Through mass production from farm to the grocer, he was able to lower the price of a bottle of ketchup from $1.75 (one to two days’ wages for most workers) to $0.35 for his best quality and 10 cents for the basic ketchup. […] HJ Heinz made every American family “richer.”

(4) Why So Few Matt Levines?

Matt Levine, famous for his Money Stuff newsletter, is arguably the best columnist in finance. There isn’t really an equivalent “Matt Levine for X” in other fields. Why? Is finance a uniquely good source of daily content? Is he a uniquely good writer? It’s a bit of both. The formula for Matt Levine’s writing comes down to: (1) first principles explanations, which cause readers to experience the “illusion of depth” in their understanding of the complex plumbing of finance (e.g., the legal mechanics of proxy contests, M&A, bond market liquidity, reverse stock splits, etc.).; (2) a news cycle that provides ample fodder for repeated illustrations of those complex underlying mechanisms of finance; and (3) lots of repetition, which doesn’t get boring because of his (4) excellent use of humor. His prodigious output is also a result of his liberal quoting from others’ articles, to which he adds commentary. His commentary is often better than the articles!

(5) Avoiding Bad Advice

“Giving advice to a startup founder on what path he or she should pursue is altogether different than giving advice on what entrée to order in a restaurant. [Founders] don’t have “multiple shots on goal” [and their incentives] are profoundly not like their investors […] Founders are working in one company at a time and putting many years of their lives into the effort […] they’re deploying their entire youth […] their entire soul […] their nights and weekends away from families, and […] even their physical and mental wellbeing. […] One wrong move and it can all crumble to nothing.”

(6) How Europe Crushes Innovation

Labor laws are turning Europe into a submerging market… “When firing people is costly, as it is in most of Europe, employers are reluctant to invest in risky ventures. […] Companies wanting to enact mass layoffs start with consultations with unions. A social plan social is drafted. Strikes inevitably ensue. Politicians get involved and badger the employer into firing fewer people than it originally planned. […] An American firm shedding workers will incur costs equivalent to paying those sacked for 7 months and be done with it. In Germany, those costs amount to 31 months of wages for each employee let go. In France, [it is] 38 months. […] When Bosch and Volkswagen, two German industrial titans, recently announced their own layoffs, the timelines stretched to 2030. […] Job security was once Europe’s comfort blanket. It now feels more like a straitjacket.”

(7) How Rockefeller and His Partners Built Standard Oil

The Gilded Age monopolies were built on increasing returns to scale. Rockefeller’s early days in oil refining demonstrates this. Standard’s early advantage came from a larger refinery and much tighter cost controls (which improved working capital turnover). But the real breakthrough with scale came from the Cleveland Massacre of 1872, when he leveraged railroad rebates to scare 25 local competitors into selling within one month. With 100% of Cleveland’s refining capacity, Standard Oil had the scale to crush fragmented markets elsewhere. The article understates the competitive dynamics that Rockefeller faced after 1872 (new oil fields, Empire Transport’s rail cars, the Tidewater pipeline threat, etc.), but makes a good case that refinery economics, not transport rebates, drove Standard’s early dominance. Economies of scale beat everything else.

(8) Relationships Are Co-Evolutionary Loops

“The best scientists aren’t those who are the most intelligent; they are the ones who tweak their theories further than others, in close contact with experimental reality. […] The same can be said of relationships. You need to keep a certain rate of improvement for things not to break. If you are too slow in adapting to each other, you will grow apart. […] Writing down everything that went wrong helped us [to] pinpoint the bottlenecks [and] forced us to put words to our goals, values, and assumptions, opening them for discussion and refinement. Is it really what you want? What are you willing to give up to achieve it? How do you feel about it?”

(9) Behind the Collapse of an Auto Parts Giant: A $2 Billion Hole and a Mysterious CEO

First Brands was a rollup of auto parts suppliers based in Ohio that recently went bankrupt. They “accumulated a total of some 25 different brands including Trico windshield-wiper blades, Fram motor oil, and air filters and Raybestos braking components. […] Revenue reached $5 billion last year.” The latest rumors are that their aggressive leverage—more than $6 billion in loans funding their acquisition spree—was partly built on fraudulent A/R factoring: “[the directors] have so far found around $2 billion more in borrowings than the underlying accounts receivables.” In other words, they double-borrowed against A/R. First Brands imploded so quickly because their slow growth gave them no margin for error, and tariffs brought the house of cards down: “selling replacement auto parts to […] AutoZone and Walmart did not offer up tremendous growth potential once acquisitions stopped. Between 2023-2024, [organic] revenue only rose 1.3%.”

(10) “You guys into deep value investing or what?”

“The moment you open the 10-K of a company trading at 0.3× tangible book, a succubus of deferred tax assets emerges, whispering “temporary impairment” while you realize the CFO’s last name matches the auditor’s. […] Instead of clarity, you start convincing yourself that the land value alone justifies enterprise value. Lord help you when the catalyst never arrives. You thought you were buying a cigar butt with one puff left. It turns out it was soaked in diesel and lit by management’s stock-based comp. But it’s too late — you’ve averaged down three times and told your followers this was “the most asymmetric setup of 2025.” Thoughts and prayers. Deep value was the intelligence community’s final psyop — to convince men that “margin of safety” still exists. The only safety is selling to the next fool before the delisting notice hits.”

Miscellaneous

Info edge with SEC phone geolocation data. Book review of Against the Grain. The business of online reputation management. Argument to ban fast food advertising. America has passed peak woke. How Verisk built a P&C insurance data moat. The medieval origins of the Vikings. Growth stocks and St. Petersburg Paradox. Some impacts of paying college athletes on sports. Baghdad is the latest boomtown. The screwworms are coming. Jared Kushner loves doing land deals. Duolingo is the #1 customer of OpenAI tokens. Mao’s persecution of sparrows. The Carnival of Aalst.

(1) Thomas Peterffy: The Making of a Market Maker

This is a rare interview with Thomas Peterffy, the founder of Interactive Brokers (a large online retail brokerage). The article has tons of crazy origin story anecdotes: gathering scrap metal in the streets of post-war Budapest; being rescued from homelessness in NYC by an alcoholic ex-monk and his gypsy palm-reading girlfriend; arbitraging silver certificates through the Federal Reserve; discovering Black-Scholes several years early; hijacking the Quotron data feed in his office; and (you have to read this part) rigging a mechanized mannequin to handle keyboard-based trade entry to comply with NASDAQ’s initial restrictions on automated trading. These stories don’t cover much about the business side of Interactive Brokers (there is plenty about that online already), but it does capture just how much hustle it took to build a great business.

(2) The Rise and Fall of the Hanseatic League

“In 1356, representatives from dozens of Hansa towns gathered in Lubeck for [the] first Diet of the Hanseatic League. The Diet sent an envoy to Bruges, another major Hansa kontor, to inform the Hansa merchants in Bruges that, henceforth, their negotiations with local authorities would only be valid if they were approved by the Hanseatic League. […] Membership of the Hanseatic League was national, and the privileges were meant only for Lower Germans dealing abroad in kontors in London, Bruges, Novgorod, and Bergen, outside the region. The League went to great lengths to stop foreigners from benefitting. This limited their possible membership, which meant that, as the merchants from other places started to compete in the Baltic, they were able to undermine the League’s collective bargaining power.”

(3) The Secret Corporate Empire That Captured A Breakaway State

In Transnistria, a Russian-aligned breakaway state of Moldova, a single company called Sheriff runs the economy. They control: “A cognac factory. A major football team. Supermarkets. Petrol stations. Banks. A textile factory. A TV station. A hotel. A mobile network. A casino. A caviar and sturgeon farm. A reported string of crypto mining sites.” Sheriff was started after the collapse of the USSR by a senior police officer in Tiraspol (hence “Sheriff”), and the conglomerate exploited the post-Soviet vacuum to monopolize the economy. Sheriff got started on cigarette smuggling through the port of Odessa (official imports suggest every resident, including newborns, would need to smoke 12 packs weekly to match Transnistria’s imports). Its secretive sole owner, Viktor Gusan, basically controls the entire breakaway country, and blocks reunification with Moldova.

(4) AI Will Not Make You Rich

“Anyone investing in the next big thing must answer two questions: First, how much value will this innovation create? And second, who will capture it? Information and communication tech (ICT) was a revolution whose value was captured by the startups and led to thousands of rich founders, employees, and investors. In contrast, shipping containerization was a revolution in which value was spread so thin that in the end, it made only a single founder temporarily rich.” It’s very hard to build moats around an “obvious” innovation because, once returns are obvious, capital flows in and competition moves fast enough to prevent anyone from getting dominance: “If there had been time to build moats, there might have been a few dominant containerization companies, and the people behind them would be at the top of the Forbes 400. [But] moats take time to build and, unlike [computers], the adoption of containerization wasn’t a surprise—every business with interests at stake had a plan immediately.” Counterpoint is Gianluigi Aponte.

(5) The Sagrada Familia Takes Its Final Shape

This year, Antoni Gaudí’s Sagrada Familia will finally be finished after 142 years. Construction of Barcelona’s famous church moved slowly from 1882 until a new director in 2012 applied modern engineering methods and increased tourism revenue (it’s now the most popular tourist attraction in Spain). The church is famous for its unique design with hyperboloid geometry, organic shapes reflecting nature, bright ceramic tiles, and more, which Gaudí used because “the world had seen enough ogival windows and flying buttresses.“ But in the intervening century, Barcelona mostly abandoned Catholicism, leaving the church as a tourist destination for “Tik Tokers to [witness] the grandeur of God.” Sagrada Familia is now world’s tallest church, and above it… only sky.

(6) Zohran Mamdani is Promising Lots of Things He Can’t Actually Do

“The risk from electing an NYC mayor with a vivid imagination is limited […] Mayors cannot set income or business taxes. Increasing them would require the state legislature to act, along with the governor’s signature. [For] rent, buses, and minimum wage, NYC’s mayor has limited control. He appoints a board that sets the rent for 50% of rented flats, the rent-stabilized ones. But boards don’t necessarily do the mayor’s bidding. The board voted 5-4 in June to raise rents on one-year leases by 3% and on two-year leases by 4.5%. Mr Mamdani also wants to build 200,000 units of affordable housing by borrowing $70 billion. That would exceed NYC’s debt limit by $30 billion, and therefore would require state approval.” Much better to be the Parks Commissioner!

(7) Whistleblower Rewards: A Paper Trail Tax and a Fraud Subsidy

“It’s very hard to design an incentive system for whistleblowing fraud that doesn’t in some way encourage the fraud. Any reward for stopping large frauds is an incentive to make them larger, and the fraudster generally has better information than law enforcement about how long the charade can continue. The other popular system for spotting fraud, activist short-selling, has much better-aligned incentives, since the short-sellers can’t easily encourage their target to break the law. A better whistleblower program would be [to] relax reporting requirements to encourage more companies to go public, so that outside investors have a stronger financial incentive to identify fraud—instead of perpetuating more of it.”

(8) Here Is How We Can Fix The Robocall Problem

Regulations like the TCPA and STIR/SHAKEN were supposed to fix the robocall epidemic, but the problem is only worsening. Scammers now operate primarily overseas, AI-generated voices make their scams easier, and loopholes continue to pop up with outdated telecom infrastructure (e.g., limitations on copper landlines for STIR/SHAKEN). This article proposes various regulatory solutions: (1) penalize the carriers that complete overseas robocalls; (2) close the loopholes that currently allow lead generation forms to share blanket consent with countless third-party callers (in progress!); and (3) strengthen enforcement of existing standards like STIR/SHAKEN to fight the spoofed numbers at the source. But if those ideas don’t work, there’s always Nomorobo.

(9) Review: The Hard Thing About Hard Things, by Ben Horowitz

A good review of one of the few good “management” books, with parallels to war, honor culture, and some great quotes: “The fundamental job of a CEO is to make a continuous stream of high-stakes decisions, while maintaining excellent quality of decision-making, despite not having time to think properly about any of them. “Oh no, they will experience decision fatigue.” Yes, but they need to suck it up and produce a constant stream of excellent decisions at a very high decisions-per-second anyway.” Another contender for quote of the year: “For millennia, a young man of uncommon energy and ambition had many ways to achieve his dreams. But for a few terrible decades, the only thing he could do was become a tech bro.” Related tweet.

(10) Joan of Arc

Joan of Arc is a national hero of France for turning the tide against the English in the Hundred Years’ War. The many miracles attributed to her—inexplicable skills with riding horses, placing artillery, commanding infantry, debating abstruse theology, and more—are unusual because of their detailed contemporaneous accounts. For Catholics who accept her sainthood: “Why [did] God stretch out his hand for this war in particular?” For atheists: “[If] you sort all the people in the world by how good evidence they are for God, [you won’t even] expect one in eight billion people to be that impressive by chance.” The article covers some secular explanations for her divine guidance, but in the end, most explanations fall short. Another mystery of the faith?

(0) Miscellaneous

Getting rich from rock aggregates. China’s near-christianization during the Taiping Rebellion. Annual cost of climate change is 0.02% of global GDP. Lessons from the WPP advertising rollup. Russian espionage. Reality of SMB acquisitions. Everyone wants to do accounting rollups. Zillow’s move into mortgage lending. Latest on AI existential risk. The benefits of sun exposure. Elections in Guyana. Absurd details about the US healthcare star system. The 50-pound book that runs Citadel. The newest play for AI data is recording calls. AI impact on Constellation Software.

(1) The Bitter Lesson vs. The Garbage Can

The infamous garbage can model of corporate decisions: “Teams [were] assigned to create process maps of their company, tracing what the organization actually did, from raw materials to finished goods. […] The CEO, after being walked through the map, sat down, put his head on the table, and said, "This is even more f— up than I imagined." The CEO realized that not only was the operation of his organization out of his control, but that his grasp on it was imaginary. […] One thing you learn studying (or working in) organizations is that they are all actually a bit of a mess. In fact, one classic organizational theory is actually called the Garbage Can Model.” Implications for enterprise AI: “Scaling AI [in] companies is hard because traditional automation requires clear rules and defined processes; the very things Garbage Can organizations lack.” Or maybe not: “the Bitter Lesson [of AI research] is that encoding human understanding into an AI tends to be worse than just letting the AI figure out how to solve the problem by adding enough computing power until it can do it better than any human. […] In a world where the Bitter Lesson holds, the despair of the CEO [is] misplaced. Instead of untangling every broken process, he just needs to define success and let AI navigate the mess. In fact, the Bitter Lesson might actually be sweet: all the undocumented workflows and informal networks that pervade organizations might not matter. What matters is knowing good output when you see it.”

(2) The ‘Smart’ Restrooms That Can Solve America’s Public Bathroom Crisis

Turns out there is already a company pursuing the idea of subscription toilets. Throne Labs is a venture-backed startup building subscription toilets (SaaS has a whole new meaning…). They’ve deployed over 100 toilets in the US, including a multi-year 100-toilet contract with the LA metro system. I wonder if this asset-heavy approach is better than the asset-light third-party model of a toilet pass? The best business model is probably a hybrid model of using the first-party toilets serving as public billboards for marketing (like the Carvana vending machines), then increasing network density with third-party toilets. If I started a competing service, I’d call it Royal Flush.

(3) Europe’s Crusade Against Air Conditioning is Insane

Europe’s residential A/C penetration is 20%. The result is an estimated 175,000 excess deaths from heat exposure in Europe. “The death rate from heat in Europe is almost twice the death rate from guns in America.” The official explanation for foregoing A/C is that its energy consumption worsens climate change, but the cultural backdrop is that many Europeans believe “A/C is the ultimate American solution to a problem […] many Europeans still think exposure to cold air can make you sick, and […] that A/C is for rich people.” Meanwhile, buried in the methodology of climate change literature, the largest estimated harms from climate change are excess heat-related mortality.

(4) The Outsized Impact of AI Logistics

Autonomous drones could change domestic shipping economics by replacing hub-and-spoke warehouses with point-to-point delivery for small parcels. Labor and fuel dominate LTL costs, which makes warehouse-based aggregation sensible, but these have corresponding costs with sorting, dock congestion, and inventory holding costs. The article covers the unit economics of drone flight versus current truck-based land transportation. The main assumptions required for drones to work are energy density (part of fuel costs) and regulatory clarity. Based on China’s experience, drones will have big benefits for rural parcel delivery and urban food delivery.

(5) On the Hunt for Nightcrawlers in the Worm-Picking Capital of the World

Most nightcrawlers sold for fishing bait in the US come from a few fields in southern Ontario. “The first recorded mention of the Canadian nightcrawler—a marketing name for the Lumbricus terrestris, dew worm, or earthworm—was only in 1902, he says, and the early bait worm industry was largely a side hustle by young boys looking to make a little spending money. It was only in the 1920s that people began hiring for the work, and only after WWII, when fishing became a popular weekend leisure activity, that the industry really took off.” But the industry is falling on hard times because of shortfalls in immigrant labor and because the youth don't want to inherit the family business: “everyone I spoke to also pointed out that worms aren’t a business that the next generation might be keen to take on, and it’s not a business you can easily sell off.”

(6) Emotional Agents

The next frontier of AI development will be improving their emotions. The dystopian stories of AI boyfriends and girlfriends are only the beginning: “Imagine if a teddy bear talked back, had infinite patience, and mirrored emotions. [Children would] never want to surrender their teddy.” The same critique of smartphones frying people’s brains might be supercharged with AI: "The central concern is not whether our connection with machines will be close and intimate (they will), nor whether these relationships are real (they are), nor whether they will preclude human relationships (they won’t), but rather who does your emotional agent work for? Who owns it? What is it being optimized for? Can you trust it to not manipulate you? These are the questions that will dominate the next decade.”

(7) Condoize Parking Spaces to Create Timeshares

Another business idea I looked into years ago… “Parking garages are mostly operated as pure rentals — a central management company that rents individual spots. But nothing (as far as I can tell) generally stops parking garages from condoizing like rental buildings do and selling their spaces to owners.” Entire parking garages in NYC sell for $30-$50,000+ per space, but “[individual] parking spaces regularly [sell] for $175,000 or above. [It] seems like there is an arbitrage in parking condoization.” In practice, the arbitrage does not work even for purely consumptive buyers (hint: look into individual sale data supporting the $175,000+). Even in theory, for investment purposes, the minimum efficient scale required for amortizing fixed administrative costs leads to owning multiple parking spaces, ideally in a single location…

(8) The Package King of Miami

How a student’s package scam came crashing down: “A UMiami student had been charged with orchestrating a cyber-scam that allegedly cost retailers millions of dollars [and] was facing up to 45 years in prison.” He used various methods of “refunding fraud” such as “[…] Fake Tracking ID, wherein the scammer returns an empty box to a retailer but edits the shipping label provided by the company to an address that is slightly different from the warehouse where returns are meant to go. The package gets scanned by UPS when it is picked up, allowing the customer to claim a refund, but it will never arrive at its destination. [Another method] involves writing addresses in special dissolving ink, which is legible when the package is picked up [by] the delivery company but invisible by the time it arrives at its destination. [The] gold-standard method is what’s known as “insider” — scammer slang for compromised retail- or shipping-company- employee accounts that allow the scammer to directly submit false information about a package.”

(9) Merton Real Options Framework for Debt and Equity

“Merton’s key insight is that if you can model the Firm’s Assets as a stochastic process, you can use the same Black-Scholes-like formula to model the Firm’s Equity as a Real Call Option on the Firm’s Assets. […] In the world of Real Options, owning Equity is like holding a Long Call Option on the Firm’s Assets, with the Strike Price being the face amount of the Firm’s Liabilities/Debt. […] Holding Debt is like writing (selling) a Put Option on the Firm’s Assets, with the Strike Price being the face amount of the Liabilities/Debt. […] Asset Volatility benefits the Long Call Option position and hurts the Short Put Option position, and this is where the Real Options framework yields extremely powerful insights into divergent corporate incentives [for equity versus debt].”

(10) The Wildest Arbitrage Story You've Ever Heard

I wouldn't call this the “wildest” ever, but it’s a good story with a good takeaway. The preferred stock of the Middlesex Water Company was historically convertible into its common stock on a 1-for-12 basis. Up until 2024, an investor could buy one preferred share for $95, then immediately convert it into 12 common shares trading at $50 each, turning $95 into $600… Easy money! But there’s always a catch. The opportunity was small and illiquid: the preferred shares had basically zero volume and the public float was under $1 million. True arbitrage is rare, and when it exists, it looks like this—obscure, illiquid, non-scalable—not “people don't fully appreciate Costco’s moat.”

(0) Miscellaneous Fun

Behind the success of Hot Ones celebrity interviews. The problem of legal delay. Why REA Group has so much pricing power. Bruges catching a stray. Bolivia’s kingdom of coca. South Africa’s failed affirmative action. Louis XIV was a ballet dancer. Origin of the TBPN podcast. Medicare cuts are hitting health insurers. Limits of pricing power at LVMH. Annual report on the tow truck industry. Risks of MSC’s port acquisitions. Corruption crackdown in Albania. Privatization of Fannie and Freddie. Geopolitics are heating up Svalbard. Supply chain of drug cartels. Supply chain of stolen cars.

(1) Do U.S. Ports Need More Automation?

“The ports that have adopted automation aren’t necessarily particularly efficient […] the most automated port [in the US], Los Angeles, [is] very close to the bottom of worldwide rankings.” Some reasons why port automation isn’t a magical fix for port processing times: (1) automated container handling struggles with real-world conditions more than expert crane operators (e.g., handling fog, shifting in the waves, etc.); (2) the high fixed cost of automated cranes financially struggles in the balance with high variance container volumes (variable labor cost is more easily aligned with variable container shipments); (3) automation of loading/unloading doesn't fix the other bottlenecks in the port (e.g., availability of truck chassis); and (4) lack of port coordination: “[Most] US ports lack appointment systems for trucks to arrive or pick up cargo; instead, trucks arrive randomly, often creating large amounts of congestion as they wait to load and unload.”

(2) A Priority Pass for Bathrooms

I’ve thought about this idea for years… Ever wandered around in NYC (or any other big US city) and needed a bathroom? There aren’t many public bathrooms; you often have to pay the “Starbucks tax” to use one in a coffeeshop. The (proposed) solution: a subscription pass for toilets. Toilets are a classic under-utilized asset that could be put in a “network” for subscribers to use, with fees paid to the restaurants, shops, etc., to make their semi-public toilets available. Smart locks could control access and tracking (so they can ban anyone who clogs the toilet). To increase adoption, this could initially be given out as a credit card rewards perk. The author thinks that this could generate >$100M in annual sales. That sounds a bit high, but who knows… I know someone who pays for Equinox just to have a toilet anywhere in the city.

(3) CEO Explains How He Faked Results in $300 Million Meltdown

The story of the failed Indonesian aquaculture startup eFishery. This Softbank-backed unicorn sold a remote-controlled fish feed dispenser—a decent idea!—but ran into the classic problems of using hype and fundraises to mask underlying problems in the business. Among other financial games, eFishery paid farmers to re-route existing off-market transactions onto their “marketplace,” then booked that GMV as revenue (“90% of eFishery’s revenue came from businesses with less than 2% of gross profit margin”), and raised over $300 million on that rising revenue. A second trick was using poorly-underwritten loans to farmers to purchase their products, then obscuring the bad debts and lack of cash conversion. Sounds like the real fish food was the investors.

(4) The Rise and Fall of the Stock Market’s Music King

Hipgnosis was a COVID-era rollup of music royalties. It didn’t go well. Some reasons: (1) overly-aggressive capital deployment fueled by cheap money (“Armed with cash from repeated equity raises and millions of dollars in debt, they orchestrated acquisitions of 29 catalogs in 2019 and 72 in 2020”); worsened by (2) overpaying for acquisitions (“Before [Hipgnosis], catalogs often traded in the range of 10x earnings. Hipgnosis stunned the industry by paying twice that. [The multiples] were definitely north of 20x.”; and (3) “creative” accounting to keep capital flowing (“Hipgnosis frequently [said] that it had owned 100% of catalogs when in many cases it had purchased only revenue streams, which [is] like being entitled to rent from a building but not owning the real estate.“). When the rollup music stops, the song always sounds the same.

(5) Apple + Anthropic?, Apple’s Fall, and Apple’s Options

Stratechery on Apple’s AI dilemma: “While I as a consumer would prefer they take the ChatGPT option — ChatGPT is for me more important than the iPhone — if I put my business analyst hat on I think they should pay Anthropic. Most consumers aren’t me, and Apple will ultimately be both more comfortable and retain the greatest control over its destiny by — from most users’ perspective — magically delivering them a better Siri (powered by Anthropic) than simply giving up and slotting in ChatGPT. Or here’s another idea: finally accept that the company that is most aligned with Apple’s own business interests is the one Apple’s leadership hates the most: Mark Zuckerberg would welcome support for Llama’s development — he’s certainly paying enough!”

(6) Why Skyscrapers Are So Short

“Since legal restrictions place limits on what would otherwise be economical to produce, they have a way of becoming optimization targets. No modern building is a better illustration than the 5 + 2 podium, a type of building that exists almost entirely due to building code provisions. US building codes allow the use of light-framed wood construction for multifamily apartment buildings so long as the wood portion does not exceed 5 (sometimes 6) stories in height. And buildings taller than 7 stories are considered “high-rise” construction, a designation that brings with it many burdensome and expensive code provisions. The result is that multifamily buildings often consist of a “podium” of 1-2 stories of concrete ([with] parking, retail space, or amenities), which has 4 or 5 stories of wood apartments placed on top of it. Because it’s so economical, a large fraction of [US] residential construction in urban areas is podium construction.”

(7) Entrepreneurship Through Acquisition

“In a self funded search, where you have students raising money to buy $1 million-ish EBITDA companies one and a half million dollars and buy them at four times and finance them at 80%, the math is just extraordinary. If you buy something for 4x, you are generating a 25% return on assets before any growth. You lever that at 80% with debt that costs 8 or 10% and the return is astronomical. […] Most people discover that it is a highly personal activity […] they are going to meet people [and] going to go door to door, [they] are [meeting] every business owner in the [market]. […] The vast majority of the value created [is] upfront [because] the magic is in the multiple [and because of] the difficulty of finding a company and getting a transaction done.”

(8) Valve Conquered PC Gaming. What Comes Next?

Valve is an unusual and lucrative business. Its main product is Steam, which is like an App Store plus Facebook plus Reddit for the PC gaming industry. Steam earns about $11M in revenue per employee (compared to ~$2.5M per employee for Apple) and has >60% operating margins. Steam originated from the need for PC game developers to distribute game updates in a way that could avoid piracy, but evolved into a de facto storefront and social network for the fragmented PC gaming industry. The company’s extreme secrecy and operating quirks (e.g., “no hierarchy”) are covered in the article, but I would guess that its success came in spite of (not because of) those oddities.

(9) Hillbilly Elegy – The Culture of White American Poverty

A review of JD Vance’s biography. Vance focuses on the self-perpetuating culture of white rural poverty, and a central theme is the victimhood culture. Their life “happens” to them, rather than being the result of their choices. Some recurring bad choices are a lack of emotional regulation, discipline, planning, and hypocrisy. “Vance’s neighbor growing up was an obese woman whose favorite pastime was sitting on her porch, railing against Democrats and their stupid welfare systems which the poor greedily leeched off of. Meanwhile, [she] had lived off of welfare since the day she had left high school. When pressed, she argued that she was one of the few people who deserved a little help, while most welfare users were abusive parasites.” How he escaped: “The biggest turning point was joining the Marines out of high school. The way he describes it, the Marines acted like a crash course in middle-class culture. That gave him a reliable authority, enforced discipline, and showed him that he had underestimated his own control over his life.”

(10) Against "Brain Damage" - AI Can Help, Or Hurt, Our Thinking

“Our fear of AI damaging our brains is actually a fear of our own laziness. [AI] offers an easy out of the hard work of thinking, and we worry we will take it. We should worry. But we should also remember that we have a choice. […] This principle becomes even more critical in writing. Many writers insist that "writing is thinking," and while this isn't universally true (I generated a pretty good Deep Research report on the topic if you want the details), it often is. The act of writing, and rewriting, and rewriting again helps you hone the ideas. If you let AI handle your writing, you skip the thinking part entirely.”

(0) Miscellaneous Fun

Best European cities. Best restaurants in Antwerp. Regulatory driver of local banking rollups. How duct tape saved Apollo 13. El Salvador’s descent into dictatorship. Death of a fugitive carbon credits broker. The war on rats in NYC. How to buy small private islands in NYC. Why the English language doesn't use accents. The history of the NY Produce Exchange. Critique of stablecoins. The rise of Bad Bunny. Why we should let the LLMs daydream. Creative use of transshipment tariffs. Pivot at Substack. Update on Appointment Trader. Building the SKIMS brand. BrewDog’s preferred stock mess.

(1) Green-Wood Cemetery’s Living Dead

Green-Wood is the largest cemetery in NYC. Despite a slow start after launching in 1838, Green-Wood gained popularity with “celebrity” burials like former NY governor DeWitt Clinton. Famous burials soon followed: Samuel Morse, Boss Tweed, Charles Tiffany, Horace Greeley, Henry Ward Beecher, Henry George, and Leonard Bernstein. Green-Wood has more than 600,000 permanent residents on its 478 acres, among the most densely “populated” parts of NYC: “A grave at Green-Wood is the only real estate that some New Yorkers ever own.” Spaces are running low: “Green-Wood will be out of room for burials in 5 years.” But as that deadline looms, they are turning to tourism with concerts, movies, and tours. The cemetery hosted 300 events last year, and some are quite popular: “A recent panel featuring gravediggers sold out.” People are dying to get in!

(2) The Doomsday Cult’s Guide to Taking Over a Country

South Korea has a weird number of end-of-the-world Christian cults (an estimated 3% of South Korea’s population belongs to a Christian cult). One of these cults, known as Grace Road, has been repeating the Jonestown Guyana playbook in Fiji. Grace Road emerged in South Korea in 2008, and then moved to Fiji in 2014. Since then, its 400 refugee cult members created a tropical business empire: “Between 2015 and 2025 they set up six beauty salons, five True Mart supermarkets, a dental clinic, a construction firm, five patisseries, three Snowy House ice-cream parlors, a chain of eight Korean restaurants called Grace Kitchen, nine pizza parlors, two Italian trattorias and a fried-chicken restaurant called Fierce Chicken.” Don't drink the Flavor Aid.

(3) The Capital Asset Pricing Model: Risk, Reward, and Reasons

“To get a good [return], you need a good reason to expect it: some justified belief that there's a particular asset that is mispriced, and a good meta-reason to think nobody else has considered it. In a competitive market, you need not just a theory about the asset in question, but [also] a theory of why your competitors haven't spotted it. Yes, "I'm a better investor than they are" is a theory—but do any of your competitors pick stocks despite consciously thinking they're bad at it? […] You should assume that the average market participant is as self-confident as you are. […] The opportunity cost is the index. […] Investing ends up being like many other hobbies: it takes time and effort, and if you're very good and somewhat lucky, you might break even.”

(4) The Original Bourse at Bruges

The first stock exchange wasn’t in Amsterdam or Antwerp, but Bruges. This now-quaint tourist town in Flanders (Belgium) became the dominant trading hub for Hanseatic League merchants (from the North Sea) and Mediterranean merchants (mostly Florence and Genoa), with special focus on textiles, commodities, and FX. The first exchange was a plaza, the Beursplein, located in front of the hotel owned by the Van Der Beurse family, which gave the modern name to stock exchanges in many languages (bourse, bursa, etc.). The story of Bruges is interesting for various reasons: “For one, a prominent family became the etymological origin of the very word ‘bourse,’ but more significantly, [Bruges] marks a turning point in Europe’s economic history, the transition from Mediterranean trade to Atlantic trade dominated by the French, English, and Dutch [… and] the [creation of] practices [that were] the template for every exchange that followed.”

(5) How This Refugee Became Albania’s First Billionaire

Samir Mane is Albania’s first billionaire. But he’s not an oligarch! He fled communist Albania as a student in the early 1990s, then returned a few years later to start the country’s first electronics retailer (Neptune), selling western products like washing machines. Later, he built Albania’s first shopping mall and grocery store, while also expanding his retail investments and land holdings across the Balkans. Now, his company BALFIN Group is building luxury villas in advance of the coming Albanian tourism boom. Retail, real estate, and tourism. Skanderbeg would be proud.

(6) Capitalism, Socialism, and Social Desirability Bias

“Republicans’ critics often claim that they favor a world where “anyone who can’t afford health care dies.” Yet in reality, virtually zero Republicans favor such a world. At most, they oppose the expansion of medical programs that already spend over a trillion taxpayer dollars per year. The brutally honest argument against government-funded health care is that the cost of saving the marginal life is too damn high. Spending $100,000 each month to keep an 85-year-old on life support is a terrible use of taxpayer money; a faithful steward of the public interest would pull the plug. Yet flesh-and-blood Republicans don’t merely avoid this brutal truth. They make the opposite argument, accusing the [Democrats] of supporting “death panels” to deny taxpayer funding for cost-ineffective coverage.” But nobody will vote for pulling the plug on grandma.

(7) Asurion: The Early Days (‘95-‘01) with Kevin Taweel

Asurion started out as a Houston tow truck dispatch company, but found its way into cellphone insurance through a small insurance acquisition that let them cross-sell roadside assistance and phone protection through the same telecom carrier relationships. By 2001, cellphone insurance had completely overtaken the original roadside business, riding the wave of Americans buying increasingly expensive and fragile smartphones. The returns tell the story: “A dollar invested in the original purchase...has grown at a compound annual rate of over 61%...translating into an MOIC north of 5275x.” Among others, some lessons: (1) cross-selling into existing distribution channels is lucrative; (2) cross-selling niche insurance with a low yearly price and standardized underwriting is also lucrative; and (3) mobile phone adoption was a powerful secular tailwind.

(8) The Illusion of Self-Improvement: Why AI Can’t Think Its Way to Genius

LLMs have ingested a good portion of the corpus of written human knowledge, but they have not produced many knowledge breakthroughs (yet). Why? Because “thinking harder” mistakes the process of how we gather knowledge. Most breakthroughs and insights come from external and novel information—from interacting with the real world—rather than from endless reflection on existing knowledge. Columbus didn’t discover the New World just by thinking deep thoughts; Fleming didn't discover penicillin just by thinking deep thoughts; and so on. This is important for understanding both the current limits of LLMs, and how those limits could be overcome: (1) tool-augmented reasoning with access to fresh data; (2) multi-agent verification with cross-checking; (3) environment-grounded tasks with external inputs; (4) external feedback loops; and more.

(9) How a Tiny Middleman Could Access Two-Factor Login Codes From Tech Giants

When a company sends a 2FA code via SMS (e.g., Uber alerts), the SMS rarely goes directly from the company to the recipient. The SMS is passed into a chain of subcontractors that interconnect the telecoms. The telecoms lease their “global titles” (essentially their telecom address), allowing subcontractors to handle SMS routing. But SMS is rarely encrypted. Bad actors set up shop as subcontractors and step into the telecom chain to read the unencrypted messages, often with a goal of (what else?) stealing cryptocurrency: “About 20 Israeli cryptocurrency investors [had] their 2FA codes intercepted by manipulating telecom traffic. The investigation traced the manipulation to a Global Title registered to SMSRelay, [where a security breach] allowed an attacker to access the investors' email and crypto accounts by capturing the login codes sent via SMS.”

(10) August Belmont and the NYC Subway

The story of 1800s financier August Belmont and his role in the construction of the NYC subway: “The Rapid Transit Subway Construction Company was formed in 1900 to carry out the work. Another company, the Interborough Rapid Transit Company or IRT, which would actually operate the system, was established two years later. The company operated the subway free of [city] taxation and had the right to collect a five-cent fare from all passengers. [The] IRT was responsible for paying the interest charges on the [loan from the city], which amounted to 3.25% per year. [The] subway caused Belmont great financial trouble. Fares were fixed at five cents [but] inflation after WWI meant that the cost of inputs like coal and wages were rising. By 1921, Belmont’s Interborough Rapid Transit Company was bankrupt.”

(0) Miscellaneous

The continued success of Walmart. Backstory of the egg shortage. The financial post-mortem on Thrasio. Shenzhen’s stolen iPhone building. Inside scoop on the Builder.ai fraud. Another Duolingo interview. Obvious critique of stablecoins for FX. Predicting SCOTUS decisions with AI. Selling geckos online. Deepfake marketing in healthcare. Consensus was wrong about tariffs and inflation. The crackdown on NYC biking. The mess of addresses in India. Election results in Suriname. Underrated societal harm of noise. Continued decline of Germany’s work ethic. Interview with Applovin CEO. The history of the National Guard. Wild corporate espionage. France’s Catholic comeback. Wagner Group stabilized the Central African Republic. Underrated Buffett anecdote.

(1) The Discovery of Copper

Copper is a critical natural resource for modern electrical equipment. The world has been worried about running out of copper since the days of Edison, but we always find more. How? Because of productivity improvements in copper mining. Daniel Jackling invented modern copper refining in 1906 with the “flotation separation” method. This enabled extracting copper from lower grade ores: “Enormous chunks of rock are blasted out, ground down into a dust as fine as face powder and separated in flotation tanks. Then these concentrated ores undergo further pummeling and processing. The final step [involves] electrolysis: 98.6 percent pure copper slabs are dunked in tanks containing an acidic solution, where they sit for nearly two weeks while an electric current is run through them. What you end up with – after metal atoms in the anode, the slab you started with, have migrated to the cathodes – is a sheet of 99.9 percent pure copper.” With over two millennia of refining the refining process, we have taken the time required to extract a ton of copper from over 40 years to just 21 days. Productivity is about doing more with less.

(2) No-Parking Zone: The Perils of Finding a Spot in NYC

One of the best articles I’ve ever read about the insanity of subsidized on-street parking. End-to-end, NYC’s 3 million on-street parking spaces would stretch to Australia. These occupy 13x the real estate of Central Park, yet 97% of the spaces are free (read: subsidized). It is obvious that prices are below market: (1) garage parking (the market price) costs upwards of $500 per month, and (2) NYC residents go to absurd lengths to avoid paying for their parking (residents spend 200 million hours per year driving around to find parking). “Free parking is not actually free. We [indirectly] spend as much on parking—via taxes, higher rents, and subsidies—as we spend on Medicare.” The obvious (but politically unpopular) solution is dynamic and market-based pricing: “Charge enough that there is always an open parking space on every block.”

(3) Book Review: Zero To One

Peter Thiel’s Zero to One is, in part, an indirect critique of the “efficient markets hypothesis” (EMH) in the context of startups. The basic logic is that, if the market for startup ideas were efficient, then all good startups would already be created. Since good startups are still being created, this disproves startup EMH. He mischaracterizes EMH—it’s not about possibility, but predictability—but otherwise lays out a strong (and EMH-consistent!) argument that “secrets” (EMH-speak: “info that isn’t yet available to the market”) are behind most successful startups.

(4) The Dawn of AI Rollups

“There are a few characteristics of the markets/sectors where AI rollups will be particularly effective. In no particular order, these markets should be: (1) highly fragmented; (2) large and growing; (3) lacking technology expertise and reluctant to buy software; (4) high revenue, low margin; (5) driven by human knowledge work (human knowledge labor is a majority of OpEx); (6) relationship-based, so businesses and their customers have sticky, long-term, trust-based contracts. [The] very best rollups will buy businesses that own unique and valuable datasets that can be used to train models […] the ideal sectors [are] ones where rollups are not already ubiquitous [so that] acquisitions can still happen at reasonable prices (ideally 1-3x EBITDA).”

(5) A Dark History of the World’s Smallest Island Nation

Nauru is the cautionary tale of the resource curse. This Pacific micro-nation transformed from a German colony to a phosphate powerhouse after guano deposits were discovered in the 1900s. By independence in 1968, 35 million tons of phosphate had left its shores—enough to fill dump trucks from NYC to LA and back. Nauruans briefly became rich, with a trust fund valued at over $1 billion and a per-capita GDP second only to Saudi Arabia. This wealth fueled absurdities like (of course) everyone buying sports cars on an island with one road. Once the phosphate ran out, Nauru took a shady turn with offshore banking, passport sales, and hosting an Australian migrant detention facility. Meanwhile, the sovereign wealth fund from phosphate shrank to $30 million by 2004, leaving a poor population with not much to do other than playing bingo.

(6) Where We Are Headed: A Rough Sketch of the Near Future

How AI agents are going to change the world: “Think about the mechanics of knowledge work. At the most basic level, you’re operating a keyboard and a mouse. Did you click the right button on the user interface? That’s pretty easy to verify. Did you successfully retrieve the right piece of information, or pick the right tool? That’s more subjective, but in a constrained setting, not that hard to verify. Did you write a good memo? Trickier, but with automated grading, it’s achievable. Did you fill out this form correctly? This one ranges from “trivially easy to verify” and “incredibly difficult,” but for many forms, it will be doable. As you go about your day, occasionally stop and think to yourself, “would it be easy to cheaply verify that I am doing this task correctly?” The answers vary, but I suspect you’ll find that the answer is often “yes.” This has implications for what the near-term economic consequences of agents are likely to be.”

(7) Are Restaurant Reservation Apps Worth It?

For those seeking a restaurant reservation in NYC, there is an entire ecosystem of apps to help. Dorsia has emerged as a leader, with its approach of charging both restaurants and customers—recently securing a $50.4 million investment that values the company at $146 million. The once-popular secondary markets for reservations (notably Appointment Trader, a one-man business with $7.2 million in revenue (!) last year) have been effectively outlawed by the recent Restaurant Reservation Anti-Piracy Act. Restaurants prefer Dorsia's method because they share the price discrimination economics, and the restaurants also have gotten creative by using minimum spend requirements that monetize table access while crediting it back against the meal.

(8) How Shell Entities and Lack of Ownership Transparency Facilitate Tax Evasion

“Three principal reasons explain the ability of tax evaders and others to continue to hide their identities as beneficial owners and operators. [First] is a legal framework in many jurisdictions that promotes lack of ownership transparency. [Second] is that those who abuse shell entities need the services of gatekeepers like accountants, lawyers, and TCSPs. [Third] is layering, or chaining of numerous shell entities in different jurisdictions, [which makes it] impossible for forensic accountants and tax authorities to discern the real identity of beneficial owners.”

(9) Stablecoins are Bank Deposits

Stablecoins sound like narrow banking to me: “A stablecoin issuer is a bank and a stablecoin is a bank deposit. This is not complicated. If you hand me money and I invest it, and in return I give you something that is a liability for me and an asset for you, and that is redeemable by you on demand and at par, I am a bank and the thing I have handed you is a deposit. It doesn’t matter if that thing also works as an intermediary in a crypto market, a token in a cross-border payment app or gets you a gumball out of a gumball machine. I’m a bank, you’re a depositor, we’re in this together. […] There are two main questions: Are these new sorts of banks, called stablecoin issuers, risky? And do they solve a problem that needs solving?” How is sending stablecoins different than Venmo? I’m only half kidding…

(10) The Disapproval of the Crowd

Palantir CEO Alex Karp: “We have today privileged a kind of ease in corporate life, a culture of agreeableness that can move institutions away, not toward, creative output. The impulse […] to smooth over any hint of conflict within businesses and government agencies is misguided, leaving many with the misimpression that a life of ease awaits and rewarding those whose principal desire is the approval of others. […] A certain psychological resilience, and indeed indifference, to the opinion of others are required if one is to have any hope of building something substantial and differentiated. […] As the comedian John Mulaney has said, “Likability is a jail.”

(0) Miscellaneous

Exotic tourism in Algeria. The bizarre story of metaverse “unicorn” Infinite Reality. The 1500s papal conclave prediction market. The mysterious buyer of the Boston Celtics. The secret to Elon Musk’s productivity. Cuba’s doctor diplomacy. Rising healthcare costs due to hospital consolidation. Predictions about AI’s impact on writing. Relatable habit of parenthetical writing. Undervalued real estate in Baja California. Update on the Mauritius economy. Profile of Brad Jacobs. Improving NYC’s biking laws. The unraveling of the King of Davos. Backlash against golden passports. The adventures of Barbecue in Haiti. Origins of GT Dave’s Kombucha.

(1) What It’s Like to Retire in Belgium

“Antwerp has been my home for almost three years. It is a vibrant, multicultural city famous for its busy port, its fashion and design scenes, the local diamond industry, and Flemish art masters. […] For roughly the same amount of space as I had in California, I pay 75% less in rent. Spacious one-bedroom apartments aren’t difficult to find here and rarely exceed 1,200 euros (equivalent to $1,320). There is a lot to see and do […] Neighborhood-scale parties and fairs happen every month, as do major music festivals, bike races and international events. […] Happily, I no longer have a car […] I can take a day trip [by train] to Leuven, Brussels, or Bruges and return home by midnight for less than $25. As for flights, Brussels Airport is a $20 train ticket from Antwerp. […] I have no complaints about Belgian healthcare […] The cost of every health procedure I’ve had in Belgium has added up to less than my annual copay and $2,000 deductible of my U.S. policy.”

(2) How the World’s Top Shipping Company Became a Hub for Drug Trafficking

The world's largest shipping company has an unlikely second business line: cocaine smuggling. The Mediterranean Shipping Company (MSC), founded by secretive Swiss billionaire Gianluigi Aponte, handles 20% of global cargo (even larger than Maersk). Aponte built his empire from scratch in 1970, buying secondhand ships and focusing on underserved routes with a business model that prioritized price over punctuality (in the early days, MSC jokingly meant "maybe ship comes”). While Aponte was building his legitimate enterprise, the Balkan Cartel was establishing a parallel operation within it. A network of Balkan gangs, offshoots from the 1990s Balkan wars, infiltrated MSC's operations by recruiting company employees, particularly on South America to Antwerp routes. It’s a great article about the history of MSC, the logistics of drug trafficking with mid-sea drug transfers, and the DEA’s methods of tracking and busting those drug shipments.

(3) Book Review of “Believe” by Ross Douthat

Book review with a defense of Christianity. Amusing questions to consider: “There are people who will tell you that Jesus did not rise from the dead, but that He was a great moral teacher with a lot of good advice. But if He didn’t rise from the dead, then wasn’t it actually the worst advice ever? The example Jesus provides is an example of how to get killed: first you make people jealous, then you make them mad, then you confuse and demoralize your followers, then you refuse to speak in your own defense. If the real story ended with the Crucifixion, then surely it’s a story about what not to do. The lesson we would draw from it would be, “Don’t imitate Jesus, imitate the worldliness and adaptability of the Jewish elders, imitate the power and cruelty of Rome.” The whole question then, the one on which our interpretation of this very strange story ought to hang, is what really happened on the third day after the execution.”

(4) How a Secretive Gambler Called ‘The Joker’ Took Down the Texas Lottery

The secretive gambler Zeljko Ranogajec turned 25.8 million $1 lottery tickets into a $57.8 million jackpot by purchasing nearly every possible combo in the Texas state lottery. His (entirely legal!) scheme exploited a loophole in Texas' lottery system, which allowed online vendors to establish ticket-printing operations rather than requiring retail in-store purchases. He enlisted struggling startup Lottery.com, who convinced the Texas Lottery Commission to deliver dozens of ticket-printing terminals directly to four workshops he established for the operation. The math itself was straightforward (buy all combinations when the jackpot exceeds the cost). Ranogajec’s genius is in finding the regulatory loopholes and coordinating the logistics to exploit them.

(5) 50 Things I Now Know About Investing Because I Am 50

Great words of wisdom: (1) “Nothing else has cost me more money than speaking to CEOs;” (2) “[An] inconvenient truth of the stock market is that very few people get rich investing. You have a much higher likelihood of getting rich from starting and building a business. Trying to get rich in the stock market is tempting for all sorts of reasons, but it is insanely difficult;” (3) “Alpha is earned where no one else is looking […] if placing the buy order doesn't make you break out in a cold sweat, it's probably not a real contrarian investment;” (4) “I can directly trace back quite a few of my bad decisions to my lack of focus at the time […] there is no such thing as a day off;” (5) “Read OLD books […] I’ve generated outstanding investment ideas by reading out-of-print books that others had long forgotten about. You could even go as far as calling this approach one of my secret weapons – not just with regard to investing, but life generally.” I could quote every single piece of advice in this article. Timeless wisdom from a lifetime in public markets.

(6) 50 Things I’ve Learned Writing Construction Physics

Everyone is writing 50 things posts this month… some great takeaways on construction, energy, logistics, and more: (1) there is no easy explanation for why construction isn’t getting cheaper—common theories like design variety, regulations, a lack of innovation, etc., all fall short, and the best explanation is that (for good reasons) on-site construction is hard to automate; (2) “A key bottleneck in semiconductor manufacturing is high-purity quartz, most of which comes from a single town in North Carolina: Spruce Pine;” (3) “AT&T spent substantially more on telephone infrastructure in the 1960s than NASA spent on the Apollo Program;” and a great conclusion from his writing: (4) “Things that seem like recent developments often have key predecessors going back decades or even centuries. What seem like historical inevitabilities are often highly contingent products of chance and circumstance. Intuitive explanations [are] often wrong, and even when they’re right, the full story is often much deeper and more interesting. Causes are often complex, and what seem like simple problems often tenaciously resist solutions.”

(7) Apple and the Ghost of Companies Past

Microsoft and Intel lost out on the mobile era (relative to their market position in the 1990s) because they attempted to apply the PC paradigm to mobile, rather than rethinking the user interface. Apple risks the same mistake with trying to pigeonhole AI into an “app” paradigm, which undergirds the iPhone’s dominance. The iPhone could be an open platform for others’ models, rather than just Siri, but Apple’s focus on privacy (and ecosystem control) could cede market share to Android: “If a similar story [to Microsoft] unfolds for Apple, it might look like this: (1) Apple misses AI because it is focused on the wrong thing (privacy). (2) Apple fails to leverage its greatest strength (the iPhone platform) into a [platform] position in AI (being the platform for the best model makers). (3) Apple’s platform falls behind the industry’s champion (Android or perhaps TBD), which [threatens] Apple’s core business (AI is so important that the iPhone has a worse user experience).” Whatever happens will be obvious in hindsight!

(8) Zyn and the New Nicotine Gold Rush

The Swedes have transformed snus nicotine delivery with their little white pouches. Zyn came from efforts to enhance the effectiveness of nicotine gum, like Nicorette, while minimizing side effects like hiccups and stomach discomfort. Zyn is a life saver for tobacco companies (unclear about consumers) that want to capitalize on the demand for nicotine while shedding the stigma and health concerns linked to cigarettes. What are the risks? Nobody really knows… although the consensus seems to be that they are “less bad?” Regardless, as one snus user summarized, “You take risks every day—by driving, what you eat, and so on,” he said. “I am a proud nicotinist.”

(9) AGI Will Not Make Labor Worthless

Some counterarguments to the concerns about AI taking our jobs: (1) labor has maintained a relatively constant share of global income at ~50% of GDP since the Industrial Revolution; (2) automation will increase the productivity of workers in ways that increase the demand for their work (i.e., spreadsheets did not eliminate accountants); (3) AI automation will create new jobs that do not currently exist; and (4) energy is not infinite, so there might still be constraints on AI automating everything—the law of comparative advantage ensures that humans still have some role in the world. But if every unit of energy will more efficiently increase production when it is powering a datacenter instead of powering a human… what happens next? Paperclips?

(10) Limits of Smart

The current methods of training AI models rely on more data to scale (data being, loosely, “info about the world”). Some believe that these models could create synthetic data to bootstrap the learning process. But consider how humans create more data to generate novel insights. Lots of data comes from physical interaction with the world, not just passively, but from experiments. In science, for example, we build particle accelerators to measure quantum mechanics, telescopes to observe the universe, drug trials to determine their impact on humans, etc. In this sense, new knowledge comes from new info, and new info comes from new experiments (e.g., interactions) with the world. Experiments to gather data require a lot of money and time—could AI accelerate the process of how novel chemicals affect humans in drug trials? It is hard to predict how AI will develop, but maybe Asimov was right: “There is insufficient data for a meaningful answer.”

(0) Miscellaneous

Jensen Huang’s hobbies are “work, email, and work.” Lots of interesting NYC quality of life statistics. The White House theory of tariffs. Original Microsoft Altair BASIC source code. How Marine Le Pen embezzled EU party funds. Elon Musk is a unicorn. Netanyahu’s march to the sea. Little Diomede Island on the front lines with Russia. Design principles for AI-first apps. De Beers crisis in the diamond market. Some comparisons between EU and US cities. Importance of corporate planning. Trump’s hardball negotiations with big law. Shopify expects all employees to use AI. Behind the scenes of Elon Musk’s harem. The financial history of Seville. Good travel post on Antwerp. Economic collapse in Cuba. Ecotourism vacation to São Tomé and Príncipe. Public ownership of US airports. Nashville church hostile takeover.

(1) 50 Thoughts on DOGE

This is the most thoughtful review of DOGE I’ve read. Two good critiques of DOGE are: (1) the focus on measurable business efficiency metrics—headcount and dollars saved—is avoiding the bigger cost of government in implicit costs (e.g., bureaucracy that costs 20+ hours per year, per person, on filing taxes); and (2) his "turn everything off" playbook from Twitter/SpaceX has cut state capacity by eliminating competent employees from programs like USAID. Two reasonable responses to those critiques are: (1) his “shoot first, ask questions later" approach might be the only way to defeat entrenched resistance in a limited timeframe; and (2) programs like USAID have real accountability problems that need addressing. The article contains other thoughtful points on the pros and cons of DOGE, but the conclusion is a good point to keep in mind: "I still find it hard to evaluate DOGE in toto. There are things I hate and things I very much appreciate, but it still feels too early to pass judgment."

(2) Is Elon Musk remaking government or breaking it?

“This does not mean DOGE has failed—yet. There are three possible outcomes. First, that just as rivals laughed at Tesla and SpaceX in their early days, DOGE will come around in time. Second, that Musk will break the government. The third, likeliest scenario is that DOGE becomes snarled up in court; many good civil servants are fired or quit; fewer talented people see government as an appealing career; and America is left with a stronger president and a weaker Congress. This would be a huge missed opportunity. Imagine the Musk of the early 2010s, the genius-builder, in charge of procurement at the Pentagon or federal infrastructure projects. Instead, America has got the late-era Musk, radicalized by his own social-media platform, flirting with authoritarian movements, stuck in the same mind-numbing partisan thinking as millions of less talented folk.”

(3) Why It’s So Hard To Build A Jet Engine

Building a jet engine is an engineering task of similar difficulty to building a leading edge semiconductor fab (TSMC), a wide body airplane (Boeing), or a reusable rocket (SpaceX). One challenge is compressor efficiency at higher speeds: over 50% of the generated energy goes to compress incoming air (the "back work ratio”), and adding efficiency requires more compressor stages that add weight until the thrust gains are offset. Another challenge is finding exotic materials that handle higher temperatures and pressures (titanium was already being replaced by exotic superalloys in the early 1960s), and measuring how those novel materials behave in high temperature and pressure environments for many years. The physics is hard, but the economics are even harder: “It’s not that building a jet engine that “works” is so difficult or expensive […] The difficulty is building an engine that meets performance targets—thrust, fuel consumption, maintenance costs, [etc.].There’s no point in designing a new engine if it doesn’t significantly improve on the state of the art. [That] means engine development [must push] technological boundaries: higher compression ratios, hotter temperatures, lighter weight, larger fans, etc.”

(4) Panthera Resources - Multi-Bagger Legal Claim?

Panthera Resources is a UK-listed nanocap ($24M GBP market cap) with a $1B+ litigation claim against India for the expropriation of a gold mine. Many bilateral trade treaties include investor-state dispute settlement (ISDS) clauses that facilitate binding arbitration of legal claims outside of corrupt local courts. Panthera, and its litigation finance backers, are using this method to get some value from its long-dormant legal claim. The story has lessons for (1) creative methods of litigation finance; (2) potential market inefficiencies in pricing inactive litigation assets inside of smaller companies; and (3) the barbell (venture-like) nature of litigation finance, although with more probabilistic precision. That said, “anyone interested in Panthera Resources should keep the old German saying: "On high seas and in courts of law, you are in God's hands" in mind.”

(5) A16Z and Benchmark-Backed 11x Has Been Claiming Customers It Doesn’t Have

11x is (was?) a hot AI startup that raised over $80 million in the past two years to create “AI sales reps.” In practice, their biggest innovation was a novel approach to revenue accounting. 11x signed new customers to an annual contract (say, for $1,000) with 3-month free trials, then booked $1,000 in monthly revenue, which investors annualized into “ARR.” In other words, cash accounting led to annualizing already-annual contracts. Many such cases! Even worse, most free trial customers cancelled before making a single payment because “their products barely work.” Yet investors are still defending them? Sometimes, the real product is the stock...

(6) Something Is Rotten in the State of Cupertino

Brutal criticism of Apple’s sclerotic culture in the context of its AI feature delays: “[Siri’s new AI features are] vaporware. These were features Apple said existed, which they claimed would be shipping in the next year. [The] fiasco is not that Apple is late on AI, [but] that Apple pitched a story that wasn’t true, one that some people within the company surely understood wasn’t true, and they set the course based on that. […] Tim Cook should have already held a meeting to […] rectify this Siri and Apple Intelligence debacle. If such a meeting hasn’t yet occurred or doesn’t happen soon, then, I fear, that’s all she wrote. The ride is over. When mediocrity, excuses, and bullshit take root, they take over. A culture of excellence, accountability, and integrity cannot abide [accepting] those things, and will collapse upon itself with the acceptance of all three.”

(7) Tyler Cowen, the Man Who Wants to Know Everything

“Cowen is famous not for a single theory but for the broad scope of his intellect. Put simply, he seems to know something about everything: machine learning, Icelandic sagas and where to eat in Bergen, Norway. “You can have specific and detailed discussions with him about 17th-century Irish economic thinkers, or trends in African music, or the history of nominal GDP targeting,” said Patrick Collison, co-founder of Stripe […] Cowen’s emotional life remains a mystery. He told me he did not experience regret. “I don’t know what the function of it is,” he said. “Is it to signal thoughtfulness? To stop you making further mistakes? It’s like revenge. I don’t understand it.” [He] also said he didn’t understand envy or anger. He didn’t know what he should be envious of. He didn’t get lonely, by himself or in company. He actually said, “Why bother?”” Teutonic energy, big time.

(8) The Race Is On To Build The World’s Most Complex Machine

The explanation of ASML’s EUV lithography machines never gets old: “ASML’s most advanced machine is mind-boggling. It works by firing 50,000 droplets of molten tin in a vacuum chamber. Each droplet takes a double hit—first from a weak laser pulse that flattens it into a tiny pancake, then from a powerful laser that vaporizes it [and] turns each droplet into hot plasma, reaching nearly 220,000°C, roughly 40x hotter than the surface of the Sun, and emits light of extremely short wavelength (extreme ultraviolet, or EUV). This light is then reflected by a series of mirrors so smooth that imperfections are measured in trillionths of a meter. The mirrors focus the light onto a mask or template that contains blueprints of the chip’s circuits. Finally the rays bounce from the mask on a silicon wafer coated with light-sensitive chemicals, imprinting the design.”

(9) Strategy Announces Proposed STRF Perpetual Preferred Stock Offering

“Strategy” (formerly MicroStrategy) is a bitcoin holding company that is getting innovative with preferred stock. I’ve never seen anything like this before… They are issuing perpetual preferred stock (not perpetual strike, but perpetual strife… never heard of that before) with paid-in-kind (PIK) features. The preferred stock cannot convert to common stock, it cannot be redeemed, and (at Strategy’s election) it could also forego payment of the cash dividend in perpetuity. In other words, they are issuing a security that will likely never return a dollar to investors. On the other hand, that could also describe bitcoin and most startups? Sounds lucrative for the bankers.

(10) The Horrors of Shared Docs

Quality content from the Economist: “Shared docs tend to get shared, and they frequently end up in the browsers of total strangers. You don’t know why they are in it, and they probably don’t either. Some people may not be identifiable at all: they have usernames like “anonymous tapir”. They might be the boss; they might be one of the interns. They are probably not a tapir. When enough people are involved, chaos ensues. Someone adds a paragraph, causing someone else to make alterations in the wrong place. [One] anonymous user spends so long on rewriting a sentence that people start to wonder whether it may actually be a tapir after all. [Countless] emails arrive to tell you that there has been “activity” in the document. [One] person is always there, day and night, but never does anything.” For even more horror, try using Notion.

(0) Miscellaneous

Feudalism as a contested concept. Hilarious review of Meghan Markle’s new cooking show. The political implications of Manus AI. Making a better AI accounting system. Why are so many unusual skyscrapers being built in Tirana, Albania? The importance of repetition in the workplace. Some theories on the OpenAI non-profit buyout. Data integration at the IRS. Using Grok’s voice mode as a reading companion. Rippling’s corporate counter-espionage. Consciousness in babies. Booming market for second hand clothing. “I wonder if the fall of Rome was this stupid.” What is an AppLovin?

(1) 50 Years of Travel Tips

People travel to retreat (vacation) or to engage (experience and education). This is advice for the latter kind of traveler: “(1) organize travel around passions instead of destinations; (2) crash a wedding (you will be obliged to dance); (3) when visiting a [new] city, get a street food walking tour; (4) go to a cemetery […] and see how it's different and the same all at once; (5) enjoyment of a trip will be inversely related to the weight of your luggage; (6) minimize the amount of time spent in transit […] it’s far better to spend more time in a few places than a little time in lots of places; (7) for a memorable trip, go without reservations; (8) visit the [local] McDonald’s;” and more! For related tips, check out Rediscovering Travel.

(2) How Coal Really Won

Some original research on the takeoff of coal for home heating at the dawn of the Industrial Revolution. The prevailing theory is that coal first gained popularity as a replacement for wood in London in the late 1500s because deforestation increased the price of wood. But this doesn’t mesh with the fact that coal was cheaper than wood for most of this period. This article argues that coal took off due to a “wood-saving furnace” developed in Germany for beer-makers, which had intended to reduce wood consumption by having a more efficient flue for heat re-circulation. This flue came with the beneficial effect of reducing noxious gas penetration into the beers, and this made cheap coal useful for brewing. Plus, details on the development of chimneys.

(3) Deep Research and Knowledge Value

OpenAI’s Deep Research feature in ChatGPT has been, for many, the moment “when they felt the AGI.” Because Deep Research seems to get bogged down by SEO slop on the internet (at least for now), it paradoxically produces the best results for the most obscure info. If AI yields the best results for obscure information, then finding obscure public info is no longer an edge: “[Deep Research makes clear] how much future economic value is wrapped up in information not being public.” In part, it’s an argument for the decline of fundamental research for public market investors: “it's going to be more difficult to harvest alpha by reading endless financial filings when an AI can do that research in a fraction of the time.” It’s also an argument for the value of prediction markets as a financial counterweight to incentivize exposing information to the public, at least through the price mechanism: “AI's capability of knowing everything that is public is going to increase the incentive to keep things secret; prediction markets in everything will provide a profit incentive for knowledge to be disseminated, by price if nothing else.”

(4) Interview with Daniel Gross and Nat Friedman About Models, Margins, and Moats

Startups often make a Faustian bargain with publicity. The media attention feels nice—it builds momentum with investors, hires, and team morale—but comes with the side effect of attracting competition. AI companies are learning this the hard way: “There are theories of how you create moats for yourself and generate margins and seven powers and network effects [and] cornered resources. But [almost] anything can be competed away[.] What matters is how much attention your industry gets—if you're just on the front page of every news publication every single day [… especially] in software [where the barriers to entry are] just devilishly low to the point where a random company in China is just able to go for gold […] your margin will just erode over time.” Growth alone doesn’t create shareholder value. More examples: Engines That Move Markets.

(5) Can AI Unlock the Secrets of the Ancient World?

The Herculaneum papyri is a collection of thousands of 2,000-year-old scrolls buried during the eruption of Mount Vesuvius. The collection contains thousands of likely lost documents, but the charring from Vesuvius made the scrolls unreadable and stuck together in a “charred log” form… until now. How does it work? The burnt scrolls are scanned in a particle accelerator, then “unrolled” by simulating how the pages would have originally looked: “[their software can] detect changes in density between the burnt manuscript and the burnt ink layered onto it […]. As they build more training data from each scroll, “unrolling” gets easier. AI can do more than automate B2B SaaS!

(6) Airships May Finally Prove Useful For Transporting Cargo

Cargo airships (think: Zeppelins) fell out of favor in the 1930s due to catastrophic accidents (infamously, the Hindenburg) and other transportation improving its cost efficiency. Airships’ inefficiency stems from variable buoyancy—it is hard to managing changing weights when loads are picked up and dropped off. But some recent innovations show promise: (1) water ballast; (2) helium compression; (3) airframe designs that improve lift using pressure differentials (similar to airplane wings); and (4) replacing expensive helium with cheap hydrogen in fireproof dirigibles. FAA notwithstanding, airships could become the biggest transport revolution since jet airliners.

(7) The Afterlife of the Roman Republic

“The way to understand the fundamental change in the structure of Roman government [during the principate] is that the [former] Roman Republic [was] not a democracy, but fundamentally a partnership of the Roman Senate (representing the aristocracy) and the People. Indeed, that is what the appellation SPQR—senatus populusque Romanus (“the Roman Senate and People”)—implies: the state [was] composed of two roughly equal entities, the People and the Senate. In contrast, [the principate] is best understood as a partnership between the Roman aristocracy, still represented by the Senate [and] emperor, with the People excluded. The Senate [is kept around as a] junior partner [since] the emperor needs the Senate in order to run the empire.”

(8) U Thant Island: The Quirky Story of NYC’s Smallest Island

“Situated halfway between the United Nations complex and Long Island City is a tiny, 1/2 acre island that appeared in the East River in the 1890s. U Thant Island, as it is known, is nicknamed after a former United Nations Secretary General from Burma. […] The island was created from the materials excavated by piano manufacturer William Steinway during an attempt to build underwater trolley tunnels in the East River from Manhattan to his company town in Astoria.” Officially, U Thant Island is a bird sanctuary and off-limits for visitors. Unofficially… kayaks.

(9) How Many New Yorkers are Secretly Subsidized By Their Parents?

Years ago, I had to move out of my apartment when the rents went up >100% on lease renewal. While it was annoying to be on the receiving end of gentrification, I was also curious about the apparently unemployed 22 years olds replacing long-time tenants. Dynastic wealth was suspected. Suspicions have been confirmed: “Someone who isn’t generationally wealthy probably cannot buy a property in New York City. In seven years of working in the industry, [he] has yet to see a first-time homebuyer who can close on a property without parents’ help […] Only about one in three adults under age 43 manage to support themselves [in NYC] without any financial help from their parents.” Here’s Part II.

(10) Only Fools Think Elon is Incompetent

“As an industrialist, Elon is unmatched by any American in the country’s history […] any of us could spend a lifetime and a burn trillion dollars and end up with [nothing] resembling Musk’s high-tech industrial behemoths. Even with zero institutional constraints in our way, we would fail to identify the best managers and engineers, [we’d] often fail to convince them to come work for us, [and] we might not be able to inspire them to work incredibly hard […] We’d also fail to elevate and promote the best workers and give them more authority and responsibilities. We’d fail to raise tens of billions of dollars at favorable rates to fund our companies [and] we’d fail to negotiate government contracts and create buzz for consumer products. […] Maybe saying that Elon has a 110 IQ makes you feel like you beat him in your online fantasy world, but out there in the actual world, he is ripping up national institutions at breakneck speed. […] Elon Musk is [the] single most capable man in America, and we deny that fact at our peril.”

(0) Miscellaneous

The quiet acquisition of the Minneapolis Grain Exchange. Background on secretive Miami Stock Exchange. Distribution of federal government employees. Soon lab-grown diamond rings will soon be $10-15 per carat. Online scams are a $500 billion industry. Trump’s plan to Make Gaza Great Again. Investing in Guyana’s Rupununi Ranch. 22 hour trading day is coming to NYSE. Only 5% of Americans use AI. Conclusion of the Minnesota Timberwolves ownership dispute. Experience is incompressible. Insanity of the health insurance appeals process. Valentine’s Day flowers are mostly imported via Miami Airport. Startup founders and business intuition. Zombie unicorns. A niche clearinghouse for eggs. Paradoxes and God. The Berkshire Hathaway 2024 annual report. Palace intrigue in Uzbekistan. The reclusive Italian billionaire Andrea Pignataro. Axiomatic analysis of the papal conclave.

(1) Capital, AGI, and Human Ambition

"Labour-replacing AI will shift the relative importance of human vs. non-human factors of production, which reduces incentives for society to care about humans while [entrenching] existing powers [...] the key economic effect of AI is that it makes capital [a] general substitute for labour. There is less need to pay humans for their time to perform work, because you can replace that with capital (e.g. datacenters running software replaces a human doing mental labour)." AI solves the two main constraints on capital substitution: implicit knowledge process automation and a scarcity of top talent. If all "intelligent" labor can be replaced by AI, then "the only [remaining] forms of human ambition [will be] playing local social and political games within your social network and class [...] AIs are better poets, artists, philosophers, everything; why would anyone care what some human does, unless that human is someone they personally know? [...] With AI, we could have the last and greatest opportunity for human ambition—followed shortly by its extinction for all of time. How can your reaction not be: "carpe diem"?"

(2) Training AI Models Might Not Need Enormous Datacenters

The Economist was ahead of the times with this article, which is especially relevant in the context of recent news on DeepSeek… Backpropagation is a method of training AI models that adjusts model weights based on the difference between predicted and actual text. Large datacenters are currently required for training frontier models because of "checkpointing" that synchronizes multiple training flows. But recent innovations in checkpointing, notably DiLoCo, are enabling training models in a distributed fashion by bundling the backpropagation steps in a way that reduces checkpointing frequency. The upshot is that, in the long run, these methods could reduce model training costs, enable training in a distributed fashion (iPhones could, in theory, be used as distributed clusters), and reduce current reliance on monolithic datacenters. I'm sure I butchered some of the nuances; this stuff is complicated. But if you're into AI, it’s worth reading the article. If you're not into AI, good luck.

(3) DeepSeek FAQ

This is a good (and technical) overview of why DeepSeek is in the news. They’ve made several innovations in AI development with cheaper training and inference costs (e.g., “mixture of experts” RAG to focus on subsets of the training data, inter-chip memory bandwidth optimization, DeepSeekMLA for more efficient memory usage within the context window, pure reinforcement learning with model training, etc.) that produced a frontier model at a fraction of the cost. The news is moving fast on everything in AI, but the more general implications are (1) AI capex might be going through the floor; (2) it’s even more unclear if the underlying models have competitive advantages; and (3) the cold war with China is heating up.

(4) Strategy Letter: Let Me Go Back!

A contrarian view on switching costs in the context of Excel beating Lotus 123 (from June 2000), which makes a useful distinction between moats around legacy and incremental users, and has a number of lessons for consumer and enterprise software companies… “The mature approach to strategy is not to try to force things on potential customers. If someone isn’t even your customer yet, trying to lock them in just isn’t a good idea. When you have 100% market share, then come talk to me about lock in. Until then, if you try to lock them in, it’s too early, and if any customer catches you in the act, you will just wind up locking them out. Nobody will want to switch to a product that is going to eliminate their freedom in the future.”

(5) Bill Gates: I Coded While I Hiked as a Teenager. Was I on the Spectrum? Probably.

Excerpt from the new Bill Gates biography on a hiking trip, writing BASIC formula evaluators, and positive reflection on the influence of his parents in his life: “If I were growing up today, I probably would be diagnosed on the autism spectrum. [My] parents had no guideposts or textbooks to help them grasp why their son became so obsessed with certain projects, missed social cues and could be rude [and not seem to] notice his effect on others. [They] gave me room to grow emotionally, and created opportunities for me to develop social skills. Instead of allowing me to turn inward, they pushed me out into the world—to the baseball team, the Cub Scouts and other families’ dinner tables. They gave me constant exposure to adults, immersing me in the language and ideas of their friends and colleagues, which fed my curiosity about the world beyond school. Even with their influence, my social side was slow to develop, as was my awareness of the impact I can have on other people. But that has come with age, with experience, with children, and I’m better for it.”

(6) Treading Alert: Greenland LLP

Trump loves doing deals… “We revisit the investment case for Greenland [...] the territory is in long-term decline. It has shrunk for 28 consecutive years, decreasing in size over the period by 11,000 square miles. Earnings growth since 2015 has been reliant on fishing [...while] Greenland has significant LBO potential, carrying virtually no gross debt, elevated leverage at subsidiaries [and] labor shortages [have caused its] maintenance expenditure [to trend] below replacement rates. Given the scale of these challenges and the illiquidity of its citizenship, M&A offers value realization for Greenland’s 56,000 independent stakeholders, who can effect a change of control via a simple majority vote. [...] Greenland is uniquely positioned [to] benefit from acceleration of cyclical climate change [and] near-term deregulation of fossil fuel extraction presents a virtuous circle: put simply, more drilling means more to drill."

(7) Lorne Michaels Is the Real Star of “Saturday Night Live”

SNL is celebrating its 50th birthday, and The New Yorker has a profile on the man behind it all: Lorne Michaels. SNL is the single most influential force in comedy—it established the careers of Chevy Chase, Bill Murray, Al Franken, Adam Sandler, Chris Rock, and basically every famous comedian of the last half-century. Lorne is the man behind the scenes pulling the strings, and while it sounds like working for him is traumatic, his methods got results: “[Lorne has] imbued “S.N.L.” with a clockwork stability. Every week, the sketches are written, the index cards shuffled, the vases of flowers replenished. New employees are still routinely terrified of him; when the office feels too friendly, whispers circulate that Michaels doesn’t like “the tree-house vibe.” Will Ferrell thinks that Michaels’s emotional withholding is part of a baseball-derived management style. “Baseball players keep the highs not too high and the lows not too low,” he said.”

(8) What “The Power Broker” Gets Wrong About Robert Moses and Ambition

Does Robert Moses embody the spirit of “building” things that is resurgent in the US? Maybe! “Others have over-learned the lessons of the “The Power Broker” and [over-simplified the story of] a deeply complex, unsettling and extraordinarily accomplished historical figure. […We] have lost faith in government and failed to comprehend that some of the Robert Moses spirit must be recaptured today if the United States is going to be a great builder again. As megalomaniacal as Moses could be, he showed that the public sector could be a force for civic improvement in America, delivering monumental public works that could stand for a century and beyond.”

(9) The Domesday Book

“In the 11th century, the King of England, William the Conqueror, [commissioned a survey] to measure the capacity of his kingdom to pay taxes. The product was the Domesday Book. […] Specifically, the Domesday Book was used to collect a tax called the geld, levied annually on landholders. The tax [originated] in the Danegeld, a tax used to pay tribute to the Vikings in exchange for pausing their pillaging. This tax was levied on all subjects, the first direct tax of this breadth in Europe. The new geld was based on a landholder’s ability to pay, [which was] based on the value of a landholder’s estates, [which was imputed from] their annual income.”

(10) Sam Altman on ChatGPT’s First Two Years, Elon Musk, and AI Under Trump

An in-depth interview with the founder of OpenAI, Sam Altman. It’s surprisingly candid, and some details stand out: (1) it has clearly been stressful and they're flying by the seat of their pants; (2) they’re clearly focused on being research-first (because winning seems to require attracting the best talent in AI research), (3) the consumer-oriented aspects of the business, notably with ChatGPT, came about mostly as a surprising accidental success, (4) it’s crazy to think about how much has happened over the past two years, and (5) they should have used Superwall to optimize their pricing: “We launched ChatGPT for free, and then people started using it a lot, and we had to have some way to pay for it. [We] tested two prices, $20 and $42. People thought $42 was a little too much. They were happy to pay $20. We picked $20. It was late December of 2022. [This] was not a rigorous “hire someone and do a pricing study” thing.”

(0) Miscellaneous

Social media and declining birthrates. Estimating the valuation of Greenland. The rise and fall of corporate bonds. Melatonin reduces jet lag. The origins of non-alcoholic beer brand Athletic Brewing. The growth of the UFC. VCs love accounting rollups. The early history of financial markets data. Detailed guide to the QSBS tax break. The problems with being data-driven. Confusing opacity of modern gender norms. Buying discounted assets in Russia. It’s easier to imagine the end of the world than the end of capitalism. Getting tricked by an AI deepfake of Brad Pitt. Personal library design inspiration. Patrick Mahomes “has turned the Super Bowl into his personal man cave.” Trump’s immigration fight with Colombia. Trump’s immigration team is pulling every lever in the administrative state.