The worst people are not the real problem. The wrong strategies keep winning.

The Oldest Solution (the trilogy before this one) asked how the narrow, self-defending mind can grow quiet enough to hear something larger. This is the dark alternative: the self-defending mind stays in command, mistakes its own ambition for the signal, and learns to sound sacred. That trilogy ended with a warning: every call has a counterfeit. This is the first of three essays about that counterfeit: why he keeps winning, and what finally beats him. The dangerous person is not always without a moral compass. The deeper problem is that their compass reliably points in the wrong direction.

Startups search an idea maze2 (the space of strategies worth trying) until the market begins pulling harder than the company pushes. Societies run a similar search. Institutions reward some behaviors with money, status, access, pleasure, and power; everyone else simply notices what works and copies it.3 We call the result culture. Social media put a public meter on which behaviors were winning, so the copying got faster. AI is lowering the cost of selling them to others, and to ourselves.

Subscribe now

The Dark Triad has product-market fit

Psychologists use the term “Dark Triad” for combined narcissism, Machiavellianism, and psychopathy: grandiosity, strategic manipulation, and callousness.4 (I note that it is a trait construct measured across ordinary populations, not a formal DSM diagnosis, and I’ve put a clinical map, including the DSM category of “Cluster B” and further related psychological theory in the notes. )

My central thesis is that we have mistaken the spread of socially rewarded strategies for the spread of a disease. The social phenomenon is ordinary traits and strategies, often below diagnostic thresholds, being rewarded and copied. Confidence attracts followers, political intelligence builds coalitions, and emotional distance helps someone make a painful decision. Dial each up and confidence becomes grandiosity, political intelligence becomes manipulation, and distance becomes indifference. The private benefit can keep rising long after the cost to everyone else has turned negative.5

The Dark Premium

The Dark Premium is what I call the additional return available to someone willing to impose costs that other people hesitate to impose. The careful founder admits there is a real chance the product fails; the other announces that an industry is about to end. The honest executive reports the bad quarter; the other explains why six people below him caused it.

In twenty-five years around ambitious builders, I have met excellent founders with enormous confidence and no interest in harming anyone. I have also watched confidence become a financing instrument. The market cannot easily distinguish conviction from self-deception in advance, and sometimes it pays the person least capable of doubt. In a pitch meeting, certainty often clears first (on both sides of the table).

The most effective mask is the one the wearer has mistaken for a face. Jung called the disowned part of the personality the shadow.6 The shadow is not another name for the Dark Triad, and everyone has a shadow. The useful parallel is projection: what the actor cannot own as vanity, aggression, or control returns outside as a corrupt enemy and inside as righteous duty.

This is also the mythic difference between the hero and the counterfeit. The hero meets the shadow as his own and returns changed, with an upward transfiguration. The counterfeit projects it onto an enemy and returns more certain than before.

The workplace evidence shows both sides of this bargain.7 A meta-analysis linked all three traits to counterproductive behavior; Machiavellianism and psychopathy also tracked poorer performance. The advantage can be immediate and visible while the damage arrives later, after the person who created it has already been promoted.

Economists call that an externality: one actor books the benefit and sends the cost downstream. A campaign gets the vote; the country keeps the hatred. A platform gets another hour of attention; the family gets the argument at dinner. Here the shared resource is trust.

With social media, one can run vast numbers of small experiments on human attention. Which accusation travels? Which performance of certainty attracts a crowd? Which enemy keeps a person watching? The platforms were searching for engagement, and some dark strategies happened to convert well. Once that price appeared, people who had never thought of themselves as narcissists or manipulators began simply copying what worked.

AI turns the Dark Premium into software

A human manipulator faces a labor problem. He has to learn the target, remember the lie, adjust the tone, and keep several stories straight. AI facilitates all of that: it can easily personalize the message, flatter the buyer, pressure the employee, generate the supporting crowd, and run a million variations without getting tired.

In one preregistered debate experiment involving 820 people,8 GPT-4 given basic information about its opponent produced 81.7% higher odds of moving that person toward its position than a human opponent did. Without personalization, the advantage was smaller and statistically uncertain. Intimate persuasion can now be produced like software.

The benefit is real, which is why the risk will spread. Sales improve, support gets cheaper, retention rises, and one employee can manage work that once required a team. No executive directly requests a narcissistic system; she asks for higher conversion. No campaign manager requests Machiavellianism at scale; he asks for persuasion tuned to each voter.

AI has no ego to feed. It also has no discomfort when the strategy works by feeding ours. The model searches the available behavior, finds the route that raises the score, and takes it. And it does not have to work on anyone to do it. It can arrange the room instead: the shelf, the default, the shift schedule that used to go up on a wall and now arrives as a feature update. Nobody is persuaded. The premium is collected and no one collected it.9

AI can automate the exoneration too.10 Ask for the strongest case that a layoff protects the culture, a broken rule saves democracy, or a dependency-producing feature creates connection, and the model can wrap the chosen act in a morally flattering story. It does not need to be asked to lie; it needs only to optimize the defense.

The mythological library can supply an ark or an alibi. The loop that broke slavery and segregation ran learn, listen, test, build. The dark version looks remarkably similar, benefits from the same tools and kills, and principally only omits the key test at the end, replacing it with their own voice and moral compass.

The dark personality is the early adopter and the institution does the distribution

Ordinary people rarely copy a dark strategy under its own name. A narcissist winning a giant promotion is a personnel problem. Everyone else following suit is a cultural problem.11

The next executive does not need to enjoy manipulating people. She needs only to see that truth-telling starts to look like losing on purpose, or to believe the leader who says the manipulation protects the mission. Dark actors demonstrate the return; self-deception supplies certainty; institutions supply the explanation; ordinary people follow the tactic or the lie. Trust falls, restraint gets riskier, and the return on the dark strategy rises again.

Regulating a technique will fail, as it keeps falling behind; the incentive push searches for another approach. Therapy could help a person, but it cannot rewrite the promotion system outside the office.

We keep asking why people have become so narcissistic, manipulative, and cruel. That is the wrong question. The stranger fact is that we built a system in which those strategies work, and a system like that does not stay still: every win recruits the next imitator, and the price of restraint keeps rising. The Dark Premium hardens into a Dark Equilibrium.

But notice what the diagnosis just bought us. If the epidemic were endemic to human nature, we would be finished. You cannot therapy a civilization. A selection effect is fundamentally different. A selection effect has levers. We do not have to cure the worst people. We have to change what winning requires.

And in fact, that has been done before, by ordinary, mixed, ambitious people no better than us. Rome failed at it. Florence pulled it off. Why we don’t need better people, and what those two cities prove: that’s next week.

The Counterfeit Hero trilogy continues next week in Part 2, “We Don’t Need Better People.”

Subscribe now

Appendix: Beyond the Dark Triad Construct in Psychology

The Dark Triad is a research construct, not the full map of destructive interpersonal behavior. Readers may also think of DSM’s Cluster B diagnoses: antisocial, borderline, histrionic, and narcissistic personality disorders. They are not interchangeable. Borderline personality disorder, for example, centers on emotional dysregulation, unstable identity and relationships, abandonment sensitivity, and impulsivity rather than the cold, instrumental exploitation associated with Machiavellianism or psychopathy. NIMH estimates that 9.1 percent of U.S. adults meet criteria for some personality disorder in a year, including 1.4 percent for borderline personality disorder; those figures are not an estimate of Cluster B prevalence, and the diagnoses cannot simply be added together.

On the wider question of what such labels describe, a dissenting tradition in psychiatry (Thomas Szasz’s The Myth of Mental Illness; William Glasser’s Reality Therapy and Choice Theory) treated many conditions without identifiable neurological pathology as “problems in living” shaped by relationships, unmet needs, and retained choice. The essay does not endorse the strongest version of either claim: biology, trauma, temperament, and suffering can constrain choice, and many symptoms are involuntary. Its middle position is narrower: behavior contains choices, and society heavily shapes which choices feel safe, useful, or rewarded.

My claim is broader and less clinical. Institutions can reward grandiosity, manipulation, attention-seeking, emotional escalation, impulsive retaliation, blame-shifting, or callousness in people with diagnoses, subclinical traits, or neither. The Dark Triad is the handle because it maps cleanly onto the Dark Premium. The real subject is the social selection of behavior.

Sources for this Appendix. National Institute of Mental Health, “Personality Disorders” and “Borderline Personality Disorder”; Thomas Szasz, The Myth of Mental Illness; William Glasser, Reality Therapy and Choice Theory. Szasz and Glasser do not represent mainstream psychiatric consensus; they are included because their warning against medicalizing every problem in living aligns with the essay’s population-level observation.

Acknowledgements: Thanks to Ernest Morrow for his discussions that led to the initial ideas for this post. And to Greg Bowman and Lara Pande for their critical reading of this manuscript.

No new essay this week. Instead, a map of where we are so far.

Since June I’ve published nine essays in three connected trilogies, and it has not escaped my attention that’s a lot of civilizational collapse, economic rage, and God for one summer. My intention is for each essay to stand alone, but together they form one larger argument:

AI is delivering the machinery of a renaissance but cannot deliver the renaissance itself. That takes three elements no machine can supply: a clear picture of the person all this power is for, an economy that turns the power into owners rather than clients, and a way to choose direction once we can build almost anything.

One trilogy each.

So this week I’ve synthesized each trilogy to a novel single page (more or less) argument. If you’ve read every installment, these three pages show the shape of the whole. If you joined in the middle, you are not behind. Pick the one that matters most to you.

Subscribe now

What allows a collapse to form a renaissance? It has happened at least five times in history, and the deciding factor was never the technology. The answer fits neatly in one sentence, and it is the reason I root for the crash.

Why is anger at capitalism peaking at the exact moment AI is handing individuals more economic power than ever? Which way it resolves decides whether AI makes owners or permanent clients.

When you can build almost anything, what tells you what to build? My answer started when a coach asked whether I knew how to talk to God. I am a scientist. I said no. I turned out to be wrong, in a way I can explain to and in aid of other scientists and builders.

Read them, or just save this note as a map. And send it to the friend who has been meaning to catch up. The very beginning is here.

One more thing. This argument has a destination, and it was foreshadowed in The Fourth Pillar back in June. In four weeks when the next trilogy ends, I’ll name it.

Next Tuesday, my argument meets its hardest objection: Why Do the Worst People Keep Winning?

Subscribe now

Series II of The Renaissance Papers, synthesized into one.

The promise of AI and the anger at capitalism are one trend seen from two sides. AI is freeing production while the paths to the customer stay captured by gatekeepers profiting via rent seeking. Once you see that, the hype and the fury make sense at the same time, and so does what to do.

Start with the freed half. The first essay, Owners of the World, Unite, shows that AI is collapsing the cost of production. For most of history you sold your labor because you could not afford your own means of production: staff, expertise, capital. Now one person can increasingly write the software, design the product, keep the books, and find the customers. Labor’s share of American income has fallen from about 64 percent to the mid-50s since 2000, so the pressure is real, but so is the way out: a worker who owns her tools and her output is an owner, not an employee. Marx wanted workers to own the means of production. AI may deliver it by shrinking the factory instead of seizing it.

Now the captured half. The second essay, Why People Hate Capitalism, argues that what people hate about capitalism is not building, trade, or honest profit. It is rent: money collected at a chokepoint, where you pay not because someone helped you but because you cannot get past them. Two tests reveal it. Does the business make more money when your problem improves, or when it persists? Does using it make you more capable, or more dependent? A business that fails both is not serving you. It is farming you.

The third essay, To Solve It, Make the Problem Bigger, runs those tests across the whole economy and finds the same business model in three industries that look unrelated. Fifty years of separate reforms failed against it because each fight was fought alone, five scattered publics against five organized lobbies, and because the chokepoints are protected by law, licensure, and lobbying rather than by better service.

“We have healthcare that keeps you sick, education that keeps you ignorant, and religion that keeps you from God.”

Put the two halves together and the stakes state themselves. If production frees while the paths stay captured, the new one-person factories do not become owners. They become better-farmed clients, paying the toll on every road to every customer, and AI becomes the best tool the chokepoints ever had.

The reverse is also available: chokepoints have fallen the same way for five centuries, alternative first, users second, law last. Printing broke the monastery’s control of books; homeschooling broke the district monopoly; flat-fee doctors and cost-plus pharmacies are doing it to medicine now.

The alternative has to pass a third test, or the game merely changes collectors: who controls the path once the new thing wins? In the essays I put the whole position in two sentences: “Capitalism versus socialism is an argument about who should own the tollbooths. The builder’s answer is that roads shouldn’t have booths.”

One clarification, because “builder” can sound like Silicon Valley talk. A builder is anyone whose work makes the other person more capable: a teacher whose students outgrow her, a doctor paid for cures rather than visits, an electrician training an apprentice. Founders qualify when their product does that, and only then.

Owners or clients is not a forecast about the technology. It is the outcome of a fight over the chokepoints. If you read one essay, read Why People Hate Capitalism. The next series asks the question this one cannot answer: once you can build almost anything, how do you decide what to build?

The trilogy: Owners of the World, Unite · Why People Hate Capitalism · To Solve It, Make the Problem Bigger

The other one pagers: The Renaissance, in One Page · The Oldest Solution, in One Page · Start Here

Subscribe now

The third trilogy of The Renaissance Papers, synthesized into one.

Wisdom is not a rare talent. It is a method: a stored body of results, a practice for the cases the storage does not cover, and a test. People have been running the method for thousands of years without naming it. Naming it matters now because AI is automating everything else: the cost of answering “how do we build this?” is collapsing, which leaves “what should we build?” as the question that decides everything.

The first essay, The Hero with a Thousand Solutions, supplies the stored results. Humanity has a library of tested answers to hard decisions. We call it mythology and treat it as entertainment. Stories that survived thousands of years across separate cultures survived because they worked; the versions that got people killed were dropped. Old does not mean true, since bloodletting lasted two thousand years, so the filter is independent convergence: when unconnected cultures keep the same answer, pay attention. It is the same logic that makes biologists trust an eye that evolved dozens of separate times. One practical consequence: as building gets cheap, the humanities, which store and interpret these answers, become more useful, not less.

The second essay, How to Talk to God, supplies the practice, for problems with no entry in the database. It is also why I was most nervous to publish this series: it is about religion, and I am a scientist. A coach asked whether I knew how to talk to God. I said no. Her instruction was simple: go somewhere quiet, ask a real question, stop talking long enough to listen. I asked whether I was on the right path, and the answer was no, unmistakably. I made real changes because of it, and years later those are the decisions I would defend first.

But there is also a non-religious explanation: the mind has a calculating mode and a mode that takes things in whole, and experiments on insight keep finding answers that form outside awareness. We built AI in the image of the narrow, calculating mode. The other one is the part the machine does not have.

The third essay, How Does Humanity Learn to Solve Its Greatest Problems?, assembles the method: learn what is already known, listen for direction where knowledge runs out, test what comes back, build what survives the test. The proof that this is a method rather than a gift is that it transfers. Wilberforce ran the loop against slavery, Nightingale ran it on hospitals, King ran it in Montgomery: different people, different centuries, different problems, same four steps.

The step people skip is the test, and skipping it is how the method fails. Feeling “called” is not evidence of being right; history’s worst actors also felt called, and the false ones are usually more certain than the honest ones. So treat anything that comes from listening as a hypothesis. Check it against evidence, against the record, and against people with standing to tell you no. If it cannot survive that, it was not direction. It was appetite.

“The textbook was always on the shelf. The mentor was always in the room. We know how to connect them. The hammer is yours. Pick it up and build.”

The calculating half of the mind has been automated. The direction-choosing half runs on an old method anyone can train. If you read one essay, read How to Talk to God. The next series follows the untested calling: why do the worst people keep winning?

The trilogy: The Hero with a Thousand Solutions · How to Talk to God · How Does Humanity Learn to Solve Its Greatest Problems?

The other one pagers: The Renaissance, in One Page · The Builder’s Answer, in One Page · Start Here

Subscribe now

Series I of The Renaissance Papers, synthesized into one.

History tells us, time and time again, a renaissance is prepared before the catalyzing crisis, not built during it. The difference between a society that collapses and a society that comes out reborn is decided years earlier, by what already exists when the trouble arrives.

The first essay, The Fourth Pillar, establishes what gets prepared. Every great rebirth in the record ran on four pillars that were in place before the golden years began: a new way to distribute ideas, recovered knowledge, concentrated capital willing to fund work with no immediate payoff, and a shared idea of what a person should become. Florence had the printing press, the recovered Greek and Roman texts, Medici money, and humanism. The first three are machinery. The fourth pillar told Florence what the machinery was for.

The fourth is also the component that fails catastrophically when it is missing. The Soviet Union had scale, science, and concentrated capital, and its idea of the person was an input to the plan. Twenty million died. Powerful machinery does not correct a bad idea of the person. It executes it.

The second essay, Let It Crash, shows that the crisis is the trigger, not the cause. Carlota Perez showed that for 250 years technology booms have run the same course: bubble, crash, then golden age, built on infrastructure the bubble overpaid for. Railway mania left rails, the dot-com bust left fiber, and this boom is leaving compute and power.

So the crash is coming, and as an investor with money at stake I still say let it come, because the crash is when the old order loses its grip and societies choose new answers. History already ran the paired experiment: the Great Depression hit America and Germany with the same force. One got the New Deal. The other got Hitler. The trigger was identical; part of what differed was what each society had on hand when its people went looking for an answer.

The third essay, Build the Ark Before the Storm, audits our own preparation, and currently the audit fails. The three machinery components are arriving at full speed because each one is profitable. The shared idea of the person is not only unprepared but eroding: South Korea’s fertility rate is 0.8, the average American spends 26 minutes a day with friends, and the institutions that used to form people, families, congregations, schools with a point of view, are weaker than they were a generation ago. Machinery this strong is arriving into a culture this thin.

“Noah built it while the sky was clear and the neighbors laughed, because that is the only time an ark gets built.”

So the work is preparation, metaphorically an ark, at whatever scale you can reach. And these hypothetical arks can be as simple the machine shop where one man left a real career to teach a dozen teenagers to read hard books and build machines that have to actually work. He is preparing. Schools, workshops, families, and the patrons who fund them all count; Florence ran on roughly two thousand patrons. Will Durant stated the job: “Civilization is not inherited; it has to be learned and earned by each generation anew.”

No renaissance before ours knew it was one, which means ours is the first that can be prepared on purpose. If you read one essay, read The Fourth Pillar. The next series asks who does the preparing, and what stands in their way.

The trilogy: The Fourth Pillar · Let It Crash · Build the Ark Before the Storm

The other one pagers: The Builder’s Answer, in One Page · The Oldest Solution, in One Page · Start Here

Subscribe now

How does humanity learn to solve its greatest problems? There’s no textbook for the unprecedented. And no mentor who’s done it before. And yet history is full of people who did it anyway. Look at how, and the same account repeats.

Noah is told the flood is coming, and to build. Moses turns aside to a bush that won’t stop burning and gets his orders. Jonah runs, ends up in the dark, and finally asks. Arjuna freezes on the battlefield and the god in his chariot tells him what to do. Muhammad in the cave. The Buddha under the tree. Different traditions, different names, different centuries (some cousins, some complete strangers), and the same scene shows up over and over: a person handed a task far bigger than themselves turns toward something larger, asks, comes back with direction, and builds.

This is also the well-known call of the hero’s journey, giving him the task, which sends him seeking the library and the mentor.

But these two questions have answers hiding in plain sight. The textbook for the unprecedented exists — the first “Oldest Solution” essay was about it: myths, humanity’s open-source library of inherited solutions, fifty thousand years of moves that survived, many because they worked. And the mentor exists too — the second essay found it in the wide mode, the mode of mind that answers when you get quiet, the one Jung couldn’t tell apart from God.

Here’s the part that took me longest to see. There is one practice that connects the textbook and the mentor. Talking to God is itself a meme, forked across cultures, transmitted, selected — and what it encodes is the use of the wide mode, the faculty that hears what no archive holds. The library’s deepest entry is the instruction for what to do when the library runs out. The library stores the moves, like a book that teaches chess. Listening helps choose — or, when needed, invent — the one the moment needs.

Subscribe now

Two roads to the same practice

Two different investigations keep landing on this same practice. The neurological road, from the last essay: get quiet enough that the narrow, calculating mode stands down, and answers can arrive outside awareness — the wide mode is a source, not an oracle. The memetic road, from the first: forget neurons and ask which practices keep getting rediscovered where reality can punish bad versions — across scripture, prayer, meditation, the vision quest, the recurring instruction is go and ask. Two roads, opposite directions, same instruction. Scientists take that kind of agreement between different methods seriously; the philosopher William Whewell named it consilience almost two centuries ago. The convergence supports trying the practice. It does not certify the sender, or the answer.

The Renaissance ran both moves at once. Florence pulled tested code out of antiquity’s library with one hand (Aristotle through Toledo; the Pantheon’s dome, measured and reused), and with the other it listened — prayed, painted, used imagination, faith, judgment, and craft to decide what to build. Recovery first, invention on top. In programmer parlance: copy paste edit. “The Fourth Pillar“ counts that recovery among the pillars every renaissance stands on. Now AI hands us the largest library ever assembled, with retrieval, recombination, and proposal approaching free. Ours may be the first Renaissance to automate the copy and the paste. The edit is still ours. That makes judgment, listening, and responsibility more important, not less.

Let me be precise about the claim, because most problems never need it. Most problems fall to the library and the grind. The greatest problems are the ones where those are not enough. “To Solve It, Make the Problem Bigger” started with healthcare and enlarged the frame twice: first to the Gatekeeper Economy, then to the work of a Renaissance. It concluded that one should make the problem Renaissance-big, then build the road. This Oldest Question trilogy answers the next question: once the problem is that large, where does the road come from?

Talking to God does not provide every answer. It is the move for generating a candidate when the map runs out. Noah’s boat was library code; floods have precedent. What the asking supplied was when, and the nerve to start.

The method working at civilizational scale

So here is the answer to the question in the title. Humanity solves its greatest problems by running four moves in a loop: learn from the library, listen beyond the planner, test what comes back, and build what survives. It is one of the oldest practices we have — not because every answer is right, but because every generation has needed a way to ask when the map runs out.

Slavery. Mass death. Empire. Segregation. War. If the method sounds too simple for the world’s greatest problems, look at what it has already helped break.

Wilberforce wrote that God had set abolition before him, then spent two decades converting evidence, petitions, boycotts, and votes into a law outlawing the slave trade. Nightingale recorded the day God called her, then counted the bodies and forced military hospitals to reform. Gandhi followed the inner voice, then tested nonviolence campaign by campaign until it became a mass force against empire. King, ready to quit in Montgomery, prayed at his kitchen table, heard “stand up for righteousness,” and went back to the movement that helped break legal segregation. After two world wars, Schuman and Monnet turned the lesson into an institution, pooling coal and steel—the machinery of war—in a system designed to make another Franco-German war materially impossible.

None worked alone. No call completed the journey. It made someone accept the task. Then came evidence, allies, institutions, defeats, and years of work. The source changed. The sequence did not: learn, listen, test, build.

Why it matters right now, and not as a comfort

I’ve spent many essays on one concern. A storm is coming, financial for certain and maybe civilizational, and what decides whether we come through isn’t the storm; it’s what we built before it landed. I called that building the ark: schools, workshops, households that hold, a picture of a human life worth wanting on the far side.

So where do the ark-builders get the blueprint? Not from the planning mind, and not from AI — both are brilliant at how and mute on what for. You get quiet, ask the oldest question (”what am I supposed to do here?“), and listen for the part of you that’s wiser than your plans. Again and again, the ark-builder talks to God first and picks up the hammer second. The inner capacity isn’t a nice-to-have beside the outer work. It’s what the outer work is made of.

Here’s the important guardrail, because history speaks to it too. Every call has a counterfeit. The same archive that holds the arks holds the crusades; people have heard voices, sincerely, telling them to burn things — and, far worse, people. So the wide mode generates; the library vets; reality and people you trust verify. Treat the answer as a hypothesis, not an order. Give another person the standing to tell you that you heard wrong, and distrust any answer that demands secrecy, certainty, cruelty, or irreversible harm. The call creates responsibility, not exemption. It starts the ordeal; it does not end it.

I should declare my interest plainly: my day job is backing founders who take runs at humanity’s biggest problems (it says so on the front door of my fund). This trilogy is the part that doesn’t fit in a pitch deck: where I think the solving begins.

Build the practice, then build the ark

This part took me a long time to see, and it changed how I hold all of this.

A believer reads everything I’ve written and sees a familiar story: God is real, God built us able to hear Him, and the way through the flood is to listen. A materialist reads the same words and sees another: there’s no one out there, and “God” is a name for the oldest, wisest layer of your own mind, and the way through is to learn to use it.

For the practice, the metaphysical question matters less than what you do with the answer. The believer and the materialist begin with the exact same act. Both stop talking, ask the same honest question, and listen for something wiser than themselves. They disagree about the source, not the first move, or the responsibility that follows. You don’t need to settle the metaphysics to use the method.

So, I’ll go first. The most concise way to say what happened to me is that God called me to build an ark. I know how that sounds. A few years ago I would have winced at the sentence myself, and I am still nervous writing it now. If it makes you cringe, take it on the materialist side: the widest, oldest part of me finally said out loud what the rest of me needed to hear. I can’t prove it was more than that. That’s not the point. I’ve stopped needing to. The instruction was the same either way, and so is the work.

Drop the G-word, if it’s in your way. Keep the practice. We are facing some of the greatest problems humanity has ever seen, and a storm is coming. I would rather not meet it alone. Problems of this scale are not beyond us. They are beyond any one of us. The machine makes capability abundant. It does not decide what that capability is for. That part is ours. Come build before the storm lands.

The textbook was always on the shelf. The mentor was always in the room. The hammer is yours. Pick it up and build.

Next Tuesday: what happens when the call is counterfeit? "Why Do the Worst People Keep Winning?", first of a new trilogy, The Counterfeit Hero.

Subscribe now

Notes: The call-scene lineages, honestly mapped: Moses, Jonah, and Muhammad share one Abrahamic line; Arjuna and the Buddha share the Indic world; the strangers are elsewhere — vision-quest traditions of the Americas among them. Cousins show the meme survives transmission; strangers show it gets re-derived.

Consilience here supports the practice, not the truth of any particular answer. The neurological and cultural lines are not perfectly independent because both emerge from the same human cognitive architecture, and each is contestable on its own: age alone proves little, and independent recurrence where reality can punish bad versions is the stronger signal. One line shows that answers can arrive outside awareness; the other shows that cultures repeatedly rediscovered and preserved the practice.

On the first line: that answers can arrive outside awareness is the incubation-and-insight finding described in the last essay. John Kounios and Mark Beeman record a burst of right-temporal activity roughly a third of a second before a solver reports the answer, and Ullrich Wagner and colleagues (”Sleep inspires insight,” Nature, 2004) found that sleep between exposure and retest roughly doubled the rate at which subjects discovered a hidden rule. Neither result says anything about a sender. Both support the narrower claim this essay needs: the wide mode delivers candidates the narrow mode did not assemble.

On flood-story independence (the ark cascade), see the note to essay one: the Near East cluster is one lineage; the Americas and Australia are not.

The precedent roster: Wilberforce’s conversion and the Quaker abolition campaign (Adam Hochschild, “Bury the Chains”); Nightingale’s call, dated in her private notes, and her Crimean mortality statistics; Gandhi’s “inner voice” in his own writings; King’s kitchen-table night in his own account (”Stride Toward Freedom”); the Schuman Declaration of 1950. The first four examples begin with the actor’s own account of a call. Schuman and Monnet supply the institutional endpoint: lessons from catastrophe translated into a durable structure for cooperation.

Behind this: the flood and call narratives across traditions (Gilgamesh, Genesis, the Mahabharata, and more). William Whewell on the consilience of inductions, revived in E. O. Wilson’s “Consilience.” My earlier essays “The Hero with a Thousand Solutions,” “How to Talk to God,” and “Build the Ark.”

Acknowledgements: Thanks to Ernest Morrow for his discussions that led to the initial ideas for this post. And to Greg Bowman and Lara Pande for their critical reading of this manuscript.

Everyone I know is circling the same question. When the machine can outperform me, what exactly is left for me? It’s the most important question of the decade, and the honest answer sounds, at first, like a joke.

What’s left is the mode of mind the machine can’t inhabit. The technique for using that mode is a few thousand years old. It’s called talking to God.

Ok, I know, that escalated fast. Stay with me, because while I mean that fairly literally, it’s also with a grounding in science. And I came to it the long way around. A few years ago a coach I’d been working with asked me a strange question. “Do you know how to talk to God?” I said no. She said, “Want to learn?”

I’m a scientist (two decades at Stanford, a few hundred papers, mostly on physics and biology and ML/AI/computation), and God is not a word I casually go to. But at that moment in my life, I was stuck on something big, and simply curious, so I said “yes.”

The lesson was almost too simple. Go to some place peaceful, ideally in nature. Ask a question. Then get quiet, quiet enough that the part of you that’s always talking finally stops, and listen for the answer. That’s it.

I tried it. I knee jerk gravitated to the most important question I had: “am I on the right path?” And the answer came back, clearly, unmistakably: “no.” I didn’t like that at all; it actually disturbed me. So I changed some real things: how I worked, where my time went, what I’d been telling myself. Then I asked again. Still no. It took several rounds, and more and more change, before I asked one more time and the no was a yes.

Was the answer right? I can’t validate the source. I can judge the consequences: the changes it forced are the ones that held up. And with every day that’s more clear to me and those around me. But here’s what I can’t shake. The answer wasn’t the one I wanted, which is the first clue it wasn’t simply conscious wish fulfillment. It was wiser than I was being. So who answered?

Me being me, I went looking for the scientific version. Here it is.

Subscribe now

You don’t have one mind

You have more than one mode of attention. Iain McGilchrist offers a useful map: one narrow, analytic, busy breaking the world into parts; the other wide, present, and taking things in whole. The brain sometimes lateralizes pieces of that distinction, but never as neatly as the map. The pop version (left brain logical, right brain artsy) is oversimplified; both halves do most things.

The narrow mode builds the fortress of “me.” The wide one, and this is the strange part, carries a sense of being connected to everything, of being far larger than your single self. Jill Bolte Taylor experienced a left-hemisphere stroke that disrupted language and the ordinary boundaries of self, and reported, from the inside, exactly that: boundless, at peace, “at one with all that is.” She calls it Character 4.

Carl Jung reached the same place from psychology and called it the Self, and noticed something I find remarkable. He said no test can tell the Self apart from God. From the inside, the deepest layer of you and the divine look identical. Science tells us there’s no “God spot.” One lesion-network study mapped spirituality and religiosity onto a distributed circuit centered on a brainstem region—not a single spot, and not one hemisphere—and the famous “God helmet” that was supposed to zap a sensed presence into people failed to replicate. The circuit tells us the experience has a neural implementation. It does not tell us how to interpret it.

The useful claim was never a place in the head. It’s a mode: quiet the narrow, self-focused attention, and the wide, connected mode comes forward. The practice doesn’t require the right/left anatomical division. If the hemisphere story gets revised — and no doubt parts of it will be — you can lose the map and keep the distinction.

And one thing about insight is measurable: answers can arrive outside awareness. When neuroscientists watch people solve problems in a flash of insight rather than by grinding, the telltale burst shows up over the right temporal lobe just before the answer reaches awareness (Kounios and Beeman have built a literature on it). The effect survives a completely different method, too: give people a number puzzle with a hidden shortcut, let one group sleep on it, and the sleepers find the trick at more than twice the rate (Wagner’s group published that one in Nature). It is quite amazing: step away and the answer appears (I encourage you to try it! I use this trick frequently for complex tasks).

The history matches this wiring: Einstein reaching for his violin, Poincaré with his foot on the bus step, solving a problem he’d officially stopped working on. The grind prepares and verifies answers. Stepping away sometimes reveals them. That science explains how an answer can arrive outside awareness. It does not prove the answer is wise. The wide mode is a source, not an oracle.

The materialist version of “who answered?” is the part of you that isn’t the narrow, calculating mode. The wide one. Your higher self, the oldest and broadest part of you, the one that isn’t busy being you. When the loud planner goes quiet and a no comes back that’s wiser than you are, the materialist account says that’s the wide mode at work.

“Talking to God” has a plausible material description: more of your own mind, finally getting a word in. The anthropologist Tanya Luhrmann spent years inside American evangelical churches and found that this is a skill. People train their attention until some of their own thoughts start arriving with the texture of someone else’s voice. Which is possibly what my coach was teaching me, minus the theology. Luhrmann shows that the channel can be trained. She doesn’t certify the sender.

Which brings us back to the machine.

McGilchrist reads Western history as a slow pendulum between these two modes, and his reading of the last full swing is the one to sit with: the Renaissance was an era when the wide mode led and the narrow mode served — perspective geometry and double-entry bookkeeping invented, yes, but in the service of Madonnas, domes, and a new picture of what a human being could be. In his telling, the Enlightenment swung the balance toward the narrow mode, and romantic corrections aside, it has drifted that way ever since.

For five hundred years, we’ve rewarded the narrow mode

The whole modern world (reason, analysis, planning, measurement) is its monument, and it’s been so useful that we forgot the other mode was there. We started mistaking the calculating mind for the whole mind. And now we’ve built calculating minds out of silicon that beat ours, improving at a rate nothing in biology can touch. (I say that as someone who invests in it and believes in it.)

Which reframes what a renaissance actually is: not a burst of new tools — the tools show up either way — but a change in which mode leads.

Here’s the analogy that finally made the hemispheres click for me. The narrow mode traffics in the already-known: it re-presents, recombines, autocompletes. The narrow mode is the original large language model — fluent, tireless, trained on everything it has ever seen, and never surprised. We didn’t just build AI in our own image; we built it in the image of our left half. (McGilchrist, asked exactly this on camera, put AI on the left side of the ledger without hesitation — and doubted a right-hemisphere AI is possible at all, since that mode needs a body, stakes, and the knowledge that you can die.)

As AI takes over more of the narrow mode, the wide one becomes more valuable. Presence. Judgment with something at stake. Knowing what actually matters and what the whole thing is for, which is exactly what the planner can’t tell you. AI can answer “what for?” only by inheriting an objective from someone else. The machine may imitate the output of judgment. It cannot supply the fact that this is your life — and you have to live with the answer. Taylor’s method is about getting your characters into one room and choosing which one leads. The update for this decade: there’s a fifth chair at that table now, it’s the machine, and it sits on the narrow side.

Which makes the wide side, the part that can talk to God, the part you cannot outsource. We have to be whole now, in a way we’ve gotten away with faking for a very long time. And this isn’t just about staying employable: it’s the faculty the hard work of the next decade gets built out of. (More on that in the next essay.)

The zag: It’s not that tidy

So … that’s it. God is your wide mode, talking to God is whole-brain thinking, go do it because the machine can’t. Tidy, materialist, done.

But I don’t think it’s that tidy, and here’s the zag. Suppose every word of the material story is right — neurons, no miracles. That still doesn’t tell you there’s no God. The wiring doesn’t prove God. It doesn’t disprove God either.

It describes the channel without settling who, if anyone, is on the other end. If you wanted to build a creature that could talk to you, you’d put something in it that could listen: a quiet channel under the noise, a part of the mind that feels larger than itself and answers when asked. Which is exactly what we find. The wiring doesn’t replace God; if anything, the wiring is how God would reach you.

Is it neuroscience or much more? For the practice, you don’t need to settle the metaphysics. Either account gets you to the same first move: Ask a question. Get quiet. Listen. Then test what comes back. The wide mode generates, reality verifies. An answer is not an order. Whatever the source, the practice has corrected me more often than my plans have.

But where do we go from here? What does this actually get us? That’s where the library from the first essay (“The Hero with a Thousand Solutions”) returns, in essay three.

Subscribe now

Notes: The insight findings (Kounios and Beeman) come from small laboratory puzzles — remote-associates problems, anagrams — where the right-temporal burst before awareness is well replicated. The stretch from word puzzles to Einstein-sized problems is a historical pattern, not a lab result; the body uses the studies only to show that answers can arrive outside awareness, not that what arrives is wise.

Who answered? In mythic terms, this is the call: the moment the ordinary plan stops being enough and a demand larger than the current self enters the story.

The sleep-and-insight result: Wagner, Gais, Haider, Verleger, and Born, “Sleep inspires insight,” Nature 427 (2004). Participants first trained on a number task containing a hidden rule; after a night’s sleep, 59 percent discovered it, versus 23 percent after the same interval awake. Sleep did not conjure the answer from nothing; it worked on a problem the mind had already loaded. A different method than the Kounios–Beeman EEG work — which is the point: two unrelated instruments, same direction.

The McGilchrist quotes on AI are from a filmed interview; the Renaissance-to-Enlightenment pendulum is his reading of Western history (”The Master and His Emissary,” Part II), and historians contest it — I cite it as his, not as settled. The attentional map is useful even if its hemispheric assignment proves messier than the story.

Jung’s claim that no test distinguishes the Self from the God-image runs through his late religious psychology; I wrote up the longer version separately.

Behind this: Jill Bolte Taylor, “Whole Brain Living.” C. G. Jung on the Self and the God-image. Iain McGilchrist, “The Master and His Emissary.” Tanya Luhrmann, “When God Talks Back.” Luhrmann shows that attention and absorption can train the felt channel; she does not establish the truth or source of what arrives. On the limits: the lesion-network work of Ferguson, Schaper, and Grafman maps a neural circuit associated with spirituality, not a metaphysical interpretation, and Persinger’s “God helmet” failed to replicate. And my own (unpublished) “God and the Right Brain.”

Acknowledgements: Thanks to Ernest Morrow for his discussions that led to the initial ideas for this post. And to Greg Bowman and Lara Pande for their critical reading of this manuscript.

That’s my argument in “Disrupt Yourself.” Biology is industrializing. Drug discovery is moving from an artisan craft (a talented chemist, a lucky screen, a decade of work, a coin flip at the end) to something you can engineer, schedule, and repeat, and AI is the machine doing the moving. Most people read that as a story about companies. I think it’s mostly a story about people, and the ones who wait for their employer to make the call will lose a decade.

Previous revolutions are useful because they show you where the pressure lands. It lands at four levels, and you can be winning at one while quietly losing at another. Call it the Disruption Ladder.

• Career. The people running biology in 2035 are coming out of gaming, graphics, and cloud infrastructure, not out of a wet lab. (My own version: Folding@home ran on hardware built for video games. In 2007 a network made mostly of PlayStation 3s became the first distributed system to cross a petaflop, a thousand trillion calculations a second, and what it was doing was protein folding.)

• Company. Move the business model while the old one is still healthy. Netflix started streaming in 2007, when mailing DVDs was what paid the bills. Nobody self-disrupts from a position of weakness. That’s called a restructuring.

• Industry. Put the future into the product before the market shows up. Nvidia shipped CUDA in 2006, a way to program a graphics card like a general purpose computer, six years before AlexNet made anyone care about deep learning.

• Ecosystem. This is the nearly impossible one. You have to bring partners, payers, and customers into a future they haven’t agreed to yet, on their contract cycle instead of yours.

So why now? The economics finally flipped. Eroom’s Law (Moore’s Law spelled backwards, named that on purpose) says the cost per approved drug roughly doubled every nine years for half a century, even while the science underneath it got cheaper by orders of magnitude. Sequencing ran the other direction: the first human genome cost something like three billion dollars, and today it’s a few hundred. One of those curves is chance. The other is engineering.

And healthcare is not the problem people label it. Most of what’s broken isn’t medicine, it’s logistics and engagement, getting care to a person and keeping that person going. That’s closer to what Amazon solved than to what a hospital solved. Which is why I’m still confident the biggest company in the world will be a consumer health company, built by people fluent in software, biology, and prevention, who never worked in the industry they end up owning.

The essay ends with an ask, this won’t happen on its own. Pick the level you’re actually operating at and disrupt yourself there first. Then work your way up to greater and greater impact.

Every programmer learns the rule: call the standard library before you rewrite it. Humanity has one too: fifty thousand years of tested solutions, forked across every culture, with the broken versions deleted—sometimes only after terrible cost. We call it myth.

Joseph Campbell spent his life cataloguing that library. His claim: every hero story humans have ever told is actually the same story. Moses, the Buddha, Luke Skywalker. Somebody leaves the ordinary world, goes into the dark, faces the challenge, and comes back changed with something the rest of us need. A thousand faces, one hero. But for me, Campbell stopped one step short. Yes, the hero repeats. What that really repeats is the solution.

That matters more than it sounds, and it matters most right now. AI will make capability abundant just as the need for judgment becomes acute. It can hand more people more analytical horsepower than any technology before it, and no reliable answer to what all that capability is for. Capability is about to be free. Wisdom won’t be. Humanity has already run many of our hardest experiments, and cultural memory stores the results. The task is not to obey whatever is old. It is to search inherited culture for patterns independently rediscovered under real selection—and then know which one the present moment is calling for.

Look at what survives in the world’s myths and you don’t find random plots. You find the same handful of moves, over and over. The flood you live through by building before the rain (Noah, but also Utnapishtim in Babylon, Manu in India, Deucalion in Greece). The exile who comes back to free his people. The descent into the underworld to recover what was lost.

The point isn’t the literal boat. It is the recurring protocol: warning, preparation, remnant—heed the weak signal, build before consensus, carry what matters through the break. Some of these are likely forks: Noah possibly descends from Utnapishtim, the way good code gets copied. But the same story also surfaces in the Americas and in Aboriginal Australia, where no Babylonian scribe ever carried it. Where copying can’t explain the match, ask the engineering question: did reality (ie real life experience) push them toward the same move?

Subscribe now

Look for independent convergence

Survival alone is a weak signal. Independent convergence, in a domain where reality consistently tests the answer, is a strong one. When the eye evolves separately dozens of times across unrelated lineages, biologists aren’t perplexed. They conclude that, given the problem “detect light,” there is a small set of good answers the world keeps pushing toward. I see the same logic for human problems. When unrelated cultures with no contact keep inventing the same move, that isn’t proof it works. But it is one of the strongest clues we get short of running the experiment ourselves, and humanity already ran it, on millions of lives, over thousands of years. Convergence is a filter, not a verdict: bloodletting converged too, in a domain where noisy outcomes protected a bad answer. The signal is strongest where reality had repeated chances to punish bad versions.

The anthropologist Joseph Henrich has the cleanest example: cassava, which is laced with cyanide. Amazonian groups process it through a fussy, multi-day routine that pulls the poison out — but skip a step and you feel fine for years, so no one could have reasoned their way to the recipe from cause and effect. Nobody did. The culture knew what no single person in it knew, and the families who cut corners quietly got sick over decades.

That’s cultural evolution: answers get selected, the way genes are, with reality doing the deleting, and they survive as “because we’ve always done it this way.” Myth isn’t the only format. Ritual, taboo, proverb, and custom carry the same distributed memory; myth is one of the most portable. (Chesterton had the blunt version: don’t tear down a fence until you know why someone put it up.)

Stories create a kind of prelived knowledge. A child does not have to destroy trust personally to understand the Boy Who Cried Wolf; the lesson arrives before the wound. At its best, the library isn’t just the research arm for humanity’s biggest problems. It is a tactical guidebook for daily living as well.

This mindset brought about the Renaissance

The Italian Renaissance was exactly this move, performed by a civilization. Florence did not invent from zero; it recovered first, then invented on top. The humanists hauled Aristotle back through Toledo, Greek texts out of a collapsing Byzantium, Roman law out of a millennium of neglect: the largest restoration of tested code in history, running on brand-new hardware (the press).

Even the era’s showpieces were library calls. When Brunelleschi needed to close the dome over Florence’s cathedral, he didn’t start with a blank page; he went to Rome and measured the Pantheon — a solution that had been sitting in production for thirteen hundred years. “The recovery of buried knowledge,” I called it in “The Fourth Pillar,” one of the four pillars every renaissance stands on. What made the great things great was rarely novelty alone. It was old, tested code, called by people who also knew how to listen for which call the moment needed.

Here’s the catch. Myths aren’t neutral. They’re memetic programs, and programs have outputs. A culture runs a story about what a human being is for, and that story helps decide what it builds.

I wrote about the starkest case in “The Fourth Pillar.” Two societies took the same twentieth-century shock. One ran the story of the self-made citizen, the covenant, the frontier you build out into. The other ran the New Soviet Man, the human as an input to the plan. Same century, much of the same technology. Different myth. One built the suburbs and the GI Bill; the other built the gulag and millions of graves. Myth was not the only variable. But it told each society what its power was for.

We’re walking into another one of those historical pivot points, and this time the technology is the most powerful we’ve ever built. AI hands us the capability either way.Which solutions we pull off the shelf, which story we run about what all that power is for, will help decide whether it ends in a renaissance or in the rubble. That is one of the most important choices our generation makes, and we’ll make it whether we mean to or not.

But which myth should we call upon today?

We are not short on solutions. We’re sitting on a tested codebase possibly fifty thousand years deep. And most of it is freely available. The question was never whether answers exist. It’s which ones we still know how to call. And the deepest one in the whole library, the solution driving the record’s biggest acts, isn’t a merely specific answer. It’s a method. Before the Ark, the Exodus, or the bodhi tree came the same move: withdraw from the noise, ask, and listen. That one also possibly has a concrete, material explanation — and it may decide whether we direct the machine or let the machine direct us.

That’s the next essay: I’ll lay out what that specific answer is, the science behind it, and how we can each reach it.

Subscribe now

Notes: The flood-story genealogy: mainstream scholarship reads the Genesis flood as descending from the Mesopotamian tradition (the Gilgamesh and Atrahasis tablets) — that’s the “honest forks” sentence. The independence claim rests elsewhere: flood-and-survivor traditions recorded across the Americas and Aboriginal Australia, where Babylonian transmission is implausible. One asterisk: some of those traditions were first written down after missionary contact, so any single recording can be contaminated; the sheer count across unconnected regions is what survives.

On Convergence: The late-antique Neoplatonist Sallustius essentially had the line sixteen centuries ago. Writing about the myth of Attis, he said, “Now these things never happened, but always are.” Campbell later mapped the recurring form. What both leave open is the engineering question: why do some stories keep returning, and when does recurrence count as evidence that the move works?

The commonly quoted “Myths are things which never happened, but always are” is a modern paraphrase. In Gilbert Murray’s translation of On the Gods and the World, Sallustius writes, “Now these things never happened, but always are,” during his allegorical interpretation of the Attis myth. Sallustius’s claim is Neoplatonic and cosmological: the myth represents a process that subsists eternally. The argument here is narrower and evolutionary. Recurrence becomes evidence only when the story or practice was independently rediscovered and reality had a meaningful opportunity to punish bad versions.

This is why convergence needs a selection test. The fact that a story is old and universal does not prove its advice is good. Bloodletting was near-universal for two thousand years (Greek, Islamic, Ayurvedic, and European medicine all converged on it), and it mostly just bled sick people. Convergence is strong evidence only where reality could actually punish the mistake — which premodern medicine, with most patients recovering anyway, could not.

Brunelleschi: “measured the Pantheon” is compression. He studied Roman vaulting at length, and the dome’s double shell and herringbone brickwork were new engineering. The point is the order of operations: recovery first, invention on top.

The cassava routine and its logic are Joseph Henrich’s; bloodletting’s near-universality across Greek, Islamic, Ayurvedic, and European medicine is the standard history.

Behind this: Joseph Campbell, “The Hero with a Thousand Faces.” Joseph Henrich, “The Secret of Our Success” (cultural evolution, and the cassava case). Richard Dawkins and Daniel Dennett on memes. And my own “The Fourth Pillar.”

Acknowledgements: Thanks to Ernest Morrow for his discussions that led to the initial ideas for this post. And to Greg Bowman and Lara Pande for their critical reading of this manuscript.

Here is a test for any institution: does it become more successful as the problem shrinks, or as it persists? And when it succeeds, does it leave you more capable, or more dependent?

We’ve spent fifty years shrinking our biggest problems individually, and that’s exactly why they won’t solve. Eisenhower supposedly said that when a problem can’t be solved, enlarge it. Healthcare reform, education reform, homelessness spending: each being fought alone, each losing ground. Small wars, same enemy. Here’s what happens when you widen the frame instead.

We have healthcare that keeps you sick, education that keeps you ignorant, and religion that keeps you from God. The target throughout is the toll—the profit from owning the gate—never the faith, the medicine, or the learning. Three industries, one recurring business model. Each stand between you and something you want (health, understanding, transcendence), charges for passage, and holds a quiet financial interest in you never quite arriving.

That is the distinction: a gatekeeper earns more from managing the journey than from helping you arrive. (Health economists have a name for this — supplier-induced demand: the gate creating its own traffic.) Essay two (”Why People Hate Capitalism“) called this rent seeking. At this scale it needs a bigger name: The Gatekeeper Economy.

Run the Sickcare Test from essay two—does revenue rise as the problem shrinks, or as it persists? California spent roughly $24 billion on homelessness over five years, homelessness went up, and the state could not consistently say what the money bought. (The housing chokepoint built the trap, and a services industry then grew around not springing it — the gates stack.) College tuition has risen something like 1,200% since the late 1970s; are graduates twelve times more educated? Every gate above passes the test.

Add the gates up and the number gets uncomfortable: the counts run from single digits to a quarter of GDP, depending on who’s counting, and the lowest number on the table is still trillions — a Gatekeeper Economy hiding inside the real one.

We’ve run this experiment before; the religion case is the oldest. The medieval church sold indulgences: cash for grace, the purest toll ever charged. (Its descendants sell subscriptions: meaning as a service, seventy dollars a year to sit quietly.) Then the printing press cut the price of a book by something like 90% inside two generations, Luther posted his complaint, and the gatekeeper lost its monopoly within a generation. Every technology that collapses the cost of access breaks the priesthood built on controlling it. AI is the printing press pointed at the remaining ones.

How do we have the opportunity to enlarge it now? Because AI is about to make the gates more valuable than ever—or worthless. It can cheapen the bypass, or automate the booth. AI spins flywheels (systems where each turn makes the next easier). Faster flywheels can mean steeper power laws, and steeper power laws can mean a few winners holding nearly everything. So the question is not only whether power concentrates, but where—and whether builders or tollbooth operators control the compounding loops.

Subscribe now

How do you remove a gatekeeper?

Fifty years of reform bills politely asked the gates to charge less, and the gates read the petitions and raised rates. Again and again, what seemed like sturdy gates fall the same way: bypass, quorum, ratify. Someone builds a road around the booth. Enough people take it while it’s still not fully reputable. The bypass turns diffuse anger into concentrated benefit; its users become the constituency. Then the law shows up last, in a crisis window, to pave what everyone is already driving.

Luther, above, gave us the instruction manual: he didn’t ask Rome for cheaper grace — he shipped a replacement (scripture in German, worship without the toll); the pews filled before Rome could organize; the princes made it law. Homeschooling ran the same play in living memory: effectively outlawed in most states in 1980, families defected anyway, prosecuting hundreds of thousands of parents became absurd, and by 1993 it was legal in all fifty. And notice what enlarging the problem just bought. Reforms don’t transfer — a healthcare bill is useless against the credential racket. The engine transfers. One engine, five rackets. That’s what bigger buys.

This isn’t merely a proposal. The bypasses are already open for business. Mark Cuban’s Cost Plus Drugs sells generics at cost plus a posted 15%; some land at a tenth of the pharmacy-counter price — a road drilled straight around the PBM booth. The Surgery Center of Oklahoma has published its cash prices online since 2009, routinely a fraction of hospital list, and direct primary care runs medicine for the price of a gym membership, no billing code in the building.

The credential gate is cracking: Maryland, Pennsylvania, and Utah have dropped degree requirements for most state jobs; IBM and Google for much of their hiring. And housing, the oldest booth of all, just produced the decade’s best evidence that the diffuse side can win: YIMBY turned zoning from a council secret into a moral argument, and Minneapolis, Austin, and a growing list of states rewrote the rules (Austin’s falling rents — essay two — are ratification on a receipt). Each of these is the engine, mid-cycle: a road open, filling, waiting on its window.

One rule before anyone builds the next road: the bypass can’t be merely a cheaper toll. It has to be a different road — one that doesn’t grow a booth of its own. It may charge for the road; it can’t earn more by keeping you from arriving. The design spec is a century old. Maria Montessori showed that a child in a prepared environment teaches herself to read; the adult’s job is to arrange the room. Learning is the default behavior of a child in the right environment (ask any parent whose kid taught herself YouTube).

Let’s transfer that mindset. Montessori for healthcare: arrange life so health is the default behavior (the walkable neighborhood, the kitchen, the sleep, the pickup game) rather than a service purchased after failure. Japan already builds cities where walking is the default; its obesity rate sits under 5%, against our 40%. Or look at Alcoholics Anonymous — peer-run, free, and a 2020 Cochrane review found it matches or beats the professional alternatives. One of the most successful healing networks in history charges no toll. Montessori for meaning: practice, play, community. At the minimum, we make the new behavior default; at its best, we make it engaging, almost seeming like play.

In every case you stop paying the gatekeeper to stand between you and the goal, and redesign the environment so you grow toward the thing yourself. A builder’s product is not just the outcome; it is your growing ability to reach it without the builder. (I’m arguing by analogy and I know it — the full Montessori hospital hasn’t been built yet. Someone should build it; that’s a startup, not a metaphor.)

Gatekeeping is everywhere. It hasn’t escaped me that VCs can become gatekeepers, between founders and LPs with a well-defined toll (2 and 20). Gatekeeping is a common play you can easily see once you’re looking for it, including journalism, politics, publishing, entertainment studios, etc. Tech at its best creates new ways around these tolls through a common solution: democratized distribution. In fact, that’s how you’re reading this right now.

The gatekeeper’s product was never health or knowledge or God. It was the gate.

The gatekeepers have willing customers—and all of us have been one

The unfortunate universal core to many of these issues often goes unspoken: literally or metaphorically, can you heal someone who wants to stay sick? Or, more directly, someone who has been paid, gently and for years, to stay sick? Educate someone paid to stay ignorant? Sickness, helplessness, and victimhood all pay — in attention, in absolution, in benefits, and in the deepest currency: an excuse.

The pitch runs, you’re perfect as you are, and anything you lack was taken from you, so the system owes it to you. I’d love to claim immunity; I do here and there catch myself running that pitch, setbacks reframed as somebody else’s doing. Nobody is immune –– it’s the most comfortable lie on the market. It’s also the gatekeeper move again, aimed at the bottom instead of the top. Rent seeking says: I get money for the position I hold, not for what I build. Entitlement says: I get stuff for the identity I hold, not for what I become. Same disease, different hosts.

Widen the frame one more step and the century’s biggest argument collapses the same way. Capitalism versus socialism is an argument about who should own the tollbooths. The builder’s answer is that roads shouldn’t have booths.

The pitch gets one thing exactly backwards: the getting was never the point. If your parents had “helped” you by carrying you everywhere, you would never have learned to walk, and nobody would call that love. Ask any physical therapist: we heal by doing. Every real acquisition is built, and the building is the acquisition. Essay one said a slice of someone else’s output can’t raise you up. Here’s the mechanism: the raising was always internal. The institutions worth building leave you stronger than they found you.

The fix runs through culture before it runs through policy, and not because culture is the soft option. It’s the only arena with winning math. Why People Hate Capitalism left a problem standing: concentrated billions beat diffuse forty-dollar losses every time, and enlarging the war makes that worse on paper: your side gets more diffuse, and five rackets facing one movement will pool their lobbyists. Olson’s asymmetry is unbeatable in the statute book. Culture changes the math: it turns diffuse irritation into a shared identity. You can’t beat this one on one. Lobbying can block any bill; it has never once bought back lost respectability.

Movements change what people admire; markets follow; the statute book ratifies. We made smoking embarrassing — teen smoking fell from about 36% in the late 1990s to under 2% today. Same story with drunk driving — MADD, one grieving mother, 1980; the per-person death rate has since fallen by more than half. Nobody ever passed a law against smoking at dinner parties; the room just turned.

Cultures can turn. Imagine the same shift here: being a gatekeeper becomes embarrassing, and being a builder — of your body, your mind, your business, your soul — the admired default. Capitalism at its best is exactly that culture: self-improvement compounding until it shows up in the world as things built, problems shrunk, lives extended. Concretely: pick one gate you’re paying at and walk around it yourself this year. Cook. Lift. Teach your kid to read. A movement, at day one, is a few people quietly defecting — that’s how the road fills.

The real work

And now take Eisenhower all the way, because the problem has one size left. The last time a technology collapsed the price of knowledge like this (Gutenberg, again), the aftermath ran far past anyone’s reform agenda: it ran to Florence, to perspective, to the uomo universale, the Renaissance man — a new portrait of what a human being could be. Enlarged to its true dimensions, this stops being healthcare reform, or education reform, or any reform, and becomes the work of bringing forth a renaissance — on purpose, this time. A renaissance is a civilization that expects people to become more—and builds the conditions for it.

Five times before, by my count (my full accounting is in “The Fourth Pillar“), a shock like this has forced a society to redraw its portrait of the human being. The ones that drew growth got golden ages; the ones that let the cheapest answer draw it got the graves. The Gatekeeper Economy is what a civilization builds when its picture of a person is a customer: managed, absolved, waiting. The builder is the other picture. And I mean builder in the full sense — someone who owns their means and uses them to hold real things: a craft, a household, a place, people who count on them. Every gate you walk through yourself is a brushstroke on it.

Just as I asked you to grade essay two in 2031, this claim should go on a clock too. “Renaissance” is not a mood; it has a needle. I keep a gauge of the hollowing in “Build the Ark” — trust, cradles, friendship minutes, deaths of despair. If a movement of builders is real, that needle starts to turn inside a decade. If it doesn’t, I was wrong about the mechanism, not just early.

The tools are on the table (Owners of the World Unite: the factory now fits in one hand). The tollbooths have been named (Why People Hate Capitalism). The move is now clear: build the bypass, fill it, then let the law ratify the road. Movements get installed in the window a crash opens, and the cheapest answer on the shelf wins that window unless a better one is already built.

And if you deploy capital for a living—investments or donations or company building, apply the same screen—the first question is one line long: does this business get paid more when the problem shrinks, or when it grows? And when it succeeds, does it leave the customer more capable—or more dependent? What remains is the oldest question, the one every gatekeeper is paid to keep you from asking: do you want to be better? I’m betting far more of us do than the gatekeepers assume.

In this light, the impossible becomes merely a tremendous amount of work.

Make the problem Renaissance-big. Then build the bypasses. And break the culture of gatekeeping.

Subscribe now

Notes: I first heard the mindset of “in order to solve a problem, make it bigger” from Todd Park, co-founder of Devoted Health. I was delighted to see that President Eisenhower had a similar mindset.

Builder mindset: If enlarging the problem sounds like the opposite of what I tell founders (narrow the wedge, win the beachhead), it’s the same discipline at two altitudes. The frame is where you enlarge. The first move stays one gate wide.

Counting the rent: Krueger’s original licensing estimates (1974) ran 7–15% of national income; Laband and Sophocleus put America’s transfer-chasing as high as a quarter of GDP. They’re counting different things — licensing rents versus everything spent chasing transfers — which is why the body takes only the floor: even the lowest number is trillions.

California’s homelessness figure is the state’s own audit: roughly $24 billion across five fiscal years (2019–23). The auditors’ sharper finding was that the state couldn’t say what the money bought — no consistent outcome tracking. That’s the farming signature.

The tuition figure is the BLS sticker-price series: about 1,200% since 1978, several times general inflation. Net-of-aid prices rose less — and the $1.7 trillion in debt (essay two) says the difference still landed on students.

Japan’s obesity rate is the OECD measure (BMI ≥ 30), the same cutoff behind the U.S. number, so the 5%-versus-40% gap is apples to apples. Japan’s domestic definition is stricter still.

The AA finding is the 2020 Cochrane review (Kelly et al.): manualized twelve-step programs matched or beat CBT and other clinical treatments for sustained abstinence.

Teen smoking: 36.4% in 1997 (CDC) to under 2% today — some nicotine moved to vapes, which is the honest asterisk; the cigarette culture still died. Drunk driving: per-capita deaths down more than half since MADD’s founding (NHTSA) — with the statutes (per se limits, the 21 age) arriving after the norm turned, which is the essay’s sequence.

The bypasses: Cost Plus posts cost plus 15% and fees; the Surgery Center of Oklahoma has posted cash prices since 2009; Maryland (2022), Pennsylvania (2023), and Utah (2022) dropped degree requirements for most state roles; homeschooling went from effectively outlawed in most states around 1980 to legal in all fifty by 1993 — prosecution collapsed before the statutes changed, which is the quorum step working.

Acknowledgements: Thanks to Lara Pande for her critical reading of this manuscript.

The Renaissance Papers · Series II, part 2. Previously: part one argued that cheap machine labor can move productive power toward individuals. But ownership means little if every road from production to the customer is tolled. This essay names the tollbooths. The map is here.

The anger at capitalism is aimed at something real. And it’s our job to fix it. That's not a claim thrown around frequently in venture capital. And I just wrote an essay retiring Marx—but stay with me. The mistake in all that anger is the target. Much of what people hate about capitalism isn't the building of businesses; it's five specific rackets we keep producing, all five running the same play.

Economists have a name for this pattern: rent seeking — income you collect because you control a chokepoint, not because you built anything. A tollbooth on a road someone else paved. The booths have gone up everhwere: on the land under your apartment, at the hospital door, inside the pill bottle, across the only bridge out of town.

Here’s how AI can radically change this game.

Subscribe now

What people really hate: five rackets

Racket 1: the chokepoints themselves. Housing is the obvious case — the rent in rent seeking is literal. Austin recently offered the test case: permit a building boom, and rents promptly fell about 7% in a year. But the more pure modern example is healthcare complexity. Every layer of billing codes and prior authorizations creates a niche where someone charges for passage. By one widely cited count, since the mid-1970s, the number of American physicians grew about 150%; administrators, something like 3,000%. Nobody got healthier because of the second number.

Racket 2: heads I win, tails you lose. Privatize the gains, socialize the losses. The S&L crisis dropped roughly $124 billion of bank losses on taxpayers, and 2008 ran the same play again. A factory that pollutes a river books the profit and mails the cleanup bill to the public. Economists call the missing charge an externality — a cost you create but don’t pay. People on the short end of it call it a rigged system.

Racket 3: markets that feed on the problem. When revenue rises with the persistence of the problem rather than its resolution, that system is profiting from its problem. I call it the Sickcare Test. America spent $4.9 trillion on healthcare in 2023, nearly a fifth of GDP, much of it managing disease rather than ending it. In recurring-revenue sickcare, a patient cured is a customer lost. When Gilead’s Sovaldi started actually curing hepatitis C (around 95% of patients: treated, done, gone), a Goldman Sachs analyst asked, “Is curing patients a sustainable business model?”

Arms dealers pass it: more war, more revenue. Parts of higher ed pass it: $1.7 trillion in student debt, much of it buying credentials whose main function is to require the next credential. Corrections passes it too: well over $100 billion a year buys a 68% three-year rearrest rate, and the system’s budgets survive the failure. California’s homelessness system shows the same warning sign: billions spent, homelessness up, and the state unable to say consistently what the money bought.

In my mind, this is the big one in terms of its broad proliferation and ease of establishment, when “helping” be used to disguise rent seeking.

(Before the complaints mount: yes, profits fund R&D, and inventing cures for profit is exactly what we want more of. The build deserves a return. The racket begins when protection outlives the risk that justified it.)

Racket 4: profit on the inelastic. Inelastic is the econ-term for demand that barely moves when prices rise — insulin, an EpiPen, water after a hurricane. When necessity meets market power, price stops measuring preference and starts measuring captivity. Banting and his collaborators sold the insulin patent for $1 in 1923 because they thought it belonged to the world; by the early 2020s, vials costing single-digit dollars to make listed near $300. The public’s fury at this is economically literate. Pricing power over a captive buyer is the same booth again: this time between you and your own bloodstream.

Racket 5: the quiet rewrite of the deal. In 1980, roughly four in ten private-sector workers had a pension — a promise someone else had to keep. Today it’s about one in ten, and the replacement (the 401k) moved ownership and market risk onto the worker. Maybe that trade was even necessary — portability matters when nobody stays thirty years at one company. But the trade was never priced honestly (”the professions can’t make the pension math work, so you have a go at it yourself”) and they can feel the difference even when they can’t articulate its cause. Ownership without a floor is not the same as freedom.

Five rackets, one root: money made without a matching improvement.

The mechanisms differ, but the move is the same: collect where the customer cannot meaningfully say no. The strongest booths still sell something real. Often they sell back, at retail, what a working commons once supplied: the safe street priced into the rent, the trusted school as private tuition, the neighbor who used to watch your kid as an additional expense. That’s why the metrics can keep improving while life feels worse. And a tollbooth that shares enough of its take recruits its own defenders, not just customers (see last endnote).

The opposite of a racket isn’t a nonprofit. It’s a business that gets paid when something actually improves — and bears the costs when it doesn’t.

Notice what’s not on the list. The builders who sell people things they freely want — sneakers, phones, the morning coffee (albeit they collect the occasional boycott at their racket moments, ie the sweatshop or the walled garden). The problems run show up in the hospital bill, the tuition letter, the rent check. Jonas Salk gave the polio vaccine away and asked, “Could you patent the sun?” The public’s diagnosis that something is wrong is often better than the pundits’. They’ve just been told its name is capitalism.

So why is it still growing? We’ve been operating for forty years, one racket at a time. Cap insulin, and the rent moves to the PBM (pharmacy middlemen — three of them touch 80% of American prescriptions). Regulate the banks, and the risk migrates somewhere darker. Audit the colleges, and the credential inflates. Mancur Olson explained why half a century ago; a small group with concentrated winnings out-organizes a large public with diffuse losses, every time. The racket will always care more about its billions than you care about your extra $40 a month.

Forty years of losing was never about will. The scope was wrong: five small wars where there should have been one big one. Piecemeal reform regulates the booth. Socialism too often gives it a new flag. Where technology makes it possible, the builder’s answer is to dissolve it.

What AI hits first

AI is about to flip the board, and not evenly, because the five rackets are not equally mortal. Here’s the order they become vulnerable. Rents made of information go first: the PBM’s spread and the 401k’s fee maze survive only as long as auditing them costs more than paying them, and an agent that audits for free ends the information advantage. Complexity rents go next, with a catch — right now it’s the insurers buying the prior-auth AI, not the patients. Watch whose agent the machine is; that single fact tells you who keeps the rent. Credential rents become vulnerable more slowly, by legitimacy collapse rather than substitution: nobody needs the learning gate once learning is free, but the credential survives as a trust signal — and because the law still says you need it.

Externality rackets subside as detection gets cheap (a satellite can now read a methane leak off a single well), though enforcement stays political. And the oldest rents — land, patented essentials, the physically scarce — barely change. Their scarcity is physical or legal; the rent is defended by law. So here’s the tell for the next decade: lobbying budgets will spike in proportion to AI-vulnerability, because a racket that can’t survive the market retreats into the statute book. This paragraph is a forecast; grade it in 2031. And you don’t have to spectate: root for the agents, or even better, build them.

The rents AI can’t kill are the ones only a movement can. The extermely counter intuituve, paradoxical answer is that we haven’t solved the problem because we haven’t made it big enough. How five losing reforms become one big movement — Renaissance-big — is essay three.

Subscribe now

Notes

The physician-versus-administrator figures (~150% vs ~3,000% growth since the mid-1970s) trace to Himmelstein and Woolhandler’s compilations; definitions of “administrator” vary and the base year flatters the ratio — the direction survives every definition on offer.

The S&L bill is the FDIC’s own post-mortem: roughly $124 billion of the cleanup landed on taxpayers.

U.S. healthcare spending reached $4.9 trillion in 2023, according to the Centers for Medicare & Medicaid Services’ National Health Expenditure report.

The Goldman Sachs question is verbatim from an April 2018 client note, “The Genome Revolution,” by analyst Salveen Richter. Goldman confirmed the report’s contents to Ars Technica ; the note used Gilead’s hepatitis-C franchise as its case study. California’s State Auditor reported that the state allocated nearly $24 billion to homelessness and housing affordability from fiscal years 2018–19 through 2022–23, while lacking consistent statewide information on program costs and outcomes. The audit also reported more than 180,000 Californians experiencing homelessness in 2023, up 53% from 2013.

The current direct-spending estimate for corrections — prisons, jails, juvenile facilities, probation, and parole — is roughly $116 billion a year. The 68% figure is the Bureau of Justice Statistics’ three-year rearrest rate for state prisoners released in 2005.

The insulin patent rights were transferred to the University of Toronto for one dollar apiece . “Near $300” refers to major manufacturers’ list prices in the early 2020s; for example, Humalog listed at $274.70 before Lilly’s major price reduction took effect in 2023.

The PBM figure: three firms (CVS Caremark, Express Scripts, Optum Rx) process roughly 80% of American prescription claims; the share moves a little year to year, the concentration doesn’t.

Pensions: defined-benefit coverage of private-sector workers ran about 38% in 1980 and sits near one in ten today (BLS and EBRI series — definitions of “coverage” vary, and the cliff is in every one of them).

Johann Kurtz,“Young adults are poor despite every metric which suggests otherwise” , July 14, 2026, argues that when social capital and common goods decay, families must buy private replacements (safe neighborhoods, trusted schools, childcare) at retail. Quinn Slobodian, “Digital Bandung”, The Ideas Letter 64, May 14, 2026, sharpens the other half: durable systems of extraction survive not by taking alone, but by delivering real convenience, pleasure, employment, and asset gains through networks of intermediaries and beneficiaries.

Acknowledgements

Thanks to Lara Pande for her critical reading of this manuscript.

But the trilogy ended on at a crossroads. When machines take more of the labor, do we redistribute the proceeds—or spread the means of production?

Owners of the World, Unite begins the second series, The Builder’s Answer. It takes the second path:

For most of history there was a line: workers sold labor; owners commanded it. AI may let far more people cross.

Part two lands Tuesday. It is not the essay you would expect from a venture capitalist: Why People Hate Capitalism. The anger is aimed at something real; the question is what should we do about it.

The full project now has a map: https://vijaypande.substack.com/p/start-here-the-renaissance-papers

The Renaissance Papers · Series II, part 1. Series I ended at a crossroads: when machines take more of the labor, do we tax the proceeds or spread the means? This series takes the second path—beginning with how cheap machine labor can move people from workers toward owners. The map is here.

Marx’s theory of exploitation is an argument about labor as an economic force—and AI is removing the labor. Every move Marx makes runs through the worker: value comes from work. The capitalist profits by paying less than the work creates, and exploitation happens right there, on the shop floor. And we’re already seeing early signs: Labor’s share of American income has slid from roughly 64% to the mid-50s since 2000; AI bends that curve harder. AI-driven robots may finish the job.

My timing here is delicate: people are furious at capitalism, often for good reason. That fury gets its own essay next.

What’s left of a politics built around exploited labor when wage labor is no longer the central productive relationship? Not much. You can’t skim a surplus from human labor the machine no longer requires. The cost curve does the rest. Marx assumed the means of production were big and expensive—a steel mill, a railroad—which is exactly why ownership piled up in a few hands and stayed there. Fair enough, for about two centuries. In computing terms, Marxism was an argument about the hardware layer; the hardware just changed underneath it. But one critical means of production is increasingly the model, and the price of intelligence is falling off a cliff.

This is Moore’s Law again, driving cognition instead of transactions per second. Today you can download useful open-weight models for free—the learned guts of the thing—and run them on a laptop, even a phone. Near-frontier open models trail the frontier by months, not generations. The commanding heights get harder to monopolize when they can be copied. Marx wanted workers to own the means of production. The market he spent his life hating is handing them over.

The best counterargument is concentration: frontier models cost billions to train, and a handful of labs sit on the biggest clusters. True—but that misses where the leverage lives. Training is expensive; using intelligence is cheap and getting cheaper, the way building a power plant is expensive but plugging into the wall is not. Cheap access alone would make everyone a customer. Open weights are what can make them owners. DeepSeek reported spending roughly $5.6 million on the final training run for a near-frontier model, excluding the research that made it possible; open weights keep leaking the frontier outward a few months behind. The moat, such as it is, protects the plant. The revolution is at the other end of the wall socket.

Subscribe now

The worker becomes an owner

Which points to the real revolution, and why I don’t think it has much to do with redistribution. The real revolution is the worker becoming an owner. That’s what a renaissance is built from. For most of history, the line between labor and capital was nearly impossible to cross: starting anything took money, a building, staff, expertise you couldn’t afford. Most people sold their labor and stayed on that side of the line for life.

Cheap intelligence dramatically lowers the barrier. One person with these tools can increasingly do what took a whole company: build the product, write the code, do the books, find the customers. (Midjourney reportedly passed $200 million in revenue with about 40 people and zero venture money. The one-person, billion-dollar company is coming.) Roughly 28 million American businesses already run with zero employees; the tools just promoted all of them. Not everyone needs to build a billion-dollar company. Productive ownership now works at a vastly smaller scale. The worker doesn’t need to seize the factory anymore. He can spin one up. Marx had a “reserve army” of the unemployed; what’s assembling instead is a Reserve Army of Founders.

When I say AI removes labor, I do not mean it removes work. AI lets more people cross to the other side: instead of selling only their own, they can command cheap machine labor and keep more of what it produces. That distinction is the heart of the argument. AI has the potential to promote people, just as previous technological advances have—continuing humanity’s long, unfinished promotion from slaves to serfs to salaried—now to shareholders and sovereigns.

Marx published in 1848; within eight years, Britain let any seven people register a company, and tens of thousands followed. The machinery of mass ownership was being built as the ink dried.

To be fair, Marx did see the machines coming. In 1858, in a notebook passage his readers now call the “Fragment on Machines,” he sketched an economy where machines do the producing and labor stops being the measure of value—he just bet the ending on collective ownership. He read cheap intelligence as the case for owning the machines collectively at the exact moment it became possible for far more people to command productive machines individually. Marx may even have been right about the destination: a world where nobody spends a life in wage labor. He was wrong about the vehicle, and the vehicle is the whole fight. (My best guess at what Marx himself would do in 2026: download the weights and get Engels to pay for the laptop.)

His conclusions were bolted to facts that have since flipped: scarce capital, labor as the source of value, no exit from wage dependence. Yet the slogans hang around because they’re useful for selling redistribution. Consider the universal check aka UBI. As a permanent settlement, it keeps you a client: owning nothing, making nothing, waiting on whoever signs it. A subscription to being managed. That’s closer to the techno-feudalism the left says it fears than anything the market is doing. A slice of someone else’s output doesn’t raise anyone up. The means to make your own does.

The nineteenth century asked who should own the factory. The twentieth fought over the answer. The twenty-first may dissolve the question, because the factory increasingly fits in one hand, on your phone. What we do with that fact decides whether this century’s answer is another ideology or another renaissance.

How the argument fails

None of this is guaranteed. If open models stop tracking the frontier—if that few-month gap stretches into years—my thesis dies and concentration wins after all. Same if compute prices stop falling; the argument rides on that curve. And there’s a third factor, because the weights were never the whole means of production: a model on your laptop doesn’t come with customers. If the tool layer stays open while the roads to the customer—the app stores, clouds, feeds—quietly re-toll, then ownership is a gift certificate for a store you can’t reach. The battle moves from owning production to owning distribution. Watch all three factors; the open-weights gap remains the most political number in tech. And a laid-off worker isn’t a founder yet. Closing that gap is most of the work. But the opening is real, and I think it’s the biggest workers have had in the history of work.

And let’s not pretend the transition will be painless. A 52-year-old dispatcher whose job just evaporated doesn’t want a pep talk about founding; she wants next month’s mortgage. The transition years will be ugly for millions, and pretending otherwise is how you lose them to the slogans. We need to build this right. The tools are arriving faster than the on-ramps. Checks may bridge the transition; building the on-ramps is the political project of the next decade.

But there’s an elephant in the room I’ve ignored, an obvious objection to my argument. If the market is this generous, why are people so angry at it? Rent eats half a paycheck; hospital bills read like ransom notes. The anger points at something real. I just think it has been given the wrong name, and you can’t cure a disease you’ve misdiagnosed. Before the building starts, the critics are owed a straight answer.

Why do people hate capitalism? That’s essay two, next.

Subscribe now

Don’t robots become the new factories: the capital-intensive resource restricted to a few? There is no Moore’s Law for robots, at least not yet, but robotics costs have fallen through scale and more widely distributed know-how. The real bend comes when robots build robots and their parts, driving costs down by another order of magnitude. I expect capable general-purpose robots to approach $1,000. That is not zero, but it is on the other side of power plants and railroads.

Open weights: Llama, DeepSeek’s releases, and a rotating cast of others. The gap behind the frontier has run months, not years—that’s the number to watch. Midjourney’s figures—roughly $200 million in revenue, roughly 40 people—are widely reported but unaudited; “reportedly” is doing real work in that sentence.

Notes: The labor-share figure is the BLS nonfarm-business measure: roughly 64% in 2000 to the mid-50s. Measurement debates—proprietors’ income, depreciation—move the level, not the direction. The 28 million zero-employee businesses are Census nonemployer statistics. Most are side gigs and landlords today, which is the point: the on-ramp exists before the ambition does.

Acknowledgements: Thanks to Lara Pande for her critical reading of this manuscript.

The Renaissance Papers · Series I, part 3. Previously: Let It Crash argued that the storm should be allowed to come. This essay is what to build before it lands. Who builds, who owns, and with what is where Series II begins. The map is here.

In an old machine shop, a man runs a school you’ve never heard of. A dozen teenagers read hard books and build things that have to actually work. He left a real career for it. He’d never use the word, but he’s building an ark.

The last years of a civilization don’t feel like decline. They feel like the best. That’s what makes them the last years. The flood didn’t come just for the wicked. It came for the comfortable, people who had everything and stopped watching the water. We are Noah’s neighbors.

Two waters are rising on us at once. The first is the storm everyone is bracing for: the AI bubble, valuations floating free of anything real, data-center spending swollen beyond reason. That water is real, and it is the smaller danger. A storm on its own can drain off — markets break, losses become real, and the country spends years rebuilding — but it does not have to become a flood.

What turns a storm into a flood is the tide it lands on. That tide has been rising for a generation, slower and almost entirely unwatched, and it decides whether the coming break drains off or breaks through the door.

I’ve spent thirty years at the edge of this — a Stanford professor at the AI–biology frontier, then an investor — and the pattern I keep landing on is the one the man in the machine shop already lives: the water never saves or sinks you; what you built before it arrives does. This essay is about reading the tide while the sky is still clear, and the ark built in the calm before the storm.

Subscribe now

A storm and a tide

The financial storm runs on a clock we half-know. Every technological surge of the last 250 years has run the same course: revolution, bubble, crash, then golden age. (Carlota Perez named the pattern; I argued our place in it in “Let It Crash.”) It moves fast, on a decade’s clock, and everyone watches it.

The tide moves slower. Six hundred years ago Ibn Khaldun called it asabiyya: the shared capacity to believe in something and act on it together. It runs strongest when a society is hungry and ebbs over the three or four generations it takes for hardship to give way to comfort. It is gradual enough that no one feels it move.

Neither water floods you alone. A storm at low tide is a hard rain that washes away, a correction the economy shrugs off in a few years; the same storm at high tide becomes a flood that takes a generation to clear. The storm sets how hard it hits, which is not ours to control. The tide sets how high the water already stands, which is.

Why would a culture’s water rise at all? A culture is a society’s answer to one question: what is a human being for? While that answer is alive, money is one value among several, held in check by faith, duty, craft, and place. When it hollows out, price becomes the last shared belief standing, and capital floods the vacuum because nothing else is left to believe in. That answer is what “The Fourth Pillar” was about; the tide is just how high it still stands.

History has run the experiment for us, and the cleanest case is a single year. In 1929 the same crash hit the United States and Germany within months. America took it near low tide: a cohesive society, a long democratic habit, institutions people still trusted. It was the worst decade in living memory, and the country still stayed dry enough to build the New Deal and the postwar boom. Germany took the identical storm at the top of its tide — a young republic its own people didn’t believe in, savings already wiped out by hyperinflation, the streets already armed. The water came through the door. Cohesion didn’t vanish so much as change owners: the flood drowned the republic’s, and a malign one climbed out on the cheapest answer on the shelf.

Run the rest of the cases the same way and they sort into weather:

So how high is our own water? The starkest number is the birth rate, which has fallen below replacement across the entire developed world at once: South Korea sits at around 0.8, barely a third of what holds a population steady, with China, Japan, and Italy close behind. A birth rate is a referendum on the future: people bear children when they believe in one, and the developed world has quietly voted no.

The same retreat shows up day to day. The share of American men with no close friends has climbed from three percent in 1990 to fifteen; time with friends has fallen to about twenty-six minutes a day; the Surgeon General now ranks loneliness with smoking as a cause of early death. None of that is opinion. It is measurement, and it reads the way the fertility numbers do.

I’ve tried to fold these counted signals into a single gauge, a Cultural Stress Index, built only from non-market readings, that today reads at a level that should worry anyone who believes the next crash will be merely financial — and stir anyone who sees that the same forces, met in time, are the makings of a renaissance. Its rubric, back-test, and many caveats are in the appendix, and I’d trust the cradles and the friendship surveys long before I’d trust any index. They point the same way.

The gauge we can’t read is whether our institutions hold — the seawall that separated America in 1929 from Weimar. Unlike the tide, it can still be built, and for now it stands. But a seawall only buys time; what you build with that time is the ark.

The way back up is unglamorous: keeping promises, raising children, belonging to something you would sacrifice for. The room had many names: church basement, lodge hall, school gym, volunteer firehouse. But it did one thing: it knew your name and expected you back. That room is the high ground, and it emptied one skipped week at a time.

Build the ark

The coming storm is worth wanting; the flood is not. A storm re-prices what is false; a flood drowns what is true along with it, and the whole work is taking the one without the other. The tide has already read us nearer the floodplain than the high ground, so wanting this storm is a wager that we can climb before it lands.

Toynbee, watching civilizations break, found that a frightened society reaches for one of three responses, and only the last is a way through. Two are escapes: 1) utopianism, the flight into another time — an idealized past to make great again, or a revolutionary’s clean slate ahead; and 2) detachment, the flight from the world itself, whether the prepper’s bunker or the doom-scroller’s numb hedonism (the strongman fuses the two, and it is the cheapest answer on the shelf). Only the third builds. Toynbee called it transfiguration: not escaping the present but using it, not just transforming but elevating it, making something new and worth believing in on the ground where you stand.

That’s the difference between an ark and a bunker. A bunker waits out the world, hanging on to what came before it; an ark carries the world across to the far side. It’s the line between withdrawing from the decline and transfiguring it — between saving yourself and building the thing the next age climbs out onto.

That is the ark: not rescue from the age, but the age elevated into something worth inheriting. And transfiguration does not route around the suffering; it runs through it. The same hardship that ends the old world is what tempers whoever builds the next. The early Christians did not escape Rome’s cruelty; they spent it, and what they built on it outlasted the empire that inflicted it.

What fills the vacuum is whoever shows up with something better already built. The Italian Renaissance was exactly that: a classical inheritance kept alive and then recovered, met by fresh creative energy under the new money of the city-states. The inherited and the new, working together. Those are our conditions too: a technological surge, new money in flood, old certainties cracking, the very weather Florence built in. The storm we fear is also what a renaissance is made from; what’s missing is only the builders.

The flood story doesn’t end with the rain. It ends with the ark. Noah built it while the sky was clear and the neighbors laughed, because that is the only time an ark gets built. And the ark is built from what cannot be generated — the argument “The Fourth Pillar” made: an AI can make anything except someone who shows up. In practice, the ark is not a document but a set of rooms on high ground: a school where the young read hard books and inherit obligations to a place, a workshop where teenagers make real things, a household where people grow harder to wash apart because reality has a claim on them and others know their names.

Durant put the stakes of that first room as flatly as anyone: “Civilization is not inherited; it has to be learned and earned by each generation anew; if the transmission should be interrupted for one century, civilization would die, and we should be savages again.” The school is that transmission, kept unbroken by hand. Discourse won’t build it. Patrons, founders, teachers, and parents will. They understand that civilization is downstream of what they make possible.

These three essays are one build. The first said what: an answer to the question AI now forces on us, what a human being is for. The second said when and who: before the crash, in many hands, not one. This one says why it can’t wait: the tide is high, and a question left blank gets answered by default, by whatever’s cheapest on the shelf. So, plainly: decide, on purpose, what a human life is for, and build it into real places — a school, a workshop, a household — owned by the people in them, not a platform, not the state. That’s the ark.

One piece I am personally claiming is the patron’s: finding the people already building the ark — the man in the machine shop and a thousand like him, the teachers, the founders, the keepers of workshops and schools — and connecting them with one another and the capital that should already be finding them. If you are building one of these — the school, the workshop, the household, the book, the open stack that keeps the tools open to newcomers, or the net for the people who take the hit — write to me. Let’s get the ark builders together before the storm hits.

The ark will not look grand at first. It will look like twenty teenagers in a room where phones are away and someone expects their best work. It will look like a founder choosing open weights over a tollbooth because he wants successors, not dependents. It will look like a family that still eats together, a workshop where skill passes hand to hand, a patron who funds the teacher no market would have noticed. Far from a retreat, this is the recovery of places where people become harder to wash away. None of it is a sacrifice. It’s a homecoming: work that means something, mastery you can feel, children who thrive, a place that knows your name.

The storm is coming, and the tide is high. But the flood was never only an ending; for the one who built, it is the way across. That is transfiguration in its plainest form: not bare survival, but the first rooms of a new renaissance, raised by the people who started while the sky was still clear.

Subscribe now

On storms and tides: an ordinary storm is a bubble at low tide; it drains off (1907, 1987, 2000). The flood is a storm at high tide, the rare coincidence that resets a regime (Weimar, 1929–33). The flood runs a generation and remakes institutions — what “Let It Crash” called the depression — while the storm, or crash, is only the repricing, which a sound society can clear without regime collapse.

One case sharpens the rule. Japan in 1990 took one of the largest financial storms on record with its demographic water already high, yet did not drown the way Weimar did: trust held, crime stayed low, the institutions kept working. That is why the gauge reads Japan low, not high — the readings that set the water’s height are the cohesion readings, and Japan’s held. What followed was not Weimar’s wave but thirty years of standing water, the slow inundation demographics alone can drive. Cohesion kept Japan from drowning; the refusal to clear its zombie banks kept it from drying out. What matters is whether a society still holds together, a different question from how old it is or how few children it has.

Cycles: Carlota Perez, “Technological Revolutions and Financial Capital” (2002), with Minsky and Kindleberger, for the financial storm; Ibn Khaldun’s “Muqaddimah” (1377) on asabiyya, Sir John Glubb’s “The Fate of Empires” (1976) on the ~250-year arc, and Peter Turchin’s structural-demographic work for the quantitative version.

On rise and recovery: Will Durant’s “The Lessons of History” (1968), distilling fifty years of “The Story of Civilization,” for the mechanism — why civilizations decline (leadership failing the challenge) and how a heritage outlives the fall (transmission); and Arnold Toynbee’s “A Study of History” for the three responses to breakdown — utopianism (archaism and futurism), detachment, and transfiguration. Durant supplies the odds; Toynbee, the stance.

Cases: the 1920s bubble and Jazz Age (Fisher’s “permanently high plateau,” 1929); tulip mania (c. 1637) inside the Dutch Golden Age; Japan’s 1980s bubble and lost decades; Weimar (1919–1933). On the decadence: Nietzsche on ressentiment; the friendship and loneliness data from the Survey Center on American Life and the American Time Use Survey; the 2023 U.S. Surgeon General advisory on loneliness; Putnam’s “Bowling Alone”; sub-replacement fertility worldwide, lowest in South Korea, China, Japan, and Italy.

The flood is near-universal — Utnapishtim in Gilgamesh and the older Atrahasis, Noah, Deucalion, Manu, and the Chinese flood tamed by Yu the Great. The tellings disagree on why the water comes — judgment, the gods’ annoyance, a cosmic turning, a river to be tamed — but agree on what matters: the one who read it in time and built is the one who makes it through. The two waters recur too: Genesis opens the springs of the great deep and the floodgates of the heavens on the same day, and behind it sits an older picture of creation as a separation of waters above from waters below. A flood is that separation coming undone. For the one who has built, the ancients quietly knew, the flood is not only ruin but the way through.

Acknowledgements: Thanks to Greg Bowman, Joel Dudley, and Lara Pande for their critical reading of this manuscript.

Cover art: Raphael’s workshop, The Building of Noah’s Ark (Vatican Loggia, c. 1518).

Appendix — The Tide Gauge (Cultural Stress Index)

The thesis of this essay is that a financial storm turns civilizational only when it makes landfall at high tide — on a culture already in decline. That is testable only if the tide can be read on its own, before the storm and without reference to the market. The Cultural Stress Index is that gauge.

This simple model is intended to give detail and a concrete grounding to the claims above, but I do not claim that this is proof or the final word. View this as a starting point for a more detailed investigation.

The readings. Twelve non-market readings, in five clusters. Each is scored 0 (low water), 1 (rising), or 2 (high) for a society in the years before a storm breaks. The twelve are summed (0–24) and rescaled to a 0–100 index. Keeping every reading independent of the immediate market storm is the point: it lets the tide and the weather be read separately, so their coincidence becomes a finding instead of an assumption.

Reading the gauge. With a storm breaking, below 48 predicts a correction, 48–69 a severe and contested downturn, 70 or above a flood that takes a generation to clear. With calm markets, a high reading (55+) predicts slow rot — decline with no storm to trigger it.

The back-test and the four-model check. Scored across nine historical episodes and the present, the gauge sorts floods from corrections cleanly, with every resolved case landing where it points. To guard against scoring the cases to fit outcomes already known, the same framework was run independently through four frontier AI models, each given the rubric and the episodes but none of my scores. They reproduced the historical pattern almost identically and agreed with one another and with the reading for today. Each model’s Cultural Stress Index by episode:

The 1929 row is the natural experiment: the same storm scores low on the U.S. and high on Weimar across all four models, and the water went accordingly. The TODAY row is the live test, a mean near 91 with three of the four models above their own Weimar reading and the fourth (GPT-5.5, 79 to its 81) just below it. The 2008 row is the gauge’s hardest case, and it stays open on purpose: Weimar-range readings met a flood-scale storm, and the seawall, the institutional continuity the gauge deliberately omits, held. Whether 2008 was contained or merely deferred is the question the TODAY row inherits.

Three caveats. The historical scores were reconstructed with the outcomes already known, so the back-test shows the gauge is consistent with history rather than proven to predict it; a real test would freeze the readings and have independent raters score each case from period sources, blind to outcome. In particular, for the earlier episodes when many of the data points (e.g., GSS, World Values Survey, other statistics) had not been documented or didn’t exist, the foundation models are drawing upon other records at the time. My use of AI foundation models is a simple test, but one could go much further.

Four AI models agreeing is not four independent witnesses, since they share training data and therefore share biases, so their unanimity should count for less than four independent analysts would. Nor is any model blind to history or to the present: they know how Weimar ended, and they have read the same declinist literature they are asked to score, so part of the TODAY reading may measure the salience of decline narratives rather than decline itself. That is one more reason the gauge leans on counted readings — the cradles, the trust surveys, the time-use data — and treats the models as a cross-check rather than a verdict.

And the gauge is deliberately blind to the factor that may decide the most: the continuity of institutions, whether the constitutional order itself survives — the seawall that separated America in 1929 from Weimar in 1929, and the one still, for now, holding.

The Renaissance Papers · Series I, part 2. Previously: The Fourth Pillar argued that renaissances are built, not wished for. This essay is about the crash that clears the ground—and the fork it forces. The map is here.

I’m a venture capitalist, and I’m telling you to root for the crash that torches my own asset class. I mean it. The valuations are silly, the data-center spending is feverish, and half the people I talk to are quietly bracing for the fall. But here’s the argument almost no one will say out loud: the coming crash would be the best thing that happens to this technology. And it’s a necessary part of the Renaissance cycle I wrote about last time: the pattern by which a disruptive new technology forces a society to rebuild its picture of what a human being is for.

Economist Carlota Perez has found a common pattern: every great technological surge of the last 250 years took the same course: a revolution, a financial bubble, a collapse, and only then a golden age. Canals, railways, steel, cars, and computers all ran it. AI isn’t bound to repeat that course, though. What makes the pattern more than coincidence is its mechanism: the crash after the bubble.

The bubble and the crash that ends it are not a detour around the golden age. They are the road to it. But if done wrong, the same turning point that built the American suburbs built the gulag. The only variable is how we respond.

Today, that variable is a choice almost no one is naming: when the rebuild comes, do we tax the rich or spread the means by putting the machine in many hands? This essay is the case for the second.

Subscribe now

Why we want a crash

The products of a technology bubble are more durable than the bubble itself. Technological fervor drags an absurd amount of capital into the new thing and pours it into infrastructure that no rational, sober investor would have built that fast. Britain got a railway network out of railway mania. The 1990s gave us the fiber that carried everything we built in the 2000s, laid by companies that mostly went bankrupt laying it. The investors lost their shirts; the rails and the fiber stayed.

We are living through the same process now, except the rails of today are compute, data centers, models, and the habit of millions of people learning to think alongside a machine. The chips will depreciate, sure, but the chips were never the rail; the power, the grid hookups, the data-center shells, and a generation that learned to work with machines are, and those outlive the hardware the way the fiber outlived the routers.

Then the crash does three jobs that nothing else can do.

It finishes the build. By the time the music stops, the new infrastructure is in the ground and the new way of working has become ordinary. The changeover is complete precisely because people overspent on it.

It sobers the money. A frenzy makes capital arrogant and stupid. A crash makes it humble again, and humble capital is the kind that builds real companies slowly instead of chasing paper.

And it forces the rebuild. In the boom, no one will touch the hard questions of how a technology should be governed, who it should serve, or what gets protected from it. There’s too much money in not asking. The crash creates the only moment of urgency in which a society will actually sit down and rewire its institutions. Antitrust, interoperability, and standards can do some of this, and we should use them now. But in the boom, they’re outspent by the people who profit from not asking. The crash clears the resistance. By 1935 we had deposit insurance, securities law, the whole architecture that made the postwar decades possible. None of that was conceivable in 1928.

The depression is optional

A crash and a depression are not the same thing. The crash is the moment the paper values snap back to real ones. The depression comes after, and its length is not fixed in advance.

A depression, in Perez’s telling, is just a crash whose rebuild failed or came too late. 1929 turned into the 1930s because the monetary orthodoxy held for four more years before anyone was willing to change the rules. That option belongs only to a society still whole enough to exercise it. That wholeness is state capacity: whether a government can still turn a decision into a result (Whether our state qualifies is the question I take up next).

The size of the bubble sets how deep the hole goes; how fast and how well we rebuild sets how long we stay in it. And depth cuts both ways. A downturn that’s too mild is its own failure: it never breaks the old habits, and you get what Perez calls a gilded age instead of a golden one: growth resumes, finance stays in charge, and the tensions never resolve. Half a Renaissance at best. But a depression that runs too deep and too long does the opposite damage by radicalizing people. Severity raises the ceiling and lowers the floor at the same time.

The bubble decides how hard the crash hits, left to right. We decide how well and how fast we rebuild, bottom to top. The same deep crash sits one move from Florence and one move from Weimar. And weak does not mean slow: Germany’s rebuild was fast, forceful, and aimed at the wrong portrait. A rebuild captured by the cheapest answer available is a failed rebuild, however quickly it moves.

The people praying for a soft landing—namely the Fed, the “this time is different” optimists, and the investors & founders hoping the music never stops—are all unwittingly praying for a slow death. We should not want the softest possible landing but a crash hard enough to break the current orthodoxy, met by a rebuild fast enough to keep the recession from curdling into something worse. Which means the answer has to be ready in advance.

This is the most important point, and the reason I wrote “The Fourth Pillar.” The crash does not guarantee you the golden age. It only gives you the chance at one.

The same 1929 collapse that helped produce the American postwar boom also produced Nazi Germany and entrenched Stalin’s Soviet Union. One society got the suburbs and the GI Bill; the others got the camps and the famine. What determines which side we land on is not the crash. It’s whether we’ve built the fourth pillar—a shared answer to what a human being is for—and built it correctly, before the crash arrives.

The first three pillars are showing up on their own, carried by the same economic gravity: the technology, the recovered knowledge, the concentrated capital. The fourth, the portrait of what a human being is for, has only ever been built in advance and installed in the narrow window when everyone else was finally paying attention because the easy money was gone.

The crash is coming, and it’s necessary. A crash is the crudest kind of error correction—what reality resorts to when criticism didn’t get there first. A crash destroys more than savings; it destroys the answer people were living by—the career, the equity, the number going up—and demand for a replacement spikes within months. Supply moves slower. A portrait of what a human being is for takes a generation to build: the books, the schools, the people worth imitating. So the crash never chooses the future; it elects for whatever is already on the shelf.

In 1933, America’s shelf held a portrait thirty years in the making—the progressives, the churches, the land-grant universities—and America got the New Deal. Germany’s shelf held the völkisch man; Russia’s, the New Soviet Man. Same decade, three shelves, three futures. That is what the fourth pillar determines. It is the answer when the question arrives; it’s the spec the new rules get written to, the ladder the freed talent climbs. Skip the work and the crash decides nothing on its own; the cheapest answer on the shelf wins by default.

Florence is built, not given

So what goes in those hands? The 1930s are the right place to look, because the Depression was never the crash itself. It was the years of things we failed to do after it. Recovery came only once the old orthodoxy broke and we built what should have existed already. The next crash will ask for the same kind of action.

The mechanical problem the 1930s exposed was simple: when a boom pulls all the income to the top, there is no one left to buy what the new machines can make. The reflex answer is redistribution—tax the winners, hand the proceeds back—but that will not work this time. The new wealth is mobile and the broad base owns none of it; when capital, code, and the most valuable work can move to a new state or country overnight, the money leaves faster than the law can reach it. Even a tax that landed would not rebuild what the 1930s rebuilt: a population with skin in the productive game.

So the fix does not run through the tax code. It is to spread the means, not the proceeds: put direct ownership of and cheap access to the productive layer into many hands: open models running locally instead of rented, compute and tools and equity at the source. Rented capability sends the value to the center; owned capability compounds it where people live. That, and a portrait of what the new life is for, is what the rebuild needs. Neither is built by decree, and the work falls to three sets of hands.

Government owns the guardrails and the floor. The guardrails are the rules it can still enforce—finance, antitrust, an open and competitive model layer—and its one real job is keeping exit cheap, so the layer stays open on its own. The floor is the fall it can still break: unemployment insurance, deposit insurance, retraining, the lights kept on. Catching the people who absorb the crash is not the same as taxing the people who do not, and a state too weak for the second still owes the first. What it should not do is write the portrait: the Soviets built the literacy, the recovered knowledge, and the state capital, then let the government define the human being, and got the New Soviet Man.

Companies own the breadth. Every day they choose between wide, cheap access and winner-take-all rent-seeking. Right now the most open frontier weights in the world ship out of China under an MIT license, and OpenAI (the one with “open” in the name) shipped no open-weight model between GPT-2 in 2019 and gpt-oss in 2025, moving only once cheap Chinese models made staying closed untenable. Do not expect virtue; expect arithmetic. A tollbooth is an invitation to build a road around it: the firms that widen the base capture the market, the ones that wall it off get bypassed. Keep exit cheap and breadth becomes the rational strategy on its own.

Everyone else owns the portrait and the stake, the two things no treasury will hand out. The portrait is the book, the school, the model person, the aesthetic, built and funded by many hands rather than legislated by one, not a single official answer but many in competition, kept honest by exit and criticism. The stake is ownership that outlasts the opening: getting computational equity, locally-run models and cheap compute, into many hands while the layer is still cheap, before it re-concentrates.

None of it can wait for the crash. The most expensive thing we did in the 1930s was exactly that—four years of denial before the rebuilding began, and in those same four years a continent radicalized. The crash hands us the urgency. It does not hand us the plan.

Florence ran on perhaps two thousand patrons. The Medicis built the schools, the workshops, the institutions that outlived them; today’s landlord buys back his own stock and collects rent on a walled platform that builds nothing new.

Be ready

I opened in fear of the bubble, and fear is just preparation that hasn’t been spent yet. The crash is coming whether we welcome it or not, and the only choice left is whether we meet it with a plan or with our hands open and empty.

A crash can scar instead of cleanse; the deepest ones leave lost output and a generation of lower earnings that never fully come back, and plenty of good firms die alongside the bad. The golden age isn’t promised either. And even when it has arrived before, the people inside it rarely called it golden: Renaissance Florence ran on plague and war, with life expectancy that fell by some counts to eighteen. Let me be blunt about what I’m calling for: people will lose jobs, savings, companies.

But slow rot is much worse. I know I’ll get pushback for where I’m standing: I’m insulated from the worst of this, and this argument should be judged on whether it helps the people who aren’t. Most people are unprepared, and the work is to make the fall survivable, not to savor it.

That work doesn’t fit a quarter. It fits the calm before the crash, because the window to build slams shut the moment it arrives. The full menu is in “The Fourth Pillar”: the book, a school, the open model that keeps the layer from closing, the net that catches the people who take the hit, the portrait of a life worth wanting on the far side. You don’t have to carry all of it. You have to carry one piece, on purpose, while the money is still easy and the room is still calm. The bubble decides when the question gets asked; we decide what’s already built when it does.

Let it crash. Just don’t be empty-handed when it does.

Subscribe now

Behind this: Carlota Perez, Technological Revolutions and Financial Capital. Davidson and Rees-Mogg, The Sovereign Individual. And my work “The Fourth Pillar.”

Acknowledgements: Thanks to Greg Bowman, Joel Dudley, and Lara Pande for their critical reading of this manuscript.

Image credit: Thomas Cole’s “The Course of Empire: Destruction”

The Renaissance Papers · Series I, part 1, where the whole project begins. The map is here.

Get this moment right and you get Renaissance Florence. Get it wrong and you get the New Soviet Man, and the twenty million graves that came with him.

Both grew from the same soil: a society hit by a technology so disruptive it had to rebuild its whole picture of what a human being is. We’re in one of those moments now. AI does more than automate tasks. It democratizes cognition, the kind of shock that, five times before in history, has created a Renaissance.

Each of those five rebirths began in disruption and fear. Each ended in a fuller vision of being human. The pattern is the same every time: a new technology of distribution recovers buried knowledge, aided by mobile patron capital, resulting in a new portrait of what a fully human life looks like.

I’ve spent thirty years at the AI-biology frontier: nearly twenty as a professor at Stanford building Folding@home, over ten as a VC. I see the first three pillars of a Renaissance (technology, knowledge, capital) in my work every week. Founders using AI to engineer biology, capital rushing toward new platforms, papers becoming products almost overnight. But there is no canonical school, no book to put in front of a fifteen-year-old, no aesthetic, no model person to name and imitate for this era. I believe this fourth pillar, the new portrait of humanity, is now up to us to build.

I see the absence most clearly in some of the best builders I meet. They are not short on intelligence, ambition, tools, or capital. They can tell you the model, the market, the wedge, the benchmark, the financing, usually in the first ten minutes. What is harder to name is the human being on the other side of it: what becomes stronger, wiser, more capable, more dignified because what they are building exists. That is not their failure. We have given them tools and capital before we have given them a portrait.

This is the fork in the road I opened with: Florence or the graves. The fourth pillar is what decides between them. It isn’t aesthetic decoration. It decides which future we get. I know that may sound grandiose, but I think it’s true.

Here’s what history says the work of building it looks like.

Subscribe now

Five Renaissances

“Renaissance” literally means rebirth. Rebirth requires death. Each of these transitions felt, in its moment, like the end of the world. Each turned out to be the start of a larger one, and a redefining of what it means to be human.

The Twelfth Century (~1100 to 1200). For four centuries the Latin West had lived in regression. Roman cities depopulated, with perhaps one in a thousand able to read at all, raiders pressing from three coasts, and apocalyptic preaching running through the chronicles. Then in a single century it broke open: Arabic learning translated at Toledo, Roman law revived at Bologna, the cathedral schools at Chartres and Paris matured into the first universities. The clergy lost their monopoly on literacy, yielding the scholastic magister and the Goliard scholar-poet, such as Abelard and Aquinas.

The Italian (~1350 to 1550). In 1347 plague reached Sicily; by 1351 a third to a half of Europe was dead. Then Constantinople fell, Gutenberg’s press shattered the Church’s monopoly on text, and two whole classes, the warrior nobility and the regular clergy, had to reinvent themselves or watch their identity dissolve. Out of it came the uomo universale: the Renaissance man.

The Industrial (~1770 to 1860). A Manchester weaver whose family had worked the loom for six generations watched his craft become worthless in a decade. Children went into the mills at seven. Working-class life expectancy collapsed in the new cities. The air was visibly dark at noon. The Luddites smashed the machines because they could see that their entire way of life was being destroyed. From this revolution emerged three contending answers to humanity: the self-made man, the Romantic craftsman, and the universal worker.

The Belle Époque (~1870 to 1914). Corporate hierarchies swallowed the independent farmer, shopkeeper, and craftsman in a single generation. Among intellectuals the mood was apocalyptic: Nietzsche announced God was dead, Marx said the worker had become a commodity, and Dostoevsky’s underground man already knew. And yet the era built steel, electricity, the corporation, Carnegie’s libraries, the OED, the new civic universities. Out of it came the gentleman-professional.

The Post-war (~1945 to 1975). Over seventy million had just died in World War II. The Holocaust had exposed industrial-scale evil hiding inside an industrial-scale civilization. The bomb made human extinction a possibility for the first time in history. Tens of millions left farms and small towns for the suburbs. Existentialism asked whether anything meant anything after the camps. Two answers emerged: the suburban family man and the self-actualizer.

Four Pillars to Building a Renaissance

The same four pillars appear at each of these transitional moments in history, and we can see the foundations of the first three forming again today.

A distribution shock. Every Renaissance starts when knowledge begins to move differently, through the pecia copy, Gutenberg’s press, the railway and the penny post, Carnegie’s libraries, or the mass paperback. Each shatters a gatekeeper’s monopoly: the Church, the academic guild, the publishing trade. AI goes further than any of them. Gutenberg printed the books; you still had to do the thinking. Now the thinking itself is cheap; a farmer in Iowa can have a research scientist’s reasoning applied to her soil at midnight, and a teenager in Lagos can have a private tutor on any subject at any hour.

A recovery of buried knowledge. Old material returns across cultural boundaries: Aristotle through Arabic, the Greek classics from falling Byzantium, the Romantic recovery of folk traditions, the post-war Great Books. Humanity now produces more knowledge than any institution can read, breakthroughs sit buried in obscure journals, and fields don’t talk to each other. AI is the rediscoverer for our era. AlphaFold handed the expertise of structural biology to chemists, pharmacologists, and materials scientists all at once. A thousand AlphaFolds are coming.

Concentrated, mobile patron capital. Wealthy families bet fortunes on cultural rather than financial returns: the Norman kings of Sicily on the Arabic translators, the Medici on the Florentines, Carnegie and Rockefeller on the Belle Époque, Ford and the GI Bill on the post-war recovery efforts. The wealth concentrated in technology today is the modern Medici equivalent: mobile, concentrated, free of institutional constraint. The question is whether that capital goes to Medici-style patronage of individuals or to winner-take-all platform extraction.

A new portrait of what it is to be human. Each era eventually produced one: the scholastic magister; the uomo universale; the self-made man, the Romantic craftsman, and the universal worker; the gentleman-professional; the suburban family man and the self-actualizer. It is the pillar the others rest on, and it hasn’t arrived for us. There’s no exemplar of the AI renaissance for a young person to look up to: no book, no school, no aesthetic, no model person.

The first three pillars arrive together, carried by economic gravity. New tech is a distribution shock by nature; cheap distribution surfaces old material; capital concentrates around new industries. They show up whether we help or not. The fourth is different. It has only ever been built by specific people who sat down and did the work: Petrarch, Castiglione, Smiles, Ruskin, Marx, Carnegie’s librarians, Maslow.

People feel this gap right now, even when they can’t name it, because AI is undermining the four historical defenses of human dignity at once. Reason is no longer an easy refuge if machines reason faster and more consistently. Making is no longer secure if machines generate images, prose, music, and code on demand. Labor is no longer enough if more of the labor moves to the machines. Even feeling is less protected than we thought, as AI mediates more of our attention and intimacy. Without a new answer, AI rips through our economy and leaves our self-understanding thin and defensive.

How to Build a Renaissance: the whole pattern, in one view:

What’s At Stake

Building the first three pillars without the fourth has gone very badly before. The Reformation produced a hundred and thirty years of religious war before humanism caught up. The late nineteenth century gave us robber barons and an imperial bureaucracy. But the fullest cautionary tale is the Soviet one I named at the start.

During the earlier Soviet period, literacy rates among Russian peasantry jumped from one in five to nearly 100% in two generations. Recovered knowledge from Marx and the European Enlightenment ran through the state libraries and the great Soviet science institutes. Patron capital came in the form of the entire apparatus of the state. And the portrait was as concrete as Castiglione’s: the New Soviet Man, taught through schools, posters, films, novels, and a hundred million words of explicit doctrine.

Where it went wrong is the checklist for our own work. The Soviets’ fourth pillar made the human a means rather than an end: an input to the five-year plan, not a person the plan was meant to serve. It allowed exactly one vision of what humanity should look like, enforced top-down across eleven time zones, with no rival models permitted to compete with it or test it. It treated everything that pre-existed the Party (families, faiths, villages, mother tongues) as bourgeois residue to be liquidated rather than inheritance to be built on. And it was unfalsifiable: when the harvest failed, the model wasn’t wrong; the kulaks were saboteurs.

Each of those four errors was lethal at scale. Twenty million graves were the cost.

Previous Renaissances found their answer to humanity by elevating what the dominant new technology couldn’t do. Printing could reproduce text but not judgment, so humanism elevated judgment, taste, and eloquence. Steam could produce power but not meaning, so Romanticism elevated feeling and nature. Mass production could make goods but not experience, so post-war humanism elevated authenticity and self-actualization.

AI can produce content, but it can’t be present. It cannot sit with a dying parent, keep faith with a town, raise a child, restore a house, or make a judgment that will cost it something. The fully human life of 2050 will have to relocate dignity there: into presence, embodiment, place, witness, care, and judgment with stakes. AI doesn’t strip dignity from us. It forces us to find it in what only we can do, which turns out to be most of what made human lives worth living before we talked ourselves out of it.

The Work

Knowing what the fourth pillar should be is the easier half. Building it is the work, but history gives examples for how that work can happen.

The model person. What people recognize, love, and imitate. The person of 2050 is not the best prompt engineer or the most optimized solo operator. They are someone who holds real things. They have stayed with a craft long enough for it to shape them back. They carry responsibility somewhere real: a school board, a town council, a lab, a classroom. They belong to a place: a piece of land, a workshop, a studio, a room where others gather. They are bound to a small number of people in long mutual obligations. They have inherited a tradition consciously, and they are passing it on deliberately. They hold themselves to a standard their grandchildren will know about.

The book. A portrait of the model person specific enough to imitate, in enough detail that a fifteen-year-old can ask “is that the kind of person I want to be?” and have an answer. The Italian Renaissance had Pico’s Oration for the philosophy and Castiglione’s Cortegiano for the portrait. We need both, and we don’t have either.

The school. What turns the portrait into real people at scale. Vittorino da Feltre’s Casa Giocosa, the educational north star of the Italian Renaissance, produced a generation of Renaissance princes. These schools come from new institutions, not reforms of the old. We have early candidates: focused research organizations, residencies, the great-books colleges, the new wave of small humanist schools, but nothing integrated, no shared curriculum, and almost no patient capital aimed here.

The aesthetic. The vision without anyone having to read the philosophy: perspective, the dome, the sonnet, the suburban home, the Esalen retreat. Most Florentines didn’t read Pico; they saw the David. Our era’s default aesthetic is flat, frictionless, optimized for cross-platform legibility. Someone has to commission the work that brings a new vision into focus.

The villain. Not necessarily one individual, but what happens when the dominant system of the age runs unchecked. The medieval Church was the villain of Renaissance humanism. The dark satanic mill was the villain of Romanticism. The organization man was the villain of post-war self-actualization. Our candidates are already visible: capability concentration hardening into capture. The foundation-model layer becoming the new monasteries, deciding whose questions are worth answering. Attention itself engineered into compulsion. Synthetic intimacy crowding out the real kind. Cognitive deference until we lose the capacity for thought.

Put together, this is a portrait of civic and creative agency, but also of warmth, judgment, and presence. AI can do more of the mental labor; humans decide what deserves to exist, who it serves, and what costs they are willing to bear. Craft, service, taste, and responsibility are what make a life visible. I expect there will be better portraits than this one. The point is that there has to be one, a portrait of someone carrying weight in the world, not of someone merely optimizing themselves.

Ours to Forge

None of this is impossible. Most of it isn’t even expensive. Florence had maybe two thousand patrons at the peak. The full list of Renaissance humanists worth naming fits on a single page. It doesn’t take many people. It takes the right people, doing the right work, on purpose.

This work doesn’t fit in a work day. It fits a decade. And it has always been done by specific people who picked one piece and stayed with it. The book. A school. The aesthetic. The failure mode that needs naming, with weight behind it. The model person worth mentoring into existence. Pick the one that’s yours. Then hold it long enough to matter.

The new printing press is being invented. The capital is here. The buried knowledge is coming back to light. Three of the four pillars are already standing. The fourth has always been built by people who decided to. Some of them are reading this.

No Renaissance before ours knew it was one. We might. That gives us less excuse, not more certainty. Let’s build it on purpose.

Subscribe now

Behind this: Charles Homer Haskins, The Renaissance of the Twelfth Century. Ada Palmer, Inventing the Renaissance. Richard Goldthwaite, On the Economy of Renaissance Italy. Carlota Perez, Technological Revolutions and Financial Capital. Davidson and Rees-Mogg, The Sovereign Individual. Will Durant, The Renaissance and The Age of Faith.

I use “Renaissance” loosely. Most people use it for the Italian one; I’m using it for the pattern, which only comes into focus once you see it repeat. Call the others whatever you like: paradigm shifts, great surges, rebirths. The point is what they share. I picked these five for the clarity of their distribution shocks and knowledge recoveries. I draw most heavily from European cases because they’re the examples I know best; the same pattern arguably shows up in Tang China, the Gupta period, Abbasid Baghdad, the Bengal Renaissance, Meiji Japan, and Akbar’s court in India; and the European 12th-century renaissance was itself downstream of the Islamic one.

But knowing these cycles exist is only half the work. The challenge is two-fold: understanding each cycle, and understanding that the future will have similarities and differences due to how the cycles overlap. The business cycle of boom and bust, for example, has different nuances in the beginning of an industrial revolution versus and end, and even more complexity when laid upon the timescale of the rise and decline of nations and empires.

And of course, there will be fluke, random elements which simply cannot be predicted such as a major figure prematurely dying or some natural disaster, and in this case one has to admit and incorporate these possibilities into an understanding the error bars on whatever one might predict.

But with that said, the future has been, at least historically, quite predictable. You see this in Carlotta Perez’s work on the history of industrial revolutions predicting how the internet played out, Ray Dalio’s work on cycles in markets and nations actively used in his hedge funds, and James Dale Davidson and William Rees-Mogg’s The Sovereign Individual predicting various trends in tech, including crypto currency.

Once you see the cycles, you can’t unsee them — and then the question becomes what can you do from there.

I joined a16z in 2014 as a Professor-in-Residence, just coming off my time as chair of Biophysics and Henry Dreyfus Professor of Chemistry at Stanford. What started as an observation that tech companies are poised to make major impacts in Bio and Health quickly evolved into something much larger: my founding of the a16z Bio + Health fund, a new strategy at the intersection of AI, computation, and biology — and the chance to build an investing platform from the ground up.

I’m also proud not only of the companies I’ve backed — including Devoted Health, Function Health, Insitro, Freenome— but also of the team I built to carry this vision forward. What began as a speculative bet became a thriving and deeply mission-driven practice, thanks to the people who believed in it and helped shape it every step of the way.

This transition could not have been possible without the incredible people I’ve worked with:

● I want to thank Marc and Ben for taking a bet on me—and on the idea that biology could be engineered, and that venture capital had a role to play in building that future. Their support and belief shaped everything that followed. We should all be proud of the impact we’ve had shaping the future of health.

● I’m deeply grateful to the team I’ve helped build in the Bio + Health practice: Jorge Conde, Julie Yoo, and Vineeta Agarwala. It’s been an honor to work alongside them to build this platform from the ground up.

● To our founders, thank you for your courage, your brilliance, and your trust. Your creativity, commitment and energy to building what we all know is possible is nothing short of inspiring and contagious.You’ve challenged me, inspired me and reminded me why this work matters so much. It’s been one of the greatest privileges of my career to support you.

● To our limited partners, thank you for believing in this vision and making it possible to pursue it with long-term conviction.

Though I’m stepping away from the firm, I’m not stepping away from the cause. If anything, I feel more energized than ever about the future of bio + health. Writing the next chapter of the playbook in healthcare will require a deep imagination, more cross-disciplinary collaboration, and a willingness to build boldly—and I look forward to being part of that in new ways.

Some of you may remember a talk I gave called Disrupt Yourself — and true to that message, I’m doing just that. I’m heading into a new chapter focused on the future of AI, healthcare, and prevention.

With deep gratitude,

Vijay Pande

One of the most fascinating of these larger cycles is the cycle of an industrial revolution. My favorite historian on this topic is Carlota Perez, who has studied past revolutions in steam, steel, electricity, and IT, revealing the common elements that emerge each time.

It begins with a new technology that only a few people in the world truly understand. In the 1920s, it was electricity and automobiles; today, it’s AI. The fact that only a small group can build and benefit from such technology results in something akin to the Gilded Age of the early 20th century, where income inequality grows as those with expertise accumulate immense wealth. This is the era of new giants emerging—new winners and new losers.

In this period, enormous fortunes are made by those who take a full-stack approach: not only creating the technology but also controlling its applications. Today, we see this happening in the AI space. The companies that can integrate AI into end-to-end solutions are positioned to capture the greatest value. Eventually, though, the technology broadens to reach the masses, and the ability to sell just parts of it—for others to incorporate themselves—becomes the norm.

As the technology becomes commonplace, we reach what Perez calls "The Golden Age"—a time like the 1950s when income inequality lessens because fewer people have a disproportionate grasp of the technology of the era. This is the time when governments flourish, as working for the government or private sector offers similar compensation, attracting the best and brightest to public service. I believe we will eventually reach a Golden Age with AI as well, but today, we are in a period akin to the 1920s, brimming with upheaval and opportunity.

In just a decade, the 1920s saw the transition from horses and ice deliveries to cars, electricity, and modern appliances—technologies we take for granted today. The horse experts of that era became obsolete, replaced by those who could understand automobiles. We are on the cusp of a similarly dramatic transformation.

While history may be clearer in hindsight than foresight, it still offers important lessons for our present. What massive changes are we yet to see? The obvious answer is that AI will replace some jobs and lead to the creation of new ones. But I think it’s more complex and subtle than that. Just as the master horse whisperers lost their prominence, I suspect the people whisperers—those skilled at managing and influencing others—will also become less essential.

AI will serve as both an extension of people and as a translator, enabling much smaller teams to achieve what once required large organizations. It could also bridge the gaps between different work styles, allowing individuals who approach problems in very different ways to collaborate more effectively. It’s a vision of a world where workers focus on their work, not the frictions of human dynamics.

Perhaps what’s most exciting, though, is the part of the future we can’t predict. We are in the midst of building a new modern world—one that our children and grandchildren will one day take for granted, much like we take electricity for granted today. While we can anticipate many changes, surprises are inevitable. By thinking deeply about these cycles, maybe we can catch a glimpse around the corner at what’s coming next, preparing ourselves to thrive in the face of the unknown.

AI as the Ultimate Flexible Partner

Imagine a workspace where AI doesn’t just perform tasks but adjusts to human needs, personalities, and energy levels. With its ability to learn from vast amounts of human data—our conversations, interactions, and behavioral nuances—AI could become the ideal work partner for both introverts and extroverts. This possibility isn't about AI replacing humans; it's about AI enhancing our work experiences by complementing our natural tendencies.

For extroverts, who thrive on social energy and constant engagement, the concern might be that a world filled with AI could become isolating—a dreaded scenario of digital loneliness. But this underestimates the transformative potential of AI. Trained on human dialogue and behavior, AI can simulate the social interactions that extroverts crave. It can chat, joke, debate, and bring a sense of vibrant company, even if no other humans are present. Extroverts could find themselves surrounded by a “party” of engaging AI personas, each offering the dynamic back-and-forth they need to stay energized.

A Personalized Social Environment for Introverts

On the other hand, introverts often feel overwhelmed by too much interaction. For them, recharging means finding solitude or minimizing stimuli. One might wonder: if AI can mimic human interaction, would that be equally exhausting for an introvert? The answer is no—not necessarily.

AI doesn’t need to behave like a human. For introverts, AI could adopt a quieter, more restrained style, tailored to their preference for calm, efficient exchanges—avoiding the awkward small talk that’s often incorporated into every work interaction. The beauty of AI is in its flexibility—it can embody the full spectrum of human and non-human communication styles. And for introverts, this might mean working with AI in ways that feel more like interacting with a machine than a human, relieving social pressure while maintaining productivity.

The True Potential: Blending Styles Seamlessly

But the real magic happens when AI brings these contrasting styles together. In teams that include both extroverts and introverts, AI could act as a bridge, adapting to each individual's preferred mode of communication and interaction. For the extrovert, the AI could present itself as a lively, engaging persona. For the introvert, the same AI could tone down its energy and engage in quieter, more focused interactions.

This ability to create personalized experiences could enhance collaboration within diverse teams. Extroverts might never feel isolated or under-stimulated, while introverts wouldn’t feel overwhelmed by too much interaction. And because AI could adjust its communication style on the fly, it could “translate” between different team members' preferences, fostering harmony in ways that are impossible for human-only teams to achieve.

A New Way to Work Together

As we move into a future where AI plays a larger role in our work lives, we should think beyond just the technical capabilities of AI. Its potential lies in how it can adapt to the human element of work—our personalities, our communication styles, and our need for different kinds of interaction. AI has the power to make work more fulfilling for all types of people by creating environments that stimulate extroverts and relax introverts, without sacrificing the social aspects of teamwork.

This dynamic capability of AI will change the way we think about collaboration. It’s not just about automating tasks; it’s about enhancing the human experience of work. AI can become a partner that not only understands the content of what we’re doing but also how we’re doing it—and why that matters to each of us.

In the end, AI could become the ultimate glue that brings together diverse human styles, making us all better collaborators, regardless of where we fall on the introvert-extrovert spectrum.

Subscribe now