Subscribe now

Then Bogle started beating his competitors, for a reason closer to arithmetic than skill. Everyone invested in the market, added together, is the market. Since the actively managed funds charged high management fees, the average fund earned the market returns, minus the fees. Bogle’s fund also earned the market returns, because he just bought the whole market. But since he charged lower fees, he beat the average fund automatically.

The average active dollar doesn’t trail the index because the manager is dim. It trails because the manager is expensive. Bogle built the company to drive that cost toward zero. By the time Bogle died in 2019, Warren Buffett was writing that if a statue were ever erected to the person who had done the most for American investors, “the hands-down choice should be Jack Bogle.” The folly had become the orthodoxy, and the orthodoxy was correct.

It is still the most responsible, boring thing you can do with a paycheck. Don’t pick stocks. Don’t try to beat the market. Don’t pay fees to the clever broker1 in the good suit. Buy everything at once, payroll-deducted, preferably in your 401(k), for forty years, and never look at it until you retire. More than half of American households now have money in the market, much of it invested in roughly this way, most of them having never heard Bogle’s name. They followed the advice. They bought diversified index funds.

What almost nobody noticed, including, for most of his life, the man who built it, was that the advice worked because the people taking it were a minority. An index fund has no opinion about whether anything is cheap or dear, and that indifference is the entire source of its discipline and its microscopic fee. It free-rides on everyone else’s pricing: it lets the stock-pickers do the sweating and the arguing over what Apple is worth, and then simply buys their answer at whatever the answer turned out to be. That free ride is a marvelous deal when the rider is a small passenger on a very large market. It becomes something stranger when the passenger grows into the vehicle. Bogle himself, in the last years of his life, was the one who said it out loud: in a 2018 op-ed he warned that a handful of index managers were on track to own “30% or more of the U.S. stock market—effective control,” and that he did not believe “such concentration would serve the national interest.” The patron saint of the index spent one of his last public breaths worried about over-indexing.

He was right about that too, and you can see it in your own account. The fund still carries the same Vanguard name, the same trivial fee, the same reassuring pie chart. But it no longer holds what the pie chart says it holds. What was sold as the broadest, safest bet in finance has quietly become a bet on a handful of names—and concentration is only the first of the surprises. The bigger issue is what it’s done to prices.

The Thermostat

A price is information, and the cleanest place to watch it do its work is a market too dull to be worth rigging: the dairy aisle. Say milk gets expensive. That isn’t the start of a problem; it’s the start of the cure. A high price is a message—make more milk—sent to exactly the people who can. So they do. Dairies breed more cows, and a season later there’s more milk than the market wants, and the price drifts back down. The signal that announced the scarcity is the thing that ends it. Push the system and it pushes back. Economists call this negative feedback. You can think of it as a thermostat: the reason a market doesn’t overheat, quietly at work behind every honest price you have ever paid.

And there are two cures in that, not one. The high price pushes the dairies to make more milk—that’s supply. It also pushes some shoppers to buy less, to reach for the soy milk or the orange juice or to skip it entirely—that’s demand. More coming in, less going out, and the price has nowhere to go but back down. Two brakes, working opposite ends of the same number, and both of them just the market responding to a real fact: milk is scarce, do something about it.

A share of stock is built to ride those same two brakes. When a company’s stock gets expensive, it can do exactly what the dairy does when milk gets expensive: make more. It issues new shares at the rich price and plows the cash into the actual business—more stores, more trucks, more R&D, more output. More cows, in other words, and more milk, until the price eases back toward what the now-bigger company is worth. That’s the supply brake. The demand brake is a person: the investor who looks at a stock priced at a hundred and thirty times its earnings, decides that is insane, and won’t pay it. His money goes somewhere cheaper instead, and every buyer who walks away is a bid the price loses—the same downward tug as a shopper reaching for the soy milk. Two brakes, the same as the dairy, and the market is built to ride them both.

But just as Bogle feared, today the market is riding neither. One brake has been wired to run in reverse. The other has been overrun—the thin line of investors who still care about price, swamped by an army of passive index money that doesn’t. What’s left is a price that no longer answers to what anything is worth. The one signal it still reflects is the sheer volume of money pouring in.

Culling the Herd

Start with the supply brake—the one running backward.

To see how that happens, forget stocks for a moment and picture a rancher. Imagine he isn’t paid a salary; he’s paid in cattle, and his bonus rises and falls with the price of a cow. When cattle get expensive, the textbook says he should breed more of them and sell into the high price—but he won’t, because a barn full of new calves is exactly what would drive the price back down and gut his own bonus. So he does the opposite: he thins the herd. A smaller herd means scarcer beef, a higher price per head, and a fatter check for him. The response that’s supposed to cool an overheated market has quietly reversed into one that stokes it, because the man at the lever is paid by how high the price climbs, rather than how much value is produced.

This is almost exactly how a modern public company is run. Its executives are paid in stock—options, grants, packages that vest on the share price—so the share price isn’t one of their concerns. It is the concern. And the fastest way to lift a share price is not to build a better business, which is slow and uncertain and may not pay off until long after the options have vested or the executive has moved on. The fastest way is to shrink the number of shares. So when the stock is expensive, instead of doing the textbook thing—issuing new shares, raising cheap capital, plowing it into something real—the company does the reverse. It spends its cash buying its own shares back, hundreds of billions of dollars a year, retiring them, thinning the herd until each surviving share is a claim on a slightly scarcer thing.

None of this is a crime, or even necessarily foolish. Handing cash back to shareholders is respectable enough, and a buyback doesn’t change what the business underneath is worth—it’s the same company, just sliced fewer ways. But the executives running it have no incentive to breed more calves. An expensive stock is supposed to summon something real into existence: new capacity, new products, more of the actual thing the company makes, more milk. A buyback summons none of it. It produces no new milk at all—more earnings per investor, but not a penny more overall. The supply brake didn’t fail quietly. It was rebuilt into a valuation accelerator, by people whose pay depends on the pedal staying down.

The Bid That Doesn’t Read Prices

The demand brake is stranger, because it hasn’t been removed at all. The skeptic who’d refuse to overpay is still right there refusing. He’s simply been swamped—by a buyer who shares none of his hesitation because it hasn’t got an opinion to hesitate with. A passive index fund does not judge whether anything is cheap or overpriced. It can’t; that indifference is the whole source of its discipline and its microscopic fee. Money comes in, and it buys every company in proportion to what that company already costs, at whatever today’s quote happens to be, on the fifteenth and the thirtieth, or whenever your withholding hits your account, whether the market is on a tear or on sale or on fire. It is the purest price-insensitive buyer ever assembled at scale: a bid that never once asks what it’s paying. And it is no longer a niche. Index funds overtook the active stock-pickers in total assets around 2024 and now manage more than nineteen trillion dollars between them—grown from the eleven million Bogle’s fund scraped together in 1976.

On its own, that’s harmless. A market-cap fund is a proportion-keeping machine, not a concentration machine. If a company is seven percent of the market, your dollar puts seven cents into it—you’re not piling into the winner, you’re just holding your slice. Left alone, a price-insensitive fund would sit there in perfect proportion forever and distort nothing.

But it is not left alone. It is fed—every two weeks, payroll-deducted, automatic, forever. And a buyer who pays any price and never stops is almost impossible for the other side to discipline. Nothing matches it on the sell side—no automatic seller dumping shares every payday, no matter the price. Only people, and people run out. The skeptic who thinks the giant is wildly overvalued can sell his shares. Once. Then he’s out of stock and out of ammunition, and the bid is still there on the thirtieth, and the fifteenth after that. To actually walk the price down, you wouldn’t need one skeptic, or ten. You’d need essentially everyone who still cares about value to sell at the same moment, in concert, into a bid that keeps buying through all of it—and they never coordinate. Each one’s sale is swallowed by a nation’s worth of 401(k) money buying that same morning. The price stops being something the market negotiates its way to. It becomes whatever the bid is, and the bid isn’t looking at the price.

A finance professor might object that index funds are a majority of the market’s assets but only a sliver of its daily trading, and that prices are set by whoever’s trading. True, narrowly, and beside the point. Most of that trading volume is active managers selling to each other, one fund handing Nvidia to another, churn that nets to nothing and discovers no new price. What moves the level isn’t the churn; it’s the net direction of the flow, and the net direction is a one-way street. Volume is the noise of a thousand people swapping the same shares around. The flow is the quiet, relentless, price-blind flood underneath it, driving everything higher.

The man whose job is to lean against this—the skeptic, the brake—doesn’t just get outnumbered. He gets destroyed for being right. He looks at the giant at thirty-five times earnings, judges it overpriced, and positions against it, and then he gets run over by twelve straight quarters of payroll-funded buying that does not care that he is right. The active managers watch the bears die on the side of the road, and decide—of course—they’d better move with traffic. The brake doesn’t just fail. It gets fed into the engine. The market used to carry the memory of every prior mania in the form of someone willing to bet against the next one. Price-blind money is amnesiac.

Drown the price-sensitive seller—not remove him, just drown him—and the index level stops measuring anything but the size of the bid. How big is the payroll deduction, the 401(k) contribution bid this month? There’s still a number getting printed. It just doesn’t have much to do with the performance or value of any given investment in the market. Which leaves the one question the whole machine depends on: how does a bid like that never run out of money to spend?

The Bottomless Bid

A bid that never runs out of money needs a source that never runs out of money. There is one, and it isn’t a figure of speech. The government spends far more than it takes in, and it covers the gap by selling bonds; when there aren’t enough willing lenders to absorb them all at a tolerable rate, the central bank buys the rest, with money it never had and never collected. A bond goes in, and dollars that didn’t exist that morning come out—printing, with a couple of respectable steps in between so that nobody respectable has to say so. Dollars made that way pile up faster than real production to spend them on, which grows at maybe two or three percent a year, and the surplus has to land somewhere. It can’t sit in milk and gasoline, where a high price just summons more supply and sinks back down. It pools instead in the things no one can make more of on command—shares, land, the scarce and the already-built. It goes to assets.

There is supposed to be a limit on this, and the limit is the interest rate. Make borrowing expensive enough and borrowers retreat—you don’t buy the bigger house when the mortgage rate doubles, the company shelves the new plant, and the flow of fresh money slows at its source. That works on every borrower who flinches at a price, which is every borrower but one. When the rate on the government’s debt doubles, it doesn’t borrow less; it borrows more—enough to cover the spending it was already going to do, plus enough to cover the extra interest. Today, the interest bill alone is already past nine hundred billion dollars a year, more than the entire military, the largest on the planet, and every uptick gets rolled into next year’s deficit and financed with even more debt. So the one tool built to slow borrowing down speeds this borrower up: it is the single actor in the economy that answers a higher price by buying more. You can’t raise the rate high enough to shut it down, because the rate is the thing that turns it up.

And the delivery is automatic. A retirement system built over those same decades routes a slice of every paycheck into the index by default—payroll-deducted, target-dated, rebalanced by software, never once pausing to ask what anything costs or what it’s worth. The money that has to go somewhere meets a machine designed to buy regardless of price, on schedule, forever. That is the bottomless bid: not a verdict the market reaches, but the default setting on a hundred million paychecks doing the “responsible” thing.

Thirty Years of Profit

Of course, the responsible thing isn’t to buy a company at any price. Sophisticated investors value companies in proportion to their earnings. A price-to-earnings multiple is just a number of years: it’s how long the company would have to hand you its entire profit, every dollar of it, before you’d have gotten your money back. At ten times earnings, you’re buying ten years of the company’s whole output to break even. That’s a deal that pencils. At thirty times, it’s thirty years—you’ll have your capital returned around the time you retire, assuming nothing goes wrong for three decades. At sixty, you’re not really hoping to get your money back, so much as you’re hoping your grandchildren do. But today, nobody flinches at stocks that trade at a P/E of 60, even though anyone would be suspicious of the equivalent statement: you are pre-paying sixty years of the company’s entire profit.

To be fair, investors don’t just buy a company for its current earnings; they buy it for its growth, and if profits compound fast enough, a thirty-year sticker pays back far sooner. For a single exceptional company that can be right—Tesla has spent years priced at several hundred times its earnings, and sits north of three hundred even now, a number that only resolves if you assume it conquers industries it hasn’t entered yet. Maybe it will. But the index can’t make that argument, and the index is what the price-blind money buys. One company can grow into a heroic multiple; the whole market cannot, because aggregate earnings are roughly chained to the size of the economy and margins can’t climb past a hundred percent. So when the aggregate market trades at thirty-two times current earnings—more than thirty years of payback, past forty once you smooth a full business cycle—no story about disruptive growth rescues it. A high multiple doesn’t hand you the growth; it means you’ve already paid for it, and your return now rides on reality beating an expectation that’s already priced optimistically.

History keeps the receipts on this, and it keeps them in a graveyard of names that everyone once thought of as invincible. In March of 2000, Cisco Systems was the most valuable company on earth, worth more than half a trillion dollars, trading north of two hundred times its earnings. A person who bought at that price waited more than twenty-five years just to see the stock back where he’d started—not adjusted for inflation, just nominally, a quarter of a century to break even. And Cisco won. It’s still here, still profitable, still selling the plumbing of the internet. The company was never the problem. The price was. Microsoft is the strongest case the optimist has, the firm that genuinely grew into its promise and became one of the great businesses of the age, and even Microsoft handed its end-of-1999 buyer the better part of seventeen years of dead money before the stock reclaimed its old high. That is the reward for being completely right about the best franchise of the era. You simply paid too much for it, and “too much” took a decade and a half to forgive.

So paying a few hundred times earnings isn’t optimism. It’s a bet that this company sits in the thinnest sliver of corporate history—the handful that compounded at extraordinary rates across an entire human lifetime—and a bet priced as though the sliver were a certainty. The price-blind money makes similar bets every two weeks, for huge multiples on an index highly concentrated in a few names, on your behalf, without anyone in the chain ever asking whether the odds are any good.

The Diversification That Isn’t

The antidote to Cisco or Tesla or any other company failing to live up to irrational exuberance is simple. Diversify.

And diversification is the other selling point in the indexes’ favor. In principle, you bought the fund so that you wouldn’t have all your eggs in a few baskets. Except, today, the ten largest companies are nearly two-fifths of the entire index—more top-heavy than at any point in half a century, longer than most investors have even been investing, and more concentrated than the market was at the peak of the dot-com bubble. Weight every company equally instead of by size and the index multiple drops by roughly a quarter, the widest that gap has run in the fifteen years the data covers. The headline valuation isn’t a story about five hundred companies. It’s a story about ten. The fund that sold you safety-in-numbers has quietly become a concentrated bet on a handful of names.

Cap-weighting invests your money in proportion to price, not earnings. The more the market falls in love with a company, the higher its multiple climbs and the heavier it sits in the index—so the fund buys the most of exactly the companies whose price already assumes years of perfection, the ones with the most that can go wrong. That might suit a desk running high-variance bets with analysts watching every quarter. It’s a shaky foundation under a nation’s retirement savings.

The problem with high growth expectations isn’t the growth, or even really the expectations. It’s the results. Eventually, and inevitably, companies fail to deliver on the hype. A quarter will come that will disappoint. There are some number of GPUs, or some number of Teslas, or some number of data centers that will simply be enough. The market will be saturated. In a low-multiple business, the price corrects down proportionally, and a well-diversified portfolio muddles on. But in high-multiple businesses, a disappointment doesn’t simply reprice this quarter, it reprices expectations over the next several decades. And in highly concentrated, and highly correlated, indexes the disappointment tends to be contagious.

That’s the downside of being only a little wrong. The crueler half is the upside of being mostly right: a company priced for perfection that delivers perfection has merely performed to expectations already priced in, and the multiple drifts back to earth anyway. Disappoint, and the concentrated, correlated names fall together. Deliver, and the gain was prepaid. Heads you lose; tails you don’t really win.

The Boom Loop

Market crashes always earn a sort of infamy among investors. Black Monday, the Dot-Com Crash, the GFC. We remember the sinking feeling when suddenly everything is repriced down, dramatically, all at once. The memories are visceral, emotional, almost physical.

But we always forget the crash up. The boom loop that preceded each of those notorious days. It runs like this: a price-blind bid pushes the market up, and the rising market makes indexing look like the only safe choice, which pulls in more money, which widens the bid and pushes the market up again. The higher the number climbs, the better the advertisement—every record high brings in the next dollar, and the next dollar lifts the price again. And nothing in the machine leans the other way: a government that borrows more the more it costs to borrow keeps the cash coming, the cash floods into assets because there’s nowhere productive left for it to go, and the index funnels it into fewer and fewer names at higher and higher multiples. Every part feeds the next, and the whole thing runs one direction. Up.

At the end of the day, there’s exactly one surefire way to lose your investment. Buy high, sell low.

So the next time the index closes at a record and the headline calls it a resilient economy that just won’t quit, don’t ask whether it’s a bubble—that’s the old question, and it’s asking for a date nobody can give you. Ask the harder one: what does the price signify, and what is the company worth? You’ve been buying a smaller and smaller share of the same economy, at a longer and longer payback, through a machine built so that nobody in it ever has to decide whether it’s worth it. The number goes up. But that part was never in question.

The thing the public was being asked to worry about was that some kid with a laptop and a chatbot might one day make a hit movie without paying Hollywood dues first.

Subscribe now

This is a real concern for the incumbents. The Hollywood ecosystem is one of the most concentrated wealth pyramids in America. WGA minimum scale for an original screenplay starts at roughly $80,000 and tops out near $150,000—a floor that pays the worst-paid WGA member, on a single project, more than the median American household earns in a year. Working SAG-AFTRA actors clear $1,200 a day at scale. Established writers clear seven figures. A-list actors clear nine figures per blockbuster. The studio executives clear ten. Every union victory in this fight defends an “injustice” that most Americans can only dream of.

In some sense, the strike won. The clauses are in the contract. But it’s a Pyrrhic victory. It does not stop a kid in Austin with a laptop from making the next breakout hit with a chatbot. It only guarantees that when one does—and one will—Hollywood will be contractually obligated to keep making movies the slow, expensive way while the rest of the world does not. The victory locks in a cost disadvantage. That’s the real story, and what happens next is predictable.

The Argument Has Been Made Before

“This new technology will destroy jobs” is a reliable feature of every wave of technological displacement in history. It’s also, in retrospect, the side of the argument that loses every time, at least in the long run.

In 1894, New York City had about 150,000 horses producing 2.5 million pounds of manure a day. The manure-removal industry was substantial. So were the stable industry, the carriage industry, the feed industry, and the long chain of small businesses that assumed horses were essential to how cities worked. When the automobile arrived, every one of those industries sounded the alarm—in the language of public safety, public health, and public morals. Cars were too fast. Cars were too loud. Cars would frighten children. New York seriously debated, and in some places imposed, laws limiting cars to walking speed. The carriage drivers were the loudest voice, because the carriage drivers had the most to lose.

You know how this ended. The displaced industries are footnotes. Nobody alive thinks New York should have stuck with the horse, and nobody alive even remembers the smell. They don’t have to.

The travel agents went the same way. When Expedia and Travelocity arrived in the late 1990s, the trade press filled with warnings about lost expertise, fraudulent bookings, missed connections, and the death of personalized service. Some of the warnings were even true. The industry lost about two-thirds of its agents within a decade, and air travel became cheaper and more accessible than at any point in human history. The people who lost their jobs deserved sympathy. What they didn’t deserve was a regulatory regime requiring human review of every online booking, externalizing the costs of their irrelevance onto every traveler everywhere.

The coal miner of the 2010s heard the same melody in a different key. His industry needed to die for the climate, the editorial pages explained. The just-transition support would arrive. He could retrain into solar installation, or move to a city, or—the favorite line—learn to code. He didn’t get the just-transition support, of course. The solar jobs went to West Texas, not West Virginia. The towns hollowed out. The editorial pages turned to the next round of advice for the next group getting displaced, as they always have.

This is creative destruction. The labor market adjusts. Progress requires winners and losers. Only the losers fail to see it.

This Time It’s Different

Every previous wave of displacement automated muscle and motion. The loom, the assembly line, the container ship, the self-checkout—these replaced work that was physical, done by people whose labor was too. The advice was always the same: get an education, work smarter, not harder. Predictable advice from the class whose work is done with words, numbers, and credentials. The one that doesn’t have to worry about where their electricity comes from, just as long as it comes when they power on their laptop.

AI runs the other way. It automates the words and numbers. It commoditizes exactly the kind of work the credentialed class does—drafting, summarizing, analyzing, reviewing, classifying, recommending, advising. The knowledge workers. The first jobs visibly threatened by AI are not in coal towns or factories, this time. They are in law firms, marketing departments, junior consulting bullpens, radiology clinics, and editorial offices.

Which means, for the first time, the people inside the displacement zone are the people writing the columns explaining why displacement is good. Except, of course, this time they say it’s bad.

They are not a small group. They run the New York Times and the Atlantic and most of the journalistic institutions still standing. They run the universities and the academic journals. They run the state and federal regulatory agencies. They run the major NGOs and the major foundations. They run the legal profession, the medical licensing boards, the accreditation organizations, the standards bodies, the HR profession, the compliance industry, and the policy think tanks. They’ve run the executive branch of every administration since Eisenhower.

The state bar association can require, with great moral clarity1, that every AI-generated legal brief be reviewed and certified by an admitted attorney. The medical boards can mandate physician sign-off on every AI diagnostic recommendation. The academic journals can declare AI-assisted scholarship a form of academic fraud. The HR director can write the policy requiring human review of every algorithmic hiring decision. The compliance officer can have AI outputs themselves require compliance officers.

What every previous displaced class did was complain. What this one can do is mandate.

The Class Schumpeter Predicted

This isn’t accidental, and it isn’t new. The dynamic was anticipated, in unsettling detail, eighty-four years ago.

In 1942, the Austrian economist Joseph Schumpeter published Capitalism, Socialism and Democracy. The phrase the book is famous for—creative destruction—has been so thoroughly absorbed by business journalism that it’s become cliché. The book’s actual subject is something different. Schumpeter argued that capitalism would be destroyed, not by its failures but by its successes. The specific success in question was the expansion of higher education.

A wealthy society can afford to credential more and more of its young people. But the economy can only absorb so many credentialed workers into status-appropriate roles. The surplus, Schumpeter wrote, would “swell the host of intellectuals in the strict sense of the term whose numbers hence increase disproportionately. They enter it in a thoroughly discontented frame of mind. Discontent breeds resentment.”

This stratum—what we might call the technocrats, or the professional-managerial class, or the credentialed class—would become the system’s most articulate enemies. Not because they suffered the most under capitalism. Because they suffered the contradiction of being trained for a status tier the system didn’t need. And their hostility, Schumpeter insisted, was structural rather than principled. The intellectual works in a domain that isn’t, in any meaningful sense, tested by markets. Ideas are evaluated by other intellectuals. Tenure committees grade tenure candidates. The peer reviewer reviews the peer reviewer. A system in which ideas are evaluated by whether anyone voluntarily pays for them is intolerable to the intellectual class, because the intellectual’s entire self-concept depends on having superior discernment to the masses. The market, on the other hand, caters to the masses.

Schumpeter thought this class would, in time, try to kill capitalism and replace it with socialism. He got the first half almost right and the second half almost entirely wrong. The credentialed class didn’t need to overthrow capitalism. It found something more durable. It routed around the market by colonizing the institutions that produce the market’s rules. The regulatory state, the academic credentialing apparatus, the licensing boards, the standards organizations, the major foundations, the prestige media, the HR profession, the compliance industry—every layer of intermediation between economic activity and economic outcome—would be staffed and run by this class. Their labor would be insulated from market discipline not by abolishing markets but by making themselves the gatekeepers and toll collectors of every transaction important enough to matter in the market.

This is, more or less, the world we live in. It is also the world that AI is crashing into.

What That World Does Next

The mandates being drafted right now are diagnostic. It’s as true of screenwriters, as it will be of CPAs or HR generalists. Any job that requires legislation or a professional body to keep AI out is a job that has finished its run as value creation and started its transition into rent extraction.

That is not a durable position.

Even if US models are prevented from writing legal briefs, or providing medical advice, or doing the thousands of other jobs that credentialed knowledge workers do, it won’t matter. People will just use Chinese models, or Indian ones, or Zimbabwean ones, or whichever jurisdiction is not willing to defend the American rent-seeking class. The work that survives will be the work that didn’t need the mandate in the first place—the kind customers voluntarily pay for, the way they have voluntarily paid for skilled work for as long as customers and skills have coexisted.

For the young adult mapping a course in life, this changes the advice. The strategy that worked for the mid-twentieth century—get the degree, get the license, take your seat of professional privilege—is now the worst possible bet you can make. Credentialed gate-keeping is not going to outperform AI, with the wind at its back and trillions of dollars of capital behind it. But the path of building something a customer voluntarily pays for—a product, a service, a piece of software, a company—has never been cheaper or more accessible. The tooling that used to require a Series A, a team of ten, and a network of credentialed intermediaries can now be assembled by one motivated person with a credit card and the same chatbot the rent collectors will try to legislate against. The next decade is going to be the worst time in modern history to be a rent collector, and the best time in modern history to be a builder.

Once again, there will be winners and losers. The good news is we’re still picking teams.

The Filter

The next time you read a piece in a major newspaper about hallucinations in AI legal briefs, or an essay in a magazine warning that AI threatens journalistic integrity, or an open letter from academic researchers calling for a pause on AI development, there’ll be some truth in it. There always is.

But they’ll never say the obvious part: this time, they’re the ones getting displaced.

The journalist warning about misinformation has a job in journalism. The professor warning about cheating has a job in academia. The lawyer warning about hallucinations in legal briefs has a job in law. The policy researcher warning about alignment risk has a job at a policy think tank. None of them is wrong, exactly. All of them are arguing for restrictions on AI in their own labor market, framed as protection of the public, in the same way the carriage driver argued that horseless carriages were too dangerous for the streets and the travel agent argued that online booking was a haven for scammers.

So when they run the headline about AI hallucinating a bad legal brief and Congress is debating mandatory human review of all AI legal work, you’ll be reading the 2030 version of horseless carriages should be limited to walking speed. The only real question is whether or not you get cars, or that old horse shit.

But here’s the part worth noticing: the algorithm hadn’t changed. What changed was that people knew about it. Customers who discovered dynamic pricing started timing their orders differently, comparing prices across platforms, switching to competitors. The behavior the algorithm was trained on—years of purchasing patterns from people who didn’t know the price was adjustable—was no longer the behavior it was facing. The model was optimized for a world that stopped existing the moment the model became visible.

Subscribe now

This isn’t a bug in Instacart’s pricing algorithm. It’s a fundamental property of any system that tries to set prices for participants who can react to them. The moment the outputs become known, the inputs shift.

Zillow learned this expensively in 2021, when it tried to buy homes with an algorithm. The Zestimate generated an offer; the homeowner decided whether to accept. But the homeowner knew things the algorithm didn’t—the outdated kitchen, the noisy neighbors, the pending assessment. Owners whose homes were worth more than the offer sold on the open market. Owners whose homes were worth less sold to Zillow. The algorithm’s errors were random, but the market’s choices weren’t. They were systematically harvested by sellers who had their own opinion of what the home was worth, regardless of what their “Zestimate” said. Zillow wrote down $528 million in a single quarter and shut the program down.

Why It Feels Like It Should Work

Netflix knows what you want to watch before you do. Spotify builds playlists that feel like they were curated by a friend who’s known you for years. Amazon has your package pre-staged at a local warehouse before you’ve clicked “buy.” The experience of living inside these systems is the experience of being known—predicted, anticipated, served. Every year the predictions get better. Every year the algorithms learn more. The intuition writes itself: given enough data and enough compute, you could do this for everything.

And for most of what these systems do, the intuition is correct. Recommending a song costs nothing if it’s wrong, you just hit “skip.” Suggesting a new route to a driver costs one notification. Preloading a warehouse costs shelf space for inventory that needs to be stored somewhere anyway. The stakes are low, the feedback is fast, and the person on the other end mostly doesn’t know or care how the prediction was generated. Spotify doesn’t need to understand why you wanted that song at 2 AM. It just needs to know that people who listened to the last four songs you played tend to listen to this one next. Pattern matching. It’s what neural networks were born to do.

But an economy isn’t a playlist.

The Oldest New Idea

In 1920, Ludwig von Mises published “Economic Calculation in the Socialist Commonwealth.” The standard reading—then and now—is that central planning fails because the planner can’t gather enough information. Millions of producers, consumers, preferences, constraints—no bureaucracy can collect and process it all. The market distributes the computation across every participant, with prices as the output. It’s an information-processing problem, and the market is the best available computer.

This reading is tidy, intuitive, and wrong.

The Soviets built Gosplan—a planning agency that set prices and production targets for everything the economy produced, from steel and tractors to toothbrushes and children’s shoes. Five-year plans. How many tons of pig iron, how many combines, how many meters of cotton fabric. And they could get the data. They had the census reports, the factory output logs, the agricultural surveys. Analysts in Moscow could tell you what came out of Magnitogorsk last quarter down to the metric ton. The data was real. The problem was processing it—coordinating millions of moving parts with pencils, telegrams, and a bureaucratic apparatus that moved at the speed of government. Presumably, given better technology, the problem shrinks. Given AI, maybe it disappears altogether.

Most of Mises’ readers—even sympathetic ones—took him literally: the market is a better computer than a bureaucracy. Hayek, a generation later, took him seriously. Hayek noticed that the information the planner needs isn’t the information Gosplan was collecting. A market doesn’t just need to know how much wheat there is. It needs to know how much wheat is worth—to a specific person, at a specific moment, who could also buy rice, or corn, or skip the purchase entirely because she heard flour might be cheaper next month. That’s not a fact waiting to be collected. It’s a fact that doesn’t exist until after someone decides it for herself.

What the Algorithm Would Need

Gosplan could count the wheat. Any sufficiently powerful computer can count the wheat. Wheat supply is a fact about the world—it sits in silos and on loading docks and can be measured. But price isn’t supply. Price is supply and demand. And demand doesn’t sit anywhere.

Walk onto a car lot. You came to look at the truck. The color’s not great in person. You reconsider. The sedan catches your eye—the trunk is bigger than you thought. Actually, maybe the truck’s not so bad after all. Your preferences aren’t a fixed quantity waiting to be measured—they’re generated in the interaction, shaped by the prices, the alternatives, the salesperson, your morning, your mood, and a hundred variables that didn’t stabilize into a decision until the moment you either signed or walked away.

The market didn’t read that preference. It produced it. The act of deciding generated information that didn’t exist until the decision was made. The transaction is the information—and not just for cars. How much demand is there for an invention that won’t be invented until next year? The question doesn’t have an answer. Not because the answer is hard to find, but because it doesn’t exist yet.

This is why nobody bets on gravity, or at least, nobody bets against it. It’s too predictable—there’s nothing to wager on. But people bet on sports all the time, billions of dollars a year, because human performance is irreducibly unpredictable until it happens. The models help at the margins. They can calculate the odds. But nobody knows what a specific person does in a specific moment until they’re in it and they do it. The odds are probabilities. The outcome is binary. Either they scored, or they didn’t.

An economy is eight billion people in that moment simultaneously, each decision shifting the conditions for every subsequent one. The algorithm needs next week’s answers to generate this week’s prices. The market generates this week’s prices and lets next week take care of itself.

Why Compute Doesn’t Help

A climate model can be wrong about ocean temperatures in 2040 and it doesn’t matter to the ocean. The ocean doesn’t read the forecast.

An economic model that predicts bread will cost $4 next month changes the behavior of every baker and buyer the moment the prediction is published. Buyers stock up at $3.50, creating a demand spike that pushes the price to $4.25. Bakers overproduce in anticipation of a windfall. The price crashes to $2.75. The prediction was wrong because it was right. The model’s output became an input to the system it was modeling.

This is how grocery shelves go bare the night before a snowstorm. Not because there isn’t enough bread—because the expectation that there won’t be enough bread produces exactly the shortage it predicted. The algorithm doesn’t sit outside the economy and observe it. It lands inside and changes everything it touches.

Google Maps runs into this at 5 PM every weekday. Route everyone around the traffic jam on the freeway and the side streets become the jam. Google’s solution isn’t a better prediction—it’s a deliberately worse route. The algorithm serves individual drivers suboptimal paths so that the system clears. It solved the problem by reinventing the mechanism a market already uses: give everyone slightly different information, so they each make slightly different choices.

And even that only holds while drivers trust the algorithm more than their own eyes. The moment you suspect Google is giving you a worse route, you stop following it. You take the turn that looks faster. The car behind you does the same. And now the algorithm is optimizing for a road full of people who all think they know a shortcut.

The Strongest Test

AI is the strongest possible version of the argument Mises already defeated. Not a bureaucrat with a clipboard—a near-omniscient intelligence with access to more data than any human institution has ever processed. If managing global production were really about processing speed or information access, this would be the answer.

It isn’t. The observer—however powerful, however fast—is still an observer. It can read the market’s outputs but not generate them, because the information only exists when real people make real decisions at real prices. An AI that enters the market doesn’t replace it—it joins it. Amazon’s pre-positioning, Uber’s surge pricing, Spotify’s recommendations—these are the best-equipped market participants in the history of commerce. But they’re only participants. A faster car doesn’t replace traffic.1

Every generation produces a new version of this confidence. The Soviets thought central planning would outperform markets because they had the right theory of what people needed. The techno-optimists think AI will outperform markets because they have better data on what people do. Same error, better hardware.

Mises argued against central planners with pencils. He was right—and not because pencils are slow. The strongest technology ever built hits the same wall: you cannot allocate resources better than a market, because the market is generating the information you’d need to beat it. An economy isn’t a system to be solved. It’s a process that generates solutions—continuously, locally, and destructively, overwriting the last answer with the next transaction. The most powerful computer ever built can read the flea market’s prices. It cannot generate them. That takes a Saturday morning and two people willing to haggle over a lamp.

Lobbyists, pork barrel spending, back-channel campaign contributions, a no-show job for your sister-in-law’s second cousin, good old-fashioned insider trading. We complain about all of it, then nothing changes. Asking politicians to vote themselves fewer opportunities for skimming off the top—and the bottom—is a losing battle. They’re not going to collectively act against their own interests. Especially when their interests are special.

Subscribe now

But maybe we’re fighting the wrong war. Everyone wants to get money out of politics. The easier answer is to put more in.

The Pelosi Problem

Nancy Pelosi has served in Congress since 1987. Her salary, like every rank-and-file member’s, is $174,000 a year. It’s been frozen at that number since 2009. Over nearly four decades of service, her total congressional earnings come to roughly $6 million.

Her net worth is reportedly north of $270 million.

There are exactly two ways to read that. Either she’s one of the sharpest investors in American history—a savant who should be running the world’s most successful hedge fund and making billions, but who selflessly chose public service because she just loves America that much. Or she’s a cheat.

Pick whichever story you prefer. But notice that both readings lead to the same conclusion: something about this system is deeply broken. Either we’re dramatically underpaying someone with world-class financial talent, or we’ve built a system where the only way to get rich in politics is to be dishonest about it. Either way, we shouldn’t be surprised by what we get.

Pelosi isn’t even unusual. She’s just the most visible example of a pattern so common it barely registers as scandal anymore. Enough members of Congress beat the market, often enough, that it strains credulity—the kind of track record that would trigger an SEC investigation if they worked on Wall Street. We’ve built a system where the official compensation is a joke and the real compensation is whatever you can get away with. And then we act shocked with what they get away with.

The Talent Problem

Here’s a question nobody asks: why would anyone competent want the job?

Stephen A. Smith spent months teasing a 2028 presidential run, then killed it on Hannity’s podcast: “I got to give up my money. Yeah, I ain’t giving up my money, Sean.” Smith makes roughly $33 million a year between ESPN and SiriusXM. The President of the United States makes $400,000. To a guy like Smith, that’s not a pay cut. It’s practically nothing.

The point isn’t whether you’d vote for Stephen A. Smith. The point is that the people at the top of their fields—the ones who’ve proven they can build things, manage complexity, command audiences, negotiate under pressure—look at the most powerful job in the country and see an expensive step down. Not a promotion. Not a calling. A financial catastrophe.

We pay the person responsible for a $7 trillion budget, 330 million lives, and the most powerful military in the world less than a mid-tier ESPN personality.1 It’s less than most tech executives. Congress makes less than the guy who manages your regional Costco distribution center, probably. And then we wonder why the talent pool looks the way it does.

The private sector figured this out a long time ago. You get what you pay for. Offer below-market compensation and you get two kinds of candidates: zealots who flame out, and grifters who plan to make up the difference on the back end. And since the zealots flame out, we mostly get the second kind.

Cheap at the Price

It’s cheaper to just pay them. Pay every member of Congress and every cabinet member $50 million a year.

There are 535 voting members of Congress, a president, a vice president, and roughly 25 cabinet-level officials. Call it 562 people. At $50 million each, that’s $28 billion a year.

The federal government spent $7.1 trillion last year. Twenty-eight billion is 0.4% of that. Less than half a penny on the dollar. It is, in the most literal sense, a blip on the budget—the kind of line item that would get folded into “other” in a detailed pie chart.

For that price, you get three things.

First, you get talent. Real talent. The kind of people who currently look at a congressional salary and laugh. When the job pays $50 million, you’re not just attracting public servants—you’re attracting people who’ve built companies, managed billion-dollar portfolios, run hospitals and universities, led engineering orgs. People who are currently solving hard problems in the private sector because that’s where they pay talented people proportionately to their competence. Make governing competitive with Goldman Sachs and you’ll get Goldman Sachs-caliber people competing for the job. Some of them might even be honest.

Second, even if they’re not, you get independence. When a member of Congress makes $50 million a year, a lobbyist’s million-dollar campaign donation just doesn’t curry much favor. You can’t buy a nine figure politician for seven figures. The entire economics of influence shift when the people you’re trying to buy are already rich. A congressman making $174,000—frozen, by the way, since the iPhone 3G—is one bad election away from being financially vulnerable. A congressman making $50 million has “fuck you” money. Which is precisely what we want them saying to lobbyists.

Third, and most importantly: you get integrity. Right now, we ask public servants to take a vow of poverty and then act surprised when it turns out they didn’t mean it. The honest people see it as a step down, so the dishonest step up. Moralizing about corruption doesn’t dissuade immoral people. Pay them enough that the legitimate compensation dwarfs whatever they could skim, and the graft doesn’t go away—but it stops being the primary business model.

$50 million attracts talented people, but it doesn’t necessarily mean they’ll care if the budget balances.

Better Incentives

So give them a reason to care: pay them a percentage of whatever surplus they generate.

The federal deficit last year was $1.8 trillion. Offer every member of Congress and the cabinet 10% of any surplus they produce. The other 90% goes to paying down the debt, and once the debt’s retired, into a sovereign wealth fund.

Under this system, a balanced budget pays them nothing extra. But turn that $1.8 trillion deficit into a $500 billion surplus, and suddenly each of those 562 officials takes home an $89 million bonus. Generate a trillion-dollar surplus and they’re each making an extra $178 million a year—hedge fund money, without having to hedge. And the rest of us get to hand a debt-free America to our children.

Would they game it? Of course they would. They’d game it furiously. The fear is that they’d gut services and put people on the street to juice the number. But they still have to get elected. A congressman who shutters the local VA hospital to pad his bonus check doesn’t have a long career. The games they’d actually play would be the ones that generate surpluses without losing votes—cutting waste, growing the economy, collecting revenue efficiently. Right now they game the system to funnel money to donors and secure “consulting” gigs. They’re going to play games. The question is which set of games do we want?

The Wealth Fund

Today the federal government manages $39 trillion in debt. There is nothing to invest, nothing to steer, nothing to grow. The entire job of fiscal management is figuring out which special interests to keep paying when there’s no money left, and how much more to borrow to do it. It’s a financial triage operation, minus the financial and triage skills.

Norway—a country of 5.5 million people—runs a $2.2 trillion sovereign wealth fund. It owns roughly 1.5% of every publicly listed company on Earth. The fund returned $247 billion in profit in 2025 alone. That’s not tax revenue. That’s investment income, generated by a pool of national wealth that compounds year after year.

Imagine the United States built something similar. Not overnight—it took Norway 35 years to build up that balance. But here’s the surprising part: Norway never paid off their debt first. Norway started the fund in 1990 carrying debt, and they’re carrying debt today. The fund didn’t replace the debt. It ran alongside it. Surpluses went into the fund. Debt got serviced separately. The two were never treated as a sequence.

At American scale, a sovereign wealth fund with an equal value per person would be worth roughly $140 trillion. At the fund’s historical average return of 6.6%, that’s $9.3 trillion a year in investment income—nearly double what the federal government currently collects in taxes. No taxes required. Borrow a trillion at 4%, deploy it at 6.6%, and the spread is the return. That’s called leverage.

And yes, there would be graft. Politicians would steer investments toward favored industries, connected firms, allied nations. There would be backroom deals and influence peddling and all the same corruption we have now.

But here’s the difference: right now, the graft is parasitic. Politicians are skimming from the bottom of a debt hole—and they keep digging. Steering trillion-dollar investments is a lot more lucrative than spending money you don’t have. With a wealth fund, the backroom deals would be with people deploying real capital into real assets. The corruption would be a tax on growth, not a tax on you.

It’s still graft. But it’s a better kind.

The Oldest Profession

The instinct to purify politics—to somehow extract the self-interest from a system built entirely on self-interest—has failed for as long as anyone’s been trying. Every ethics reform, every disclosure requirement, every campaign finance law has been met with creative workarounds by the very people the laws were supposed to constrain. The fox keeps redesigning the henhouse, and we keep acting surprised when chickens go missing.

Maybe the answer isn’t a more virtuous fox. Maybe it’s a henhouse where the fox is already so well-fed that the chickens aren’t worth the effort.

You don’t fix corruption by eliminating incentives. You fix it by making the legitimate incentives so large that the corrupt ones stop mattering. Pay them enough that the job attracts talent instead of grifters. Structure the compensation so their financial interests align with ours. Give them something worth managing instead of a pile of IOUs.

O’Rourke was right. Every government is a parliament of whores. And who wants the cheap kind?

Professional power completed the pattern in the late nineteenth and early twentieth centuries, as standardized examinations and licensing replaced guild inheritance and family apprenticeship. A surgeon’s daughter doesn’t inherit the scalpel. She takes the MCAT, survives medical school, matches into a residency, and passes board certification. The credential expires with the person who earned it.

Subscribe now

The principle was the same each time: power must be earned, not bequeathed. A title held by one generation conveys nothing about the competence of the next.

Of course, dynastic advantage persists in all three domains, and everyone knows it. A senator’s kid has name recognition, donor networks, and twenty years of dinner-table education in how power works. A general’s kid grows up on bases, absorbs military culture, and understands the institution before his first commission. A surgeon’s kid benefits from family wealth that can fund a decade of training, legacy preferences, and parents who know exactly which residency programs to target.

These advantages are real and significant. But the formal entitlement is gone. The senator’s kid still has to win the election. The general’s kid still has to pass the selection board. The surgeon’s kid still has to get licensed. Between the advantage and the power, there is a proving ground—a gate that won’t open on pedigree alone. The dynasty can tilt the odds. It cannot skip the test.

Economic Empires

Economic power has no such test.

The Walton heirs collectively hold roughly $250 billion. Sam Walton built Walmart. His children and grandchildren didn’t. They hold more wealth than the bottom 40% of American households combined, and the mechanism that transferred this power to them involved no election, no exam, no selection board, no credential check, no demonstrated competence of any kind. A will executed. A fortune moved.

In every other domain, inheritance is an unlawful advantage. In economics, it is the norm. A senator’s kid with every dynastic edge still has to win an election. An heir just has to have a pulse.

Biology and Society

Of course, parents want to leave their children what they built. That instinct is as natural as they come. But it was equally natural in every other domain. Kings wanted their sons to rule. Generals wanted their sons to command. Doctors wanted their sons to practice. The desire to pass power to your children is one of the oldest forces in human life. It is also one of the most consistently overridden by law, because the damage it does to society outweighs the comfort it provides to our biological urges.

Natural impulses get legally overridden all the time. If your kid commits tax fraud, you may want to help cover it up. The law says you can’t. If your kid fails the bar, you don’t get to practice law on their behalf, either. The instinct to protect and provide for your children is deep, real, and irrelevant. All of our laws are designed to override our base nature. That’s what a civilization is. In politics, the military, and the professions, the answer has always been to overrule this impulse—and we’re better off for it.

But the desire to provide for your children and the drive to build a billion-dollar business are not the same instinct. The people who create enormous fortunes aren’t doing it for their grandchildren. They’re doing it because they’re obsessed. Bezos didn’t grind through early Amazon dreaming of a trust fund. Musk isn’t sleeping at the factory for estate planning. The drive to build is intrinsic—that’s the entire capitalist argument for tolerating unequal rewards. The inheritance isn’t the motivator.

The real reason the exception persists is power, not instinct. The people who benefit from inherited wealth have the resources to keep the system in place. The Lords didn’t vote to abolish the Lords. The purchasers of commissions didn’t volunteer to end the purchase system. Every time, the change came not because the beneficiaries agreed but because something broke hard enough to override them. In 1990 the top 0.1% held 8.5% of the wealth. By the end of 2024, it had grown to 13.8%. The 19 richest U.S. households added about $1 trillion in wealth in 2024 alone.

Something is broken.

Estate Taxation

Here’s how the estate tax actually works, since most Americans have never had reason to learn. When someone dies, their estate is valued—everything they owned. The federal government exempts the first $15 million per individual from taxation.1 For a married couple, that’s $30 million. Everything above the exemption is taxed at 40%. Everything below it is untouched.

Most people can intuit the obvious loophole with this—just give it away on your deathbed. So our laws handle that too. Anyone can give $19,000 per year to any recipient, tax-free. But larger gifts count against the total estate exemption—give your daughter $1 million this year, and your estate exemption at death drops by $1 million.2 The system treats large gifts and inheritance as two draws from the same pool.

The median American estate is roughly $300,000. The current exemption is $15 million. Only about 0.1% of estates currently owe anything at all. And even those estates, after trusts and avoidance strategies, pay an effective rate of about 3.5%. The estate tax collects roughly $35 billion a year on over $1 trillion in transfers. Functionally, it’s more of a suggestion than a rule.

Every argument against raising taxes is an argument about incentives. Income taxes discourage work. Capital gains taxes discourage investment. Sales taxes discourage consumption. These objections have force because they describe real behavioral distortions imposed on living people making real decisions.

Dead people don’t make decisions. A dead person has no labor supply to reduce, no investment to defer, no business to relocate, no income to shelter. The entire economic case against taxation is a case about distorting behavior, and a dead person’s behavior cannot be distorted. The deadweight loss of an estate tax is zero, because the taxpayer is dead weight.

This makes the estate tax unique in the tax code—the only tax where the standard economic objections don’t apply, because the entity being taxed no longer exists. The only tax with no economic cost is the one we barely collect.

Democratic Principles

The correct rate is 100%, because that’s the principle we already agree with. Nobody inherits 60% of the king’s title. Nobody inherits half a general’s stars. The credential, the rank, the office—they either transfer or they don’t. In politics, the military, and the professions, they don’t. A small exemption—$5 million per person, $10 million per couple—handles the obvious: the family home, the Mustang dad lovingly restored, the good china, the heirlooms that represent a family, not a dynasty. For the vast majority of families, nothing changes.

For wealthy families, most of the real advantage continues uninterrupted, too. Children still inherit relationships, access, family name, reputation—the same dynastic advantages that persist in every domain where formal inheritance has been abolished. The senator’s kid still has the donor network. The general’s kid still has the institutional knowledge. The same advantages apply here.

The financial tools for taking care of your family are already in the tax code. Parents can give $19,000 a year to each child, each grandchild, each in-law—anyone they want, as many recipients as they want, every year they’re alive. A couple with three kids and six grandchildren can transfer over $170,000 a year without touching the exemption. Over twenty-five years, that’s $4.25 million on top of the $10 million that passes at death. Tuition paid directly to a university, medical bills paid directly to a provider—none of that ever enters the tax system. Families can, and should, continue to help each other.

What the estate tax should eliminate is the lump-sum transfer of a dynasty—the fortune so large it replicates itself across generations regardless of what anyone does with it. That’s not a tax, it’s a principle. It’s how every other form of power in this country already works.

The revenue goes where it should have been going all along—toward the budget, toward infrastructure, toward programs like universal opportunity accounts that give every American a real starting position instead of a lottery based on who their parents were. The point isn’t to destroy wealth. It’s to stop pretending that a system which transfers dynastic fortunes is consistent with the principles of democratic and meritocratic power.

The Last Holdout

The hereditary principle has been on a 250-year losing streak. Each time it was abolished, the beneficiaries of the old system predicted catastrophe. The aristocrats said democracy would produce mob rule. The officer class said meritocratic promotion would destroy military cohesion. The guild families said open licensing would flood the professions with incompetents. Each time, they were wrong. Each time, the system that replaced inherited power with earned power outperformed the one it replaced.

Economic power is the last holdout, and the biggest. A modern dynastic fortune concentrates more power than a medieval lordship, a purchased commission, and an inherited credential combined. The principle that applies to it is not new, not radical, and not untried. It has been tested at least three times, in three domains, across three centuries.

You get a lifetime to earn wealth, spend it, give it away, and use it however you see fit. That’s the deal. That’s what a free society offers. But the deal is with the living. When it’s over, it’s over. We abolished the inherited title. We abolished the purchased commission. We abolished the hereditary credential. There’s one last holdout. Let’s abolish it.

Both emerged from extraordinary science and extraordinary scientists. Both were classified at the highest levels. What happened next split the trajectory of the global economy in two.

Subscribe now

The codebreaker machine was classified Ultra Secret—a designation above Britain’s highest standard classification. The machine itself, its specific architecture and implementation, remained classified and closely guarded for decades after the war. But Alan Turing’s foundational work—his 1936 paper On Computable Numbers, the theoretical bedrock of every computer ever built—and his related research, remained in the open literature. You can read it today, just as you could have during the war.

The bomb got classified too. But unlike the computer science, the Atomic Energy Act of 1946 classified not only the weapon—it classified the entire scientific foundation underneath it. The Act created a legal doctrine unprecedented in the history of knowledge: “born classified.” Certain categories of nuclear physics were declared state secrets from the moment of discovery—by anyone. A university researcher with no government funding and no security clearance who independently derived restricted results had, by law, produced a state secret before anyone in government had seen it. The knowledge was illegal before the ink dried.

This wasn’t intentional, or at least, it wasn’t strategic. Governments simply understood the power of bombs better than the power of bits.

A bomb is visceral—the mushroom cloud over Hiroshima was photographic proof that the physics underneath it was dangerous. Generals and senators didn’t need to understand neutron chain reactions to understand that the knowledge enabling them should never reach Moscow. Turing’s work was abstract. Math and logic. Theory based on a hypothetical thought experiment in which a machine reads an infinitely long strip of tape with 1s and 0s on it. So they classified the codebreaking equipment—that part they understood the importance of—but the idea itself was allowed to flourish in the public domain.

The Divergence

Modern computing is the result of eighty years of compounding breakthroughs on Turing’s open foundations. Transistors, integrated circuits, personal computers, the internet, artificial intelligence—each generation building on the last, each published, peer-reviewed, and available to every researcher and inventor on the planet. And available to you. The iteration loop that drives scientific progress—publish, challenge, replicate, extend—runs uninterrupted.

Physics, meanwhile, flatlined.

The most celebrated physics breakthrough of the last twenty years was the 2012 detection of the Higgs boson—a particle first theorized in 1964. Gravitational waves, detected in 2015, confirmed a prediction Einstein published in 1916. Significant accomplishments. But also confirmations of very old ideas, rather than the discovery of new ones.

The pattern holds across applied physics, too. The Boeing 707 entered commercial service in 1958, cruising at Mach 0.85. The Boeing 787 entered service in 2011, cruising at... Mach 0.85. More fuel-efficient, but the same speed, the same basic airframe, the same fundamental technology—fifty-three years apart. The biggest upgrade to the passenger experience was switching from CRT TVs to seat-back screens. And now airlines are removing those too, because your phone does a better job, thanks to computer science. Fifty years of Boeing, a prime defense contractor, shipping essentially the same commercial products despite decades of military advancements.

Materials science. Energy storage. Propulsion. Electromagnetics. Incremental refinement in secret, decade after decade, while the domains whose science stayed open compounded exponentially.

The Easy Answer

There is a serious, well-researched argument that this scientific stagnation is natural.

Physics had a golden age—roughly 1900 to 1960—that produced relativity, quantum mechanics, nuclear physics, and the solid-state physics behind the transistor. These were foundational breakthroughs: deep, accessible, and world-changing. The argument is that what followed was inevitably harder. The low-hanging fruit had been picked. Newton needed an apple. Now we need supercolliders. Tyler Cowen called the broader phenomenon the Great Stagnation. Bloom, Jones, Van Reenen, and Webb published “Are Ideas Getting Harder to Find?” and concluded that, yes, they are—across the board, research productivity is declining as each new discovery requires more people, more expertise, and more capital.

These are serious economists looking at real data. And the story is internally consistent: physics matured, the easy breakthroughs were made, and what remains is incrementally harder. The plateau isn’t a mystery. It’s a natural life cycle.

The trouble is what the explanation has to ignore to stay comfortable, not least of which is describing what Einstein produced as “low-hanging fruit.”

What the S-Curve Can’t Explain

If diminishing returns are natural—a feature of the science itself—they should hit every scientific field as the easy discoveries are exhausted. But they don’t. They hit physics, materials science, propulsion, and energy. They skipped biology, computer science, and every other domain whose foundational knowledge stayed in the open literature.

The National Institutes of Health spend roughly $47 billion a year on publicly funded biomedical research—published, peer-reviewed, available to researchers worldwide. Over the same fifty years that physics stagnated, biology produced genomic sequencing, CRISPR, immunotherapy, mRNA vaccines, and a revolution in molecular biology. No plateau. No diminishing returns. The pace of discovery accelerated. And it didn’t accelerate in isolation. Biology drew on breakthroughs in computing, chemical engineering, optics, statistics—every open field feeding into every other open field. Open science doesn’t just compound within a discipline. It compounds across the entire network, because a breakthrough published anywhere is available to researchers everywhere. Classification doesn’t just slow one field. It severs nodes from the network.

The “experiments got expensive” argument explains particle physics—you do need a multi-billion-dollar collider to detect the Higgs boson. But it doesn’t explain why materials science stagnated. Or energy storage. Or propulsion. Or applied electromagnetics. These are fields where the experiments aren’t prohibitively expensive and the commercial demand for breakthroughs is enormous. The S-curve landed selectively on the domains whose underlying science overlaps with defense classification, and left the open domains untouched.

Then there’s the timing. The physicists who built the bomb were trained in the pre-classification world—open publication, international collaboration, rapid iteration. The next generation, mentored by Oppenheimer’s contemporaries, also came up in that tradition. Both cohorts carried the knowledge base of open science in their heads. Then they retired in the 1970s. After that, every working physicist had been trained entirely within the classification regime. The open-science culture that produced the golden age wasn’t just suppressed by law. It retired out of the workforce one career at a time.

The S-curve theorists date the plateau to the same decade and call it nature. But the plateau doesn’t map to the exhaustion of discoverable physics. It maps to the disappearance of the last generation that practiced physics in the open.

What Leaks Through

Occasionally, classified technology escapes into the civilian world.

GPS was a military positioning system beginning in 1973. Civilian access was granted a decade later, with intentional accuracy degradation. Full precision wasn’t available until 2000. Twenty-seven years from invention to unrestricted civilian use. It has since generated an estimated $1.4 trillion in economic value in the United States alone. Fiber optics, developed for classified military communications, became the backbone of the modern internet. Radar, declassified after the war, produced microwave ovens, air traffic control, weather forecasting, and the foundation of modern aviation safety.

These aren’t anomalies. They’re samples. When declassified technology reaches the civilian economy, it produces outsized returns—because it enters a market that has been systematically starved of the underlying knowledge. GPS didn’t generate a trillion dollars in value because it was uniquely brilliant. It generated that value because it was one of the few things that made it into the economy. But the economic impact is only part of the story. Even once it was commercially available, it lost twenty-seven years of subsequent technological and economic compounding.

The talent tells the same story from the labor-market side. A gifted physicist motivated by hard problems might accept a lower salary—plenty of brilliant people choose academia over finance. But classification doesn’t just pay less. It strips away every non-monetary incentive that draws the intellectually driven. No publication. No peer recognition. No watching your work enter the world and reshape it. The choice isn’t between money and recognition. It’s between money and recognition, or neither. The disciplines where you can publish, build on others’ work, and see your ideas tested in the open attract the best minds. The disciplines where the best work disappears into a classified archive lose more of them every decade.

The Externality Nobody Measures

National defense is the textbook case for collective funding. No individual can opt out of the nuclear umbrella, so the cost is shared through taxation. The benefit—security, sovereignty, the freedom to build an economy without fear of invasion—is shared too. The principle is sound.

But the classification program has a non-tax cost that never makes it onto the ledger or into the conversation. The way defense R&D is implemented—classifying the science, not just the weapons—creates a compounding negative externality running alongside the positive one.

The physics presumably still advances inside the classified world, albeit more slowly. But the knowledge never reaches the civilian economy—never gets published, challenged, or extended by the engineers and entrepreneurs who would build products, launch industries, and drive the productivity gains that show up as increased wages, improving standards of living, and a broader tax base to support more R&D. This compounds the way interest compounds. Every suppressed breakthrough is a missing foundation for the next generation of breakthroughs. The drag isn’t the classified research itself. It’s the entire tree of civilian applications, products, and industries that never grew from seeds locked in a vault.

The defense budget is a trillion dollars a year. That’s the number we debate. The compounding productivity loss—eighty years of missing civilian applications building on missing civilian breakthroughs—dwarfs that number the way compound interest dwarfs principal. But nobody can put a figure on it, because we can’t measure the value of discoveries that were never made and industries that were never built. It will never appear on an appropriations request, because the cost doesn’t look like a cost. It looks like stagnation. It looks like S-curves. It looks like physics just getting hard.

The False Choice

The standard defense of born-classified is that it worked. Nine countries have nuclear weapons. Dozens more could have pursued them. The classification regime—alongside treaties, diplomacy, and deterrence—helped hold that number down.

But “worked” overstates the case. Pakistan acquired nuclear capability through A.Q. Khan’s espionage and smuggling network. North Korea built a weapon under the heaviest sanctions regime on earth. The classification of nuclear physics did not stop either program. We’re in a hot war with Iran because knowledge suppression doesn’t stop nuclear development. Any sufficiently motivated country can figure it out—the knowledge leaks, gets independently derived, or gets stolen.

What the classification regime actually achieved was delay. And delay is genuinely valuable when the subject is civilization-ending weapons. But delay is a different proposition than prevention, and it demands a different cost-benefit analysis. If born-classified prevented proliferation permanently, the sacrifice of eighty years of civilian physics might be a price worth paying. If it delayed proliferation—bought decades of time while the knowledge inevitably spread—then the question is whether eighty years of compounding civilian stagnation was a reasonable price for a slightly less fast nuclear proliferation. It might be, if we used the time to build a durable solution, but that is a much harder case to make.

The slowdown didn’t require classifying the science, anyway. We know this because we watched the other technology take a different path.

The United States developed cryptography, signals intelligence, and cyber weapons — all classified applications built on Turing’s open foundations. The NSA runs some of the most sensitive programs in the intelligence community, all of them downstream of computer science. Nobody classified computer science. The applications were classified. The weapons were classified. The underlying science stayed in the open literature, compounding for eighty years, producing a civilian economy worth trillions—while simultaneously producing the most sophisticated classified capabilities on earth.

Defense research could be handled similarly. Classify the weapon designs. Classify the engineering. Classify the enrichment techniques. But leave the physics—the foundational science that also underlies civilian energy, materials, propulsion, and a hundred other productive applications—in the open, where it could compound the way every other open science has. The Atomic Energy Act didn’t draw that line. It classified the knowledge itself. And the fields downstream of that knowledge, and our economy, have been paying the compounding price ever since.

Two technologies won the war. We opened one and got the modern world. We classified the other and got excuses.

Subscribe now

The first kind inherits it. Roughly a quarter of American billionaire wealth is inherited. Dead people don’t own things. Dead people are things. Whatever you believe about the moral right to pass wealth to your children, the economic question is different: did the transfer create anything? Inherited billions are economically inert. They employ wealth managers and tax attorneys. The argument for taxing them more aggressively than earned wealth isn’t about envy—it’s that this money isn’t doing for the economy what money is supposed to do.

The second kind makes it through finance—trading, investing, leveraging assets into more assets. This one is harder to judge. There’s nothing inherently wrong with someone trading their way to a billion dollars. The world’s most sophisticated version of the paperclip game, where over the course of a career you iteratively trade your way up from a silver spoon to a private island. But a lot of what constitutes making money in finance is gating access—charging tolls on bridges that other people should be allowed to cross on their own.

The third kind builds something. They start a company, hire people, make a product, sell it to willing buyers, and do this at such a scale that the byproduct is a billion dollars. This is the kind we need more of. Not because billionaires are virtuous. Because the process of becoming one—through entrepreneurship—is the single most powerful force1 for raising the wages of everyone who isn’t one.

And wages need raising.

The Only Thing That Raises Wages

The Bureau of Labor Statistics tracks real wages for production and nonsupervisory workers—roughly 80% of the American workforce. In 1973, adjusted for inflation, those workers earned roughly what they earn today. Half a century. Functionally zero real growth. Meanwhile, productivity per worker roughly doubled. The economy got twice as efficient at producing output. The people who made it happen saw almost none of it.

From the Industrial Revolution through the early 1970s, wages rose almost every year—not just cost-of-living adjustments or the occasional promotion, but genuine upward price pressure on labor across the board. That era lasted a century. Then it stopped. The plateau has now lasted long enough that essentially no worker in the labor force today has ever experienced a sustained rising-wage economy. Entire careers have started and ended within the flatline. We’re losing sight of what wages are supposed to do as productivity improves.

The standard explanations—globalization, automation, declining unions—all contain some truth. They also all point in the same direction once we stop treating them as buzzwords. Globalization expanded labor supply beyond domestic borders. Automation reduced labor demand by replacing workers with machines. Declining unions weakened labor’s pricing power. That’s three ways of saying the same thing: labor is not a special product. It follows the same laws of supply and demand to set prices as copper, oil, or soybeans. More supply, less demand: price stays flat or drops. More demand, less supply: prices rise. For fifty years, labor supply has expanded while the creation of new enterprises—new sources of labor demand—has not kept pace. The result is printed on every paycheck in the country.

Every entrepreneur who steps out of the job market does two things simultaneously. First, they remove themselves from the supply side. One fewer person competing for a job. Second, if they’re even modestly successful, they start creating demand—hiring people, contracting services, buying supplies. One person shifts from a net seeker of jobs to net creator of them.

Even the entrepreneur who fails does this. Start a company, hire twelve people, run it for eighteen months, and fold. That’s twelve jobs that didn’t exist before—twelve units of labor demand that one person generated by stepping out of the labor supply. And here’s the part that matters: wages are sticky. When’s the last time you took a pay cut? When the market raises wages, those wages tend to stay raised. So every burst of labor demand ratchets wages up, and the ratchet doesn’t fully unwind when the company does.

Now scale it. Imagine 2% of the working population did this—roughly three million people. Not became billionaires—just started businesses. Three million people moving from the supply side of the labor market to the demand side. Tens of millions of new positions being created, even temporarily, putting durable upward pressure on wages for everyone.

The entire negotiation shifts, because employers can no longer rely on a surplus of applicants. Now we start interviewing them.

What a Billion Dollars Actually Takes

So what’s special about the billionaires specifically?

Scale. Getting three million people to start businesses would transform the labor market. It would also require convincing three million people to do one of the hardest jobs in the economy. A single founder building a company at billionaire scale creates the labor demand of thousands of ordinary entrepreneurs combined.

The arithmetic is concrete. To accumulate a billion dollars as a founder, you generally need to build a company worth at least $5 billion—after dilution, taxes, and the long road of funding rounds. To sustain a $5 billion valuation, you’re generating somewhere in the range of $500 million to a billion in annual revenue. And to generate that kind of revenue, you need a lot of people. Revenue per employee has a practical ceiling in most industries—roughly $300,000 to $500,000 at mature companies outside of pure software. A billion in revenue at $400,000 per head is 2,500 workers at minimum and usually far more, because the ratio includes every employee, not just the ones touching revenue.

The actual numbers are bigger. Microsoft employs 220,000 people. Google, 180,000. Oracle, 160,000. Amazon, 1.5 million. Meta runs leaner than most—about 70,000—and even that is the labor demand equivalent of 35,000 entrepreneurs each hiring one person2. At the other end of the spectrum, one Jeff Bezos created more labor demand than the entire small-business sector of most mid-sized American cities.

That’s what a billion-dollar fortune actually represents. Not wealth extracted from the economy. A receipt for the creation of massive labor demand that didn’t exist before one person decided to build something. The billions are a byproduct of all that activity. Not the cause of it—the receipt.

Not Trickle-Down

This doesn’t require any trickle-down economics to work.

Trickle-down says: enrich the people at the top, and the benefits will eventually flow downward. That’s a hope about second-order effects. Entrepreneurship creates first-order labor demand. It doesn’t trickle. It hires.

A company at billionaire scale doesn’t just hire executives. It hires warehouse workers, drivers, software engineers, HR managers, accountants, janitors, graphic designers, and a small army of people whose job titles didn’t exist ten years ago. When a company like that is growing, it’s competing for workers across the entire labor market simultaneously. Every position it fills is one more employer bidding for someone’s time, at every rung of the ladder at once.

And the pressure ripples. A warehouse worker who can get $22 an hour at Amazon won’t accept $16 down the street. The other warehouse raises its wages or loses its workers. A software engineer fielding three offers pushes all three companies to bid higher. When demand rises in one part of the labor market, it raises wages in adjacent parts too, because workers have options and employers have to compete. That kind of broad-spectrum pressure is something no minimum wage law or policy initiative can replicate. Not because policy is bad. Because policy targets one rung at a time, and demand hits all of them at once.

Why We Don’t Have More

If entrepreneurship is this good for the labor market, the obvious question is: why isn’t it happening more?

The U.S. startup rate has been declining for decades. The Census Bureau’s Business Dynamics Statistics show new firms as a share of all firms dropped from roughly 12–13% in the late 1970s to about 8% by the 2010s. Adjusted for population, fewer Americans start businesses today than in the 1980s. Something is actively suppressing the thing that would raise everyone’s wages.

Some barriers are structural. Entrepreneurs don’t qualify for unemployment insurance—leave a job to start a company and if you fail, you get nothing. The safety net offered to every other worker is null and void for entrepreneurs. Young Americans graduate with six figures of student debt and need a reliable paycheck just to make their monthly payments. The very population most likely to start businesses—young, energetic, willing to take risks—is entering the workforce pre-loaded with a financial constraint that pushes them toward the relative safety of the job queue and away from entrepreneurship.

And the dominant funding mechanism makes it worse. Most people imagine that starting a business means “raising money,” but the math of venture capital excludes most of the businesses the economy actually needs. A VC fund raises $100 million from institutional investors. The managers invest a couple million each into ten companies. Eight go to zero. One breaks even. They pour the remaining capital into the last one—which now has to return enough to make the whole fund profitable. The fund needs to return $200 to $250 million in seven years, so the managers can raise their next fund. Run the math backward and you land on a simple filter: unless a company has a plausible path to a billion-dollar valuation, it doesn’t get past the associate screening the pitch deck.

Think about what that excludes. An HVAC company doing $5 million a year and employing 40 people. A regional restaurant chain clearing $2 million in profit and creating 200 jobs. A SaaS company serving a niche market, making its founders wealthy, and employing 500 people. These are exactly the businesses the economy needs—profitable, stable, creating real labor demand across the entire spectrum—and the dominant funding mechanism won’t touch them. Not because VCs are snobs. Because the fund math doesn’t work unless the company can 100x in less than a decade.

But the other barrier isn’t structural. It’s cultural. In Japan, sumo wrestlers are revered. Boys dream of entering the dohyō. Grand champions are household names—treated the way Americans treat LeBron James. And Japan gets extraordinary sumo wrestlers. Societies produce what they celebrate. America doesn’t celebrate the founder who creates 50,000 jobs. It resents “billionaires”—undifferentiated, no modifier. A culture that can’t distinguish between an heir, a trader, and someone who just created tens of thousands of jobs is not going to produce more of the third one.

But we don’t just suppress billionaires, an outcome many may count as a moral victory. We suppress many of the attempted billionaires, too. The men and women who “only” create 100 jobs, instead of 100,000. Which is a shame, because the economy doesn’t just need more unicorns. It needs more horses, too.

Ask Which Kind

The next time someone tells you billionaires are the problem, ask them which kind. The heir who did nothing? Tax the estate. The financier who gated access to capital markets? Worth a conversation. But the founder who built a company, hired tens of thousands of people, and created billions in economic activity through voluntary exchange?

That person didn’t take anything from you. They made the thing that makes your wages go up. And the economy—your paycheck, specifically—needs about a thousand more of them.

That’s what you say.

Here’s what you sign:

Either party may end the marriage at any time, for any reason or no reason, unilaterally. On exit, marital assets are divided approximately equally regardless of who accumulated them or why. Children are often placed primarily with one parent in practice. The other parent may be ordered to make ongoing payments calculated on income, not need, and enforceable by the state. There is no meaningful mechanism for establishing breach. There is little consideration of who initiated the exit or why.

Subscribe now

The vows promise permanence, sacrifice, and mutual obligation. The legal structure guarantees none of it. You stand at the altar reciting terms with no legal force, while the terms with legal force are never read aloud. No other major contract in American life works this way. The ceremony says one thing. The law says another.

How We Got Here

In 1969, Ronald Reagan signed the Family Law Act, making California the first state to offer no-fault divorce. He would later call it one of the biggest mistakes of his governorship. Within fifteen years, every state had followed California’s lead.

The cultural case for no-fault was simple: the old system was cruel. And often it was. Abused spouses, especially women, were asked to prove private harm in systems poorly designed to recognize it. Judges dismissed credible claims. Evidence was hard to obtain from inside a controlled household. People remained trapped because the court demanded proof that was often impossible to produce.

That failure was real. But the reform that followed did more than remove a bad evidentiary burden. It removed accountability from the structure altogether.

The old system had an internal logic. One party—historically, the wife—often accepted career risk by leaving the workforce to raise children and manage the household. The other accepted a long-term obligation to provide financial security. If one party wanted out, breach mattered. Adultery, abandonment, cruelty. The exit terms reflected who had broken the arrangement, and the consequences fell on the breaching party.

You can argue about whether that system was fair. But it was economically coherent. Both parties accepted obligations. Both received consideration. Exit was tied to conduct.

No-fault didn’t refine that structure. It hollowed it out.

The Option You Didn’t Know You Sold

If your business partner offered a term sheet that said, “Either of us can walk away at any time, for any reason, and you may still owe me half plus ongoing payments on any income you make anywhere,” you would leave the room. We do not accept terms like that anywhere else because we understand what they are.

That is not a bilateral contract. It is an option.

And of course people want an option. Marriage involves real risk. For women, that has historically included pregnancy, childbearing, domestic labor, and career interruption. Wanting protection against a bad outcome is rational.

But protection and an option are not the same thing.

Protection says: if the other party fails to uphold the deal, you can exit with compensation for what you sacrificed in reliance on it.

An option says: you can exit for any reason or none, and still collect.

One enforces the agreement. The other makes the agreement optional—but not for both parties equally.

That is the trick at the heart of modern marriage law. We still speak as though marriage is a binding mutual commitment, while structuring it as a relationship one party may terminate without breach and with substantial legal claims intact.

Women initiate approximately 70% of divorces, a figure that has held for decades. In any other market, when one party exercises an exit option at roughly two and a half times the rate of the other, we call that a positive expected value. No one would describe the other side’s position as a rational investment.

The Rationale That Evaporated

The defense of the current system is familiar: alimony and asset division compensate real sacrifice. Fair enough. If one spouse stepped out of the workforce for fifteen years to raise children while the other built income and assets, that sacrifice created a legitimate claim. Compensation there is not just morally justified. It is economically coherent.

The problem is that the compensation regime outlived the conditions that justified it.

Women’s labor force participation surged after 1970. Women now earn the majority of bachelor’s and master’s degrees. In most households, both partners work. In many marriages, the premise that one party sacrificed all career capital for the other is no longer the norm. But the legal defaults still reflect that older world.

Support obligations are still often keyed primarily to the payer’s income rather than the recipient’s realistic earning capacity. Custody norms still carry assumptions inherited from an era when mothers were more often full-time caregivers because they overwhelmingly were.

Even when individual cases may be arbitrated differently, rational people do not evaluate contracts by average outcomes. They evaluate exposure.

If you earn enough, support obligations can last for years, even indefinitely. Your former spouse may share a home, expenses, and a life with a new partner without necessarily changing those obligations, so long as the arrangement avoids the legal trigger that would end them. The incentive is obvious. The system can reward dependency and penalize formal remarriage.

A structure that rewards the lower-earning party for remaining lower-earning is not built to restore independence. It is built to preserve claims.

And the losses it recognizes run in only one register. Economic sacrifice counts. The loss of daily access to your children, the collapse of the household you built, the destruction of the family unit around which you organized your life—those may be devastating, but they are not compensable. The law sees money. It is far less able, or willing, to see anything else.

The Retroactive Rewrite

One of the least discussed features of the no-fault transition is that it was applied to marriages already in existence.

People who married under one legal framework did not simply receive a new procedural rule. They saw the practical terms of the institution change while they were already bound by it. The difficulty of obtaining a unilateral divorce had been part of the bargain as they understood it. Then the state rewrote the exit terms without asking either party.

If your lender rewrote your mortgage fifteen years in, you would call it a material change. When the state rewrites the practical terms of marriage after the fact, we call it progress and move on.

Whatever else no-fault was, it was not merely prospective reform. It was retroactive restructuring.

The Downstream

The U.S. marriage rate has fallen more than 60% since 1970. The median age at first marriage for men has risen from 23 to 30. The mainstream explanation is cultural—masculinity in crisis, fragile egos, boys who never grew up.

But there is a simpler explanation: when the terms of an agreement deteriorate, fewer people sign it.

Men do not need to read family law treatises to understand the risk. They see divorced fathers, brothers, uncles, and friends. They infer the terms from observed outcomes. They do what rational actors do when exposure is large and downside is asymmetric: they become more selective, delay commitment, or avoid the contract entirely.

And this is treated as a moral failure in men even though similar selectivity is considered obviously rational in women. Women are told to be choosy because childbearing and childrearing are high-cost, high-risk investments. Correct. Men make a parallel judgment about a different form of exposure: long-term financial obligations, custody risk, and legal vulnerability after exit. That is not a failure to grow up. It is what people do once they’ve grown up.

And as marriage falls, fertility tends to fall with it. Declining fertility produces the worker-to-retiree ratios that make Social Security mathematically impossible. These are not disconnected developments. A society that makes marriage less attractive should not act shocked when fewer people build families inside it—or when the systems that depend on those families start to buckle.

The Fault in No-Fault

The theory behind no-fault is that courts should not have to decide who was the bad spouse. Marriage is too intimate, too emotional, too morally tangled for that kind of adjudication.

That sounds humane. It is also wrong as a matter of contract law and basic principle.

Fault in contract is not the same thing as moral condemnation. If someone breaks a lease, the court does not determine whether they are wicked. It determines whether they breached the lease agreement. Marriage is more intimate than a rental, but the legal question is just as legible: did one party terminate the agreement unilaterally, and if so, on what grounds?

Courts are already deep in the intimate wreckage of family life. They divide assets, evaluate custody, assign support, hear abuse allegations, and issue protective orders. We have somehow decided that the relationship is too personal for accountability but not too personal for compulsory financial transfer. That is an oddly selective form of jurisprudence.

And the irony is worse than that. Under a fault-based regime, abusive conduct could matter inside the dissolution itself. Under no-fault, the formal exit terms are less tied to conduct. The spouse who was betrayed and the spouse who betrayed may stand in much the same position. The spouse who was beaten and the spouse who did the beating may still exit through a framework built to sidestep responsibility.

What no-fault eliminated wasn’t the cruelty of adversarial proceedings—anyone with a divorce decree can tell you the cruelty survived just fine. It removed the enforcement principle that gave the agreement real meaning.

Without that, marriage is not a binding covenant in any serious legal sense. It is a ceremony followed by a state-administered financial regime.

Fixing Your Marriage

This is not an argument against marriage. It is an argument that marriage law has drifted so far from ordinary principles of obligation, breach, and reliance that the institution no longer functions like the commitment it claims to be. The demographic data increasingly reflect that.

A sane reform would not abolish no-fault divorce. It would limit it.

If both parties agree the marriage is over, no-fault dissolution should remain available—mutual termination is a standard feature of any agreement. But if one party wants out and the other does not, unilateral exit should require either cause or materially different financial treatment. Not because courts should police romance, but because a contract where either party can walk away unilaterally without consequence is a contract without an enforcement mechanism. Which is to say, not a contract.

Beyond that, support should account for earning capacity, not just current income. Cohabitation and shared living arrangements should matter. Dual-income marriages should recognize the reduced financial risk and sacrifice. Custody presumptions should be genuinely equal rather than rhetorically equal and practically lopsided.

These reforms are obvious as contract principles and nearly impossible as politics. Family law is state law. Reform means fifty separate fights. The beneficiaries of the current system are organized and morally legible. The people most exposed to its downside often do not understand the terms until they are already inside them. And any proposal to make unilateral divorce less frictionless will be caricatured as an attempt to trap women, regardless of what it actually says.

So the practical alternative is obvious: fewer people will sign.

They will still live together. They will still sleep together, have children, bicker, make up, cheat, break up, and leave dishes in the sink. They will simply do less of it inside of a marriage.

For fifty years, we have treated marriage as sacred when we want to celebrate it and contractual when we want to dissolve it. Sacred when we want to shield it from scrutiny. Contractual when we want courts to transfer assets and impose obligations. The result is an institution with the enforceability of a handshake and the financial exposure of a leveraged buyout.

How romantic.

And it’s wrong.

Subscribe now

If you actually measured the economy by the volume of transactions that are cooperative versus competitive, you’d have a hard time finding much competition at all. Cooperation would be almost the entire picture. We’ve named the whole system after the thin, selective edge of market rivalry, like naming the Gulf of Mexico America the “Coast of Texas,” which, honestly, Texas might be open to.

Inside the Machine

Pick any company. Inside it, the governing principle is coordination, not competition. Engineers build components that fit together. Sales teams celebrate each other’s deals. Production lines synchronize. Nobody runs a company by having the marketing department compete with IT for survival.

We actually know what happens when companies try to make their insides competitive, because Enron tried it. Jeff Skilling’s rank-and-yank system—fire the bottom 15% every year, force employees into zero-sum competition for survival—was celebrated as Darwinian genius, right up until it wasn’t. Traders hoarded information. Teams sabotaged each other. The company that was supposed to prove the power of internal competition became the most spectacular corporate collapse of its era. That wasn’t a coincidence. Within-group competition destroys between-group performance. Always has.

The irony is that the most successful “capitalist” enterprises on Earth are internally planned economies. Apple doesn’t let the iPhone division bid against the Mac division for engineering talent on an open market. It allocates resources by committee — who works on what, how much they’re paid, what gets built. So does Google. So does every company that’s ever functioned. The invisible hand is what happens between firms. Inside them, it’s coordination all the way down.

Think About Your Week

Even between firms, the competitive layer is thinner than the textbook suggests. You don’t price-shop which grocery store to visit every Saturday—you go to the same one, because it’s close, you know where your favorite chips are, and the checkout girl is cute. You don’t wake up and scour job boards to decide who to work for today. You commute to the same place on autopilot. You don’t renegotiate with your landscaper every spring, or run a competitive bid for who gets to cut your hair this month, or comparison-shop your mechanic against every other mechanic in the tri-state area. You have a dentist. You have a dry cleaner. You have a guy.

Almost all of your economic life is repeat relationships with people you trust, conducted on the basis of familiarity and habit rather than price competition. The textbook economy—anonymous buyers meeting anonymous sellers in open price-clearing markets—is something you dip into occasionally. The relationship economy is where you actually live.

And it didn’t happen by accident. Your barber earns your repeat business by remembering how you like it cut. Your mechanic earns it by not ripping you off when he could. The restaurant on the corner earns it by knowing your name. These aren’t sentimental observations. They’re economic ones. Trust is a capital asset. It compounds. Every honest transaction makes the next one cheaper, because you skip the due diligence, the comparison shopping, the contract negotiation. The cooperative economy isn’t just warm and fuzzy—it’s efficient. It’s where transaction costs go to die.

The Three Floors

The historian Fernand Braudel spent decades studying economic life from the fifteenth century to the eighteenth, and what he found was a three-layered structure that has never been seriously challenged.

The ground floor is what Braudel called material life—self-sufficiency, routine, household production. People sowing wheat as they always had, cooking food as they always had, making clothes and tools and shelter through patterns handed down across generations. Even today, the non-market economy—home cooking, childcare by parents, favors between neighbors, DIY repairs, the uncounted labor that holds daily life together—represents 30 to 40 percent of GDP-equivalent activity in industrialized countries. A third of the economy doesn’t even show up in the economy. It’s paid in pizza and beer, when it’s paid at all.

Above material life sits the market economy—the zone of transparent local exchange where producer meets consumer, prices are visible, and competition operates within known rules. The butcher, the baker, the candlestick maker. It’s real, and it functions roughly the way Adam Smith described, and it’s most of the tracked economy.

But Braudel’s most provocative claim is about the layer above that. What we call capitalism—the zone where a few powerful actors accumulate wealth, manipulate long-distance trade, and concentrate market power—is not the fulfillment of the market economy. It’s the opposite. Braudel calls it the “anti-market.” It operates by escaping competition, not by embodying it. The great merchant houses of Amsterdam, Genoa, and Venice didn’t seek perfect competition. They sought monopoly, information asymmetry, privileged access, opacity. The entire point was to get above the transparent market layer and stay there.

So the three floors of the economy are: routine cooperation at the base, trust-based exchange in the broad middle, and monopolistic escape from competition at the top. Competition—the thing we named the whole system after—doesn’t have its own floor. At best it’s a narrow hallway.

Why We Show Up

Your barber cuts your hair for twenty bucks. But he also likes seeing your reaction in the mirror when he’s done, because it’s a craft and he’s good at it. Your coworker stays late to help you fix the presentation not because her incentive structure rewards it but because she likes you and you suck at designing slides. You use your brother-in-law as a realtor because he’s family, sort of. You sponsor the local Little League team, and the logo on those jerseys isn’t a calculated ROI play—but Coach Davis asked and you’ve known him since high school.

People work for money, sure. But they also work for pride, for purpose, for identity, for the look on someone’s face when the thing they built actually works. They show up for each other in a thousand ways that never clear a market and never appear in a model. The cooperative economy isn’t an abstraction. It’s how communities get built. How neighbors become friends. How people find out what they’re good at and offer it to each other.

That’s capitalism in its ordinary form. Not the competitive baggage we’ve attached to it—the thing you actually experience. People showing up, building trust, solving each other’s problems, and accumulating the kind of capital that doesn’t show up on a balance sheet1 but determines almost everything about the quality of your life. The most productive system humans have ever built at scale, and it works so well that we’ve stopped noticing how much of it is pure, voluntary cooperation.

What the Mislabeling Costs

Which is why getting the name wrong isn’t just a semantic quibble. Call the system “competitive” and you produce a specific, predictable distortion: you tell people that competition is what they’re losing to.

When housing is unaffordable, healthcare bankrupts families, and a degree costs six figures, the narrative writes itself: capitalism did this. The ruthless, dog-eat-dog market did this. And the prescription follows naturally—if competition is the disease, restrain it, regulate it, replace it. Burn it down if you have to.

But the cooperative economy didn’t do any of that. Housing costs aren’t high because local homebuilders are competing too hard. Healthcare isn’t expensive because doctors are locked in cutthroat rivalry for patients. Tuition didn’t triple because universities are engaged in ruthless price competition. Those failures live on Braudel’s third floor—the anti-market, where regulatory capture, monopolistic rent-seeking, and institutional self-preservation have escaped both competition and cooperation.

The mislabeling collapses all three floors into one word and tells people to be angry at the whole building, instead of the upstairs neighbors. That’s how you get a narrative where participating in the market economy feels like complicity rather than cooperation—where showing up and being useful starts to feel naive. If someone tells you the ocean is poison, you won’t swim. And you won’t be wrong for staying on the beach. You’ll just be wrong about the ocean.

Aim at the Right Floor

The problems are real. Insurance companies adding 30% overhead to healthcare—real. Zoning boards protecting incumbents from new housing—real. The tax code capitalizing deductions into home prices and locking out first-time buyers—real. Credential inflation pricing a generation out of careers that don’t really require four-year degrees—real. We can point to dozens of stripped gears in the economic machinery, and we should.

But every one of those failures lives on the third floor. Every one is an instance of powerful actors escaping the cooperative economy—using regulation, complexity, and institutional capture to extract rent from people who are just trying to show up and be useful. The third floor is the part we should be tearing apart. That’s where the bathwater is. The rest of it is the baby.

The ground floor and the middle floor—the parts where you actually live—work. They work because people show up, build relationships, and honor commitments. They work because millions of ordinary transactions, conducted on trust and habit and mutual benefit, compound into something that looks from the outside like an economy but feels from the inside like a community. The plumber who answers his phone on weekends, the contractor who has a spare part in his truck, the freelancer who delivers what she promised—they’re not winning a Darwinian competition. They’re cooperating, iteratively, with a growing circle of people who trust them. And that trust is compounding into a life.

The cooperative economy doesn’t require capital. It doesn’t require credentials. It requires showing up and being useful, and the patience to let that usefulness compound. It’s a barn raising, not a cage fight—the most accessible game in the history of the species. The tragedy of the mislabeling isn’t just that it aims reform at the wrong floor. It’s that it convinces people that not showing up is the rational move.

We named our economic ocean after its coastline. But the ocean is where we swim.

Subscribe now

Despite what the gold bugs say, a growing economy needs a growing money supply. If you make more stuff with the same amount of money, each dollar buys more, prices fall, and every producer in the economy gets the same signal—stop building, start hoarding. At that point, your mattress outperforms your factory. That’s deflation—the mechanism that turns a financial crisis into a Great Depression.

If the economy grows 3%, you need roughly 3% more money. Print exactly that much and prices hold, savers keep their purchasing power, producers get paid fairly. No inflation. No deflation. No wealth transfer. The books balance.

In practice, there are reasons to overshoot slightly. The Fed targets 2% inflation—partly because a small buffer prevents the deflationary spirals the Depression demonstrated, partly because it gives the Fed room to cut rates during downturns, and partly because workers rarely accept wage decreases from their employers, so mild inflation lets the labor market adjust without requiring actual pay cuts. That 2% is defensible. The problem isn’t the 2%. It’s the additional 2-3 points beyond it.

The Gap

Since 1971—when the dollar’s last tether to gold was severed and money creation became a purely political decision—the US money supply has expanded at roughly 7% per year.1 The real economy has grown at roughly 3%.

The US gets a larger allowance than that domestic 3% alone would suggest. The dollar is the world’s reserve currency—oil trades in dollars, international debt is denominated in dollars, foreign central banks hold trillions in dollar reserves. When the global economy grows, demand for dollars grows with it, regardless of what American factories produced that year. That’s a genuine privilege, and it widens the non-inflationary window for money creation beyond domestic GDP.

Even so. Even granting the reserve currency bonus. Even granting the 2% inflation buffer. There is an enormous gap between how much money we’ve created and how much the economy earned the right to create. The rest was fiction. Dollars that say “someone created value” when nobody did. Not counterfeit in the legal sense. They cleared the same banks, spent at the same registers. But economically? A check drawn on an account with no deposits. The money is real. The production that would justify it never was.

The unearned dollars don’t vanish. Some show up in the consumer prices that CPI faithfully tracks—your 2-3% official inflation. The rest pour into asset prices: stocks, real estate, anything scarce enough to soak up excess money. Economists don’t call this inflation. They call it “appreciation.” If you own the assets, it feels like wealth. If you’re trying to buy them on a salary, it feels like the ladder being pulled up in front of you. Same show. Different seat in the theater.

Every monetary pathology we argue about traces back to this gap. Asset inflation. Wealth concentration. The impossibility of saving your way to a down payment. Occupy Wall Street and the Tea Party. You don’t need a conspiracy theory. You need a calculator and two numbers: money created, value produced. The difference is the debasement.

The Actual Monetary Policy

From talk-show pundits to FOMC watchers, every speech about “the Fed” is arguing about the same thing: the numerator. How many dollars should we print? More? Fewer? Tighten or ease? The entire macroeconomic establishment has spent decades locked in a cage match over the speed of the printer.

But they rarely mention the denominator.

How much production earned those dollars? What did we actually build? Where’s the growth to justify all this expansion?

A company that obsesses over its spending while ignoring its revenue doesn’t have a cost problem. It has a business problem. The correct expense level is whatever your earnings can support. A company growing 20% can justify aggressive spending—even foolish spending. A company growing 2% cannot, and no amount of arguing about the budget changes that math.

If the economy grew at 5% instead of 3%, we could run the printer nearly twice as fast and nobody would get hurt. Wages would rise in real terms. Assets wouldn’t need to absorb the excess. The gap that’s been hollowing out the middle class for two generations would close on its own. The money question would answer itself.

Which means the most important monetary policy question in America has nothing to do with interest rates, quantitative easing, or Jerome Powell’s press conferences. It’s this: why does the economy only grow at 3%? What changed? And what would it take to fix it?

If the non-inflationary money supply is determined by productive growth, then everything that drives productive growth is—mechanically, not metaphorically—monetary policy. Education, regulation, infrastructure, tax policy, energy costs—they don’t just affect “the economy” in the vague way politicians mean it. They determine how many dollars the economy earns the right to create without debasement. They set the denominator. The Fed decides how much we print. This decides how much we earn.

The Doom Loop

When money creation consistently outpaces what the economy earns, capital flows toward whatever the expanding money supply makes profitable. And what excess money makes profitable is owning things, not building them. Why break ground on a factory—seven years of permitting, billions in capital, a decade to break even—when real estate appreciates automatically as the money supply expands? Why fund R&D with uncertain returns when financial engineering delivers reliable beats every quarter?

The gap is self-reinforcing. Low productivity means money creation outruns what’s earned, which inflates assets, which redirects capital from production to ownership, which drags productivity lower, which widens the gap further.

Labor productivity—output per worker per hour—averaged 2.8% annual growth from 1947 to 1973.2 The era when we built the interstate highways, the suburban middle class, and the industrial base that won the Cold War. Since then, it’s averaged roughly 1.5%. GDP still managed 3% growth, but only because the labor force kept expanding—women entering the workforce, immigration, population growth. The economy grew because more people were working, not because each worker was producing more. And that fig leaf is falling. As the population ages and birth rates collapse, the worker pipeline that subsidized GDP growth is drying up. Without a productivity resurgence, total output growth slows—and the earn-to-print gap widens from here.

The inflection in productivity coincides with the end of the gold standard—the moment money creation stopped requiring an external justification. That doesn’t mean gold is the answer, but it was a constraint. A system that lets you conjure wealth on paper without producing it in reality will, over time, produce less reality. The incentives guarantee it.

A Better Policy Debate

Why have millions of prime-age workers dropped out of the labor force entirely—not retired, not in school, just gone? Why does a semiconductor fab take five to seven years to permit in the US when Taiwan does it in two? Why does the tax code treat buying an existing asset identically to funding a genuinely new one, in an economy where the money supply inflates asset prices automatically? Why does the most technologically advanced civilization in history have lower productivity growth than it did when factories ran on steam?

These aren’t secondary questions. They are the monetary policy questions. Every point of productivity we recover is a point of non-inflationary money creation we earn. Every structural barrier we leave in place is a dollar we’ll print without earning, which will inflate an asset, which will redirect capital away from the production that would have earned it. The denominator is the whole game.

The Fed debate is comfortable because it has clear sides, familiar characters, and the illusion of a lever someone can pull. The productivity debate is uncomfortable because the answers are structural, slow, diffuse, and step on every entrenched interest in the economy. But the Fed debate, at best, optimizes the distribution of a gap that shouldn’t exist. The productivity debate is about closing it.

We did 2.8% productivity growth for a quarter century once, and those workers didn’t have computers. Now we have AI. The question isn’t whether the denominator can move. It’s whether we’ll notice it’s the variable that matters before the earn-to-print gap widens past the point of gentle correction.

The Real Printer

The printer was never the point. It was always supposed to be a receipt for production that already happened, not a substitute for production that didn’t. How much should we print? However much we earned. The answer is always that simple, and that inconvenient.

America doesn’t have a money-printing problem. It has a building problem. We print too much because we produce too little, and we produce too little partly because printing is easier.

We used to have an economy that made things worth printing money for. It’s worth asking what happened to it—and what we’d need to do to get it back.

And the rock was running out.

Subscribe now

Jevons wasn’t predicting mines would go dark next Tuesday. He was extrapolating. British coal consumption was growing at 3.5% per year, doubling every twenty years. Known reserves were large but finite. An exponential demand curve hitting a finite supply doesn’t end well. It just ends. Jevons concluded that Britain’s industrial preeminence had an expiration date—and Parliament took him seriously enough to convene a Royal Commission to investigate.

Oil had been discovered in Pennsylvania six years earlier, but in 1865 it was still a curiosity—good for kerosene lamps, not much else. The internal combustion engine was decades away. But oil turned out to be better than coal in nearly every dimension—more energy-dense, more portable, more versatile, lower emissions. Then came natural gas, cleaner still. Then nuclear fission, which packed more energy into a kilogram of uranium than a thousand tons of coal. Then solar, which harvests energy from a source that will burn for another five billion years.

Jevons saw a countdown. He assumed it was counting down to the end of industrial energy. It was counting down to the end of coal.

The Pattern

Jevons was the first to formalize the error, but hardly the last to make it.

Over a century later, in 1972, the Club of Rome commissioned a team of MIT researchers to do what Jevons had done—but bigger. Using a computer model called World3, Donella Meadows and her colleagues simulated the interaction of population growth, food production, industrial output, resource consumption, and pollution. The result was The Limits to Growth, one of the most influential publications of the twentieth century.

The conclusions were specific and stark. At prevailing rates of consumption, gold reserves would be exhausted in 9 years. Mercury in 13. Tin in 15. Petroleum in 20. Natural gas in 22. The model’s “standard run” projected global economic collapse beginning around 2020. The book sold thirty million copies. Its specific resource predictions were, almost without exception, wrong.

Not because the authors were careless — the methodology was serious enough to provoke fifty years of both qualified debate and unjustified policy. The problem was structural, and it was the same one Jevons had: the model could project consumption curves with great precision. It could not predict the inventions that would bend them.

In 1972, it was reasonable to project that population growth would outstrip agricultural capacity. Thomas Malthus had made essentially the same argument in 1798, and for most of human history, he’d been right—population pressed against the food supply, and the result was perennial famine. What neither Malthus nor the MIT model could anticipate was Fritz Haber, who figured out how to synthesize ammonia from atmospheric nitrogen—literally conjuring fertilizer from thin air. Or Norman Borlaug, whose semi-dwarf wheat varieties doubled and tripled yields across Mexico, India, and Pakistan. India’s wheat harvest jumped from 11 million tons to 20 million in six years. Borlaug won the Nobel Peace Prize for work that would eventually save over a billion lives. Today, the planet feeds eight billion people. Caloric production is now higher per-capita than in 1972, when there were fewer than four billion of us.

Proved petroleum reserves in 1972 were roughly 550 billion barrels. After fifty years of continuous extraction and presidential candidates campaigning to “drill baby drill,” proved reserves now stand at 1.7 trillion barrels—three times what the model treated as the total supply. Hydraulic fracturing, horizontal drilling, and deepwater exploration kept unlocking oil that no model had anticipated. The model assumed fixed resources consumed by growing demand. Capitalism disagreed.

In 1980, the economist Julian Simon made the pattern into a bet. He challenged Paul Ehrlich—author of The Population Bomb, which had predicted mass starvation in the 1970s and 80s—to pick any five commodity metals. If their inflation-adjusted prices rose over the next decade, Ehrlich wins. If they fell, Simon wins. Ehrlich chose chromium, copper, nickel, tin, and tungsten.

By 1990, every single one was cheaper. Ehrlich mailed Simon a check for $576.07.

Ehrlich looked at commodity prices and saw a countdown to zero supply. Simon looked at the same prices and saw a countdown to zero cost.

The countdown to zero cost won. It always does.

The pattern is so consistent that it implies something radical about how commodities actually work. Over the long run, the price of a commodity doesn’t rise, it trends toward zero.

The Price of Everything

A commodity is a product nobody really cares about. Not because it’s useless—because it’s undifferentiated. Nobody cares where their copper comes from, who mined it, or what the company’s mission statement says. Copper is copper. It doesn’t have a brand, it has a spot on the periodic table. Which means the only thing distinguishing one producer’s copper from another’s is the price. And when price is the only axis of competition, every producer on earth is doing the same thing: trying to deliver the same product for less money.

That’s a race with only one direction. Nobody is trying to make copper more expensive to produce. There’s no artisanal copper market. Every engineer, every process improvement, every new extraction technique pushes the cost down. These gains accumulate permanently across every commodity—you can’t un-learn electrolytic smelting, you can’t un-discover horizontal drilling. Each efficiency becomes the new floor, and the next improvement builds on top of it. Aluminum was once rarer than gold—Napoleon III served state dinners on aluminum plates while lesser guests ate off silver. Then Charles Martin Hall figured out electrolytic smelting in 1886 and collapsed the price by 98%. Today we wrap sandwiches in it and throw it away. The metal didn’t change. The engineering caught up to the geology.

But that’s only the first force. The second is more powerful: substitution. People don’t actually want copper. They want what copper does—conducts electricity, carries signals, moves heat. The commodity is just a delivery mechanism for a function. And delivery mechanisms get replaced. Fiber optics replaced copper in telecom. Aluminum replaced steel in manufacturing. The cost of a lumen-hour—the basic unit of being able to see after dark—has fallen by a factor of 500,000 since the Babylonians were burning sesame oil. A day’s labor in 1800 BC bought ten minutes of dim lamplight. A day’s labor today buys 20,000 hours of LED illumination. When the incumbent commodity can’t get cheap enough fast enough, something else delivers the same function for less.

Two forces. Both permanent. Both one-directional. Competition within a commodity pushes the cost down. Competition across commodities replaces the expensive one entirely. The long-run price of every commodity is zero—either extraction gets so efficient its only real cost is the shipping, or something better comes along and demand collapses. There is no third option.

Solar energy is the proof running in real time, fast enough to watch. In 1977, a watt of solar capacity cost $76. Today it costs roughly $0.20—a 99.7% decline, and the curve hasn’t flattened. Every time cumulative production doubles, cost falls by another quarter. The sun delivers 10,000 times current global energy demand to the planet’s surface continuously. It doesn’t send an invoice. The sand used to make the panels is so abundant, we sweep it out of our homes. The only real cost is the engineering, and engineering costs fall on a learning curve. They always have.

The scarcity of the old thing financed the new one. Solar was a novelty until fossil fuels got expensive enough to make it worth funding. By 2024, new solar installations generated electricity more cheaply than existing coal plants in most of the world—not new coal plants, but ones whose construction costs are already sunk. Jevons’ coal question answered itself. Not with a better mine, but with a technology he couldn’t have modeled because the photovoltaic effect wouldn’t be explained for another forty years.

Reading the Clock Backwards

Every generation faces a coal question—a resource it depends on, running out, with math to prove it. Every generation makes the same error: reading a countdown to replacement as a countdown to catastrophe. But there’s a subtler error, and it does more damage: responding to the countdown with restraint instead of invention.

Antibiotic resistance is a coal question—and a case study in getting the response wrong. Bacterial evolution is outrunning our pharmaceutical toolkit. The WHO projects 10 million annual deaths from resistant infections by 2050, up from 1.3 million today.

Paging Dr. Jevons.

The crisis isn’t biological. It’s economic. Our response to resistance has been to use fewer antibiotics—prescribe less, restrict access, steward carefully. Burn less coal. And the predictable result is that we’ve made antibiotic development one of the worst investments in pharmaceutical R&D. You can’t recoup billions in development costs on a drug doctors are told to use as sparingly as possible. So they stopped developing them. The entire global antibiotic pipeline—every drug being developed by every company on earth—contains roughly 45 candidates. For context, oncology has over 6,000. Major pharmaceutical companies have largely exited the don’t-die-from-infections space. We told the market to use less, and the market stopped making more.

Meanwhile, the alternatives exist. Phage therapy—using viruses that target specific bacteria—has been practiced for a century in Eastern Europe and is only now entering Western clinical trials. CRISPR-based antimicrobials can be programmed to kill specific resistant strains. AI-driven drug discovery has compressed candidate identification from years to weeks. None of these were in the WHO projection, because projections model the constraint, not the response. But they’ll only arrive at scale if we let the price signal do its job—reward the solution instead of rationing the problem.

The Condition

None of this is automatic. The countdown to zero cost only runs if the people solving the problem get to profit from solving it.

Britain didn’t answer the coal question by burning less coal—it developed the technologies that made coal obsolete. Borlaug didn’t feed a billion people by convincing them to eat less—he bred wheat that produced more grain per stalk. The mechanism that resolves constraints isn’t restraint. It’s invention. And invention requires an apparatus—universities, research labs, capital markets, legal frameworks—capable of converting problems into solutions.

The apparatus works when we fund it at the scale the problem demands. We landed on the moon eight years after deciding to try. The Manhattan Project took three years from inception to detonation. Fusion energy has been “thirty years away” for sixty years—but total global investment in fusion research over those six decades is roughly what the Pentagon spends in a single month. The timeline is a function of funding, not physics. When we actually commit, the timeline compresses in ways the skeptics never model—because they’re extrapolating from the underfunded version of the effort.

The real risk is never the constraint itself. It’s dismantling the apparatus that resolves constraints. Defund basic research. Politicize scientific institutions. Strangle nuclear energy in permitting for four decades. Restrict capital formation so only the already-wealthy can invest in transformative technology. Do these things and the other countdown wins—the one to zero supply. Not because the constraint was unsolvable, but because we took apart the machine that solves things.

Zero

The scarcity never goes away. It moves up the stack as each constraint is resolved by something better than what it replaced. The thing you’re worried about today will become essentially free. Then you’ll worry about the next thing. That’s what progress feels like from the inside.

But here’s the part we keep getting wrong. When the countdown appears—when the math says the resource is finite and the curve is exponential—every instinct says to slow down. Use less. Conserve. Ration. Buy time.

That instinct is exactly backward. The rising price isn’t the crisis. It’s the funding mechanism for whatever comes next. Solar didn’t get cheap because “Just Stop Oil” activists finally glued their hands to enough surfaces. It got cheap because fossil fuels got expensive enough to make the investment worthwhile. Every dollar of scarcity pricing that accrues to the old thing is a dollar of incentive to build the new one. Eli Lilly isn’t going to spend two billion dollars developing a novel antibiotic that doctors are told to prescribe as rarely as possible. But a company staring at a market where existing antibiotics don’t work anymore—where resistant infections are killing ten million people a year—will absolutely spend that money, if we let the price tell the truth.

Rationing the old thing and funding the new thing feel like the same policy. They are opposites. One stretches the countdown. The other ends it.

Running out isn’t a reason to cut back. It’s a reason to surge forward.

The word itself has been weaponized. “Austerity” sounds like deprivation—cold meals and shuttered schools. It’s deployed the way a protection racket deploys threats: nice economy you’ve got here, shame if something happened to it. But even under austerity, the government would still spend trillions. Just fewer trillions.

Subscribe now

What they’d rather you not notice is that threat contains its own rebuttal. If the spending were working—if the trillions allocated to poverty, healthcare, education, and housing were producing results proportional to their cost—the fiscal crisis demanding austerity wouldn’t exist. You don’t need to cut a budget that’s generating returns. Stable enterprises don’t hold going-out-of-business sales.

The need for austerity is itself the evidence that the spending already failed.

Money Pits and Cash Drains

Not all government spending is created equal, and the distinction that matters isn’t between “wasteful” and “efficient” spending, or between “liberal” and “conservative” priorities. It’s between spending that can end and spending that can’t.

Call the first category money pits. They’re expensive—sometimes enormously so—but they have a bottom. Rural electrification. The interstate highway system. The eradication of polio. Putting a man on the moon. You set an ambitious goal, throw mountains of money at it, and eventually voilà. The pit is filled in. The project completes. What remains is a civilizational unlock that pays dividends for generations. Nobody is still funding the Rural Electrification Administration because rural America has electricity. Mission accomplished.

The second category is cash drains—spending commitments whose very definitions guarantee they can never be completed. These aren’t pits with a bottom. They’re open pipes.

Take poverty. The federal poverty line isn’t a fixed measure of deprivation—it’s adjusted annually for the cost of living, which rises in lockstep with the economy it’s measured against. Lift every household in America by 20% tomorrow, and within a few years the threshold has followed them up. Worse, the spending is deficit-financed. Deficit spending is inflationary. Inflation raises the cost of living. And the cost of living is exactly what the poverty line tracks. That’s not filling a hole. It’s digging it.

Or consider healthcare. The federal government funds medical research through the NIH, which produces genuine breakthroughs in treatment and diagnosis. Wonderful. But each breakthrough redefines the standard of care—which Medicare and Medicaid are then obligated to provide. The spending funds the innovation that raises the costs that the spending must then cover. There is no bottom in this pit. There are always more ways to extend or improve human life, short of curing mortality itself, and each new capability becomes the new baseline that “adequate” healthcare must meet. The better we get at medicine, the more it costs to provide it universally—not because of failure, but because of success.

The critical difference: money pits produce something and stop. Cash drains produce an obligation to spend more forever. One is an investment with a terminal date. The other is a subscription with no cancel button.

America’s budget crisis isn’t driven by money pits. We’re not going broke building highways. We’re going broke because the overwhelming majority of federal spending—entitlements, open-ended healthcare commitments, poverty programs pegged to relative measures—is structurally incapable of completion. These programs don’t fail and end. They fail and expand.

The Visibility Trick

What makes the austerity threat so hollow is that the economy isn’t struggling under this weight. It’s thriving despite it. The private sector funds itself, voluntarily. No one compels the investment, forces the trades, or conscripts the workers. A hundred and thirty million people show up because the exchange works for them. They earn enough to support themselves and their families directly and still generate a tax base of trillions to fund everything else.

The giant sucking sound isn’t the economy failing. It’s the economy succeeding so thoroughly that it can finance staggering waste and keep going. When politicians warn that spending less would hurt the economy, they have it exactly backwards. The economy is the source of the money. Austerity means pouring less of it down the drains.

The standard political playbook for resisting spending cuts is called the Washington Monument Strategy, named after the National Park Service’s habit of threatening to close its most popular attraction whenever its budget faced reductions. The logic is transparent: make the public associate budget cuts with the loss of something they value, rather than with the elimination of something they’d never notice.

It works because most people reasonably assume that if a program exists, it must be doing something important. But this misunderstands the nature of government institutions. Programs don’t persist because they work. They persist because they employ people, service contractors, and generate the kind of activity that registers as GDP. A program can fail completely at its stated mission while succeeding entirely at its institutional one: perpetuation.

This is why threatened “cuts” almost never start with the back office. They start with teachers, firefighters, and air traffic controllers—the employees the public actually interacts with. The administrative layers, the compliance departments, the consultants who study whether the consultants are needed—those are never on the chopping block. They’re the ones drafting the press releases about how devastating the cuts will be.

The Austerity Paradox

The numbers confirm what the structure predicts. Federal spending on poverty has increased for sixty years while poverty rates remain largely unchanged. Education spending per pupil has roughly tripled in real terms since 1970 while test scores have fallen. Healthcare spending has grown at multiples of GDP growth while life expectancy lags every peer nation.

If these programs were working, the spending would be shrinking. Each dollar would solve a piece of the problem until the problem got smaller and the budget could follow it down. Instead, the spending grows because it isn’t working—and the programs’ structural design guarantees it never will.

So when the bill comes due and someone suggests spending less, we’re not being threatened with the loss of effective services. We’re being threatened with the reduction of a bureaucratic apparatus that has demonstrably failed to deliver on its promises.

What Austerity Actually Looks Like

Canada in the 1990s cut federal spending by roughly 20% over four years. The predicted catastrophe never materialized. The economy grew, the budget balanced, and the programs that survived were the ones that actually delivered value. Estonia after the 2008 crisis chose austerity over stimulus when every respectable economist called it suicide. Their economy recovered faster than their stimulus-spending neighbors.

Argentina offers the most recently vivid case. When Javier Milei took office in December 2023, annual inflation was approaching 300%. He took a chainsaw to the federal budget—literally, his iconic campaign prop—and every credible institution predicted collapse. Eighteen months later, monthly inflation had fallen below 3%, the government posted its first fiscal surplus in fourteen years, and the poverty rate dropped to its lowest level since 2018.

The pattern is consistent: forced discipline eliminates the lowest-value spending first, because that spending has the weakest constituencies and the least defensible outcomes. What survives is, almost by definition, what people actually need. The overhead burns off. The core remains.

This shouldn’t surprise anyone. It’s how every other institution on earth operates. When a company faces a revenue shortfall, it cuts the projects that aren’t working. When a household tightens its budget, the magazine nobody reads gets cancelled before the grocery bill. Only in government does the opposite logic prevail: threaten to cut the groceries, then use the natural outcry to protect the unread magazines.

The debate over austerity is always framed as: can we afford to spend less? The question it dodges: can we afford to keep spending like this—on programs structurally incapable of completion, funded by debt our children will service for the rest of their lives?

Austerity is not a policy choice. It’s what happens when the bond market loses patience. Spending cuts have real costs—people depend on these programs even when the programs fail at scale. But the alternative isn’t just to continue spending indefinitely. It’s forced restructuring under crisis conditions, with little control over what gets cut and no time to protect what works. Now that’s what a threat sounds like.

Money works the same way. It’s a tool. You use it to build the life you want. And yet somewhere between the invention of the 401(k) and the rise of the TikTok influencer, we convinced ourselves that accumulation was the point—that the person with the most hammers wins. That anyone who spends their tool on something that doesn’t generate a return is doing it wrong.

Which brings us to lottery tickets.

Subscribe now

The Two Dollar Verdict

Buying a lottery ticket is, according to the consensus of every financially literate person you’ve ever met, a uniquely stupid thing to do. The expected value is negative. The odds are astronomically stacked against you. You are, mathematically speaking, barely better off than simply setting your money on fire.

And they’re right. Mathematically.

They’re also applying a standard they don’t apply to anything else, including to themselves.

Is it rational to spend $400 on a wedding dress you’ll wear once? Is it rational to pay off your mortgage early when the stock market would almost certainly deliver a better return over the same period? Is it rational to spend $15 on a cocktail that takes three minutes to drink and leaves you marginally worse at everything you do for the rest of the evening?

Nearly all of our spending is irrational. That’s normal. We call it: human. We understand, implicitly, that people spend money on feelings, on meaning, on ritual, on the small daily pleasures that make the ordinary feel like it was worth showing up for. We extend that grace to almost every purchase a person can make.

But not lottery tickets.

The Dopamine Economy

Neuroscience has known for decades what personal finance hasn’t caught up to: the anticipation of a reward activates the dopamine system more reliably than the reward itself. You don’t get the rush when you win. You get it when you might.

A two-dollar lottery ticket, purchased on a Sunday night before a Wednesday drawing, buys three days of imagining a completely different life. Not a marginally better one. A different one. Three days of deciding whether you’d quit or give notice. Whether you’d tell anyone or disappear for a while. Whether you’d buy your mother a house or just pay off her car and never say where the money came from.

Most things you buy offer a slightly better Tuesday. The lottery ticket lets you rehearse a different future entirely. That’s not a minor distinction, and it’s not irrational. It’s the cheapest form of something that people with better options get for free. It’s hope.

When The Math Actually Changes

There’s a version of the lottery ticket critique that almost works. It goes like this: if you saved that money instead, it would compound into something meaningful.

This argument is correct for a certain kind of person. If you’re making $200,000 a year and you’re buying lottery tickets instead of maxing your retirement accounts, you’re leaving real money on the table. The opportunity cost is genuine.

But nobody actually cares if a lawyer making $200,000 buys a Powerball ticket. It’s the poors we’re talking about.

And that argument is even weaker than it sounds.

If you’re earning $28,000 a year and you skip the twice-weekly lottery ticket, you save sixteen bucks a month. At the end of a year you have around $200. After a decade, with optimistic market returns, you have maybe $2,500. That’s not a down payment. That’s not a retirement. It’s not even much of an emergency fund. That’s $2,500 after ten years of discipline—a number that does not materially change the trajectory of your life in any direction after a period of time that statistically outlasted the marriage the wedding dress was meant to celebrate.

The expected-value calculation that makes lottery tickets look stupid assumes you’re someone for whom small amounts of capital compound into large amounts of capital. For tens of millions of Americans, that assumption is simply false. The alternative to the lottery ticket isn’t a comfortable retirement. It’s the same life, minus the hope.

Against the plausible alternatives, a non-zero chance at transformation looks different than it does on a spreadsheet. Not smart, exactly. But not obviously dumb either. And seven days of genuine excitement per week, fifty weeks a year1, for $16 a month—that’s a price point a lot of therapists would struggle to match, whatever your copay.

But maybe you think it’s worse than irrational. Maybe you think it’s gambling.

The Shame

The line between a stupid purchase and an addiction has nothing to do with dollars. Someone funneling their entire paycheck into scratchers has a problem—but so does someone spending $300 a week on cocaine, regardless of how big a pile they can afford. The pathology isn’t the price. It’s the compulsion. We don’t evaluate Snickers bars by their worst-case consumer. We sell them to kids.

A young lawyer passes the bar and buys a Rolex. We call that a milestone. It’s not a good investment—most watches depreciate, and the money would perform better in an index fund by almost any measure. But we understand that he’s not buying a watch. He’s buying a feeling about his future. He’s buying a signal to himself and everyone around him that something has changed, that he’s arrived somewhere, that the work paid off. We find this not just acceptable but admirable.

A woman earning fourteen bucks an hour buys a lottery ticket on her way home from a double shift. We call that a tax on the poor. A failure of financial literacy. Evidence that some people just don’t understand how money works. That maybe they actually deserve what they get.

Both of them spent money they didn’t need to spend, on a feeling they wanted to have. One of those feelings is legible to the people doing the judging. The other isn’t.

The critique of lottery tickets is dressed in the language of financial rationality. But financial rationality, applied consistently, would also indict the Rolex, the destination wedding, the brand-name interview suit, the paid-off mortgage, and every other purchase humans make for reasons that can’t be reduced to a pro forma. We apply it selectively, and the selection is not random.

It’s poor.

The Point of a Hammer

People spend money on hope. They always have. They buy candles for altars, rounds for strangers, fireworks for the Fourth of July. They splurge on the good Champagne for the occasion that deserves it. They tip more than they should when the server reminds them of their kid. None of this maximizes utility. All of it is completely human, and we don’t judge any of it—because we understand, without needing it explained, that the point of money was never the fucking money.

The point was always to build something. A life that feels like yours. A Tuesday that felt worth showing up for. A moment, however brief, of genuine excitement about what might happen next.

Most of us are just trying to buy a little hope. A little something to look forward to. There’s nothing wrong with that. There never was.

The only thing worth questioning is why we reserve our judgment for the people buying it two dollars at a time.

And hey, if that’s you, good luck.

They didn’t have computers. They didn’t have stochastic models or PhD programs in financial engineering. What they had was a reed stylus, a piece of soft clay, and the compound interest formula—which they’d worked out roughly four millennia before the Bank of England opened its doors. Debts at standard rates doubled every five years. Harvests didn’t. The scribes could see where the math ended up. So they built a mechanism to deal with it.

Subscribe now

We treat financial crises like unsolved engineering problems. If only we had better models, smarter regulators, faster data feeds. The entire apparatus of modern financial regulation—Basel accords, stress tests, macroprudential surveillance. We assume the core challenge is diagnosis, like we just haven’t figured it out yet.

We figured it out before the alphabet.

The Barley-Silver Distinction

For roughly three thousand years—from ancient Sumer through Babylonia, the Hebrew kingdoms, Greece, and Rome—periodic debt cancellation was standard economic policy. Not utopian theory. Not a fringe religious practice. Documented, enforced, routine. The Sumerian word was amargi. The Babylonian term was misharum. The Hebrew version became the Jubilee. The tradition was old enough to be ancient by the time Rome was fighting over it. Different languages, same mechanism, same math.

The mechanism had three parts, repeated across centuries with remarkable consistency. Agrarian and subsistence debts—what we’d call household debt—were periodically cancelled outright. People who had fallen into debt bondage were freed. And pledged land was returned to its original holders, reversing the concentration that had accumulated since the last reset.

But here’s the part that should make every institution that’s ever claimed to be “too big to fail” nervous. Commercial debts between merchants were left intact.

The Babylonians called it the barley-silver distinction. Barley debts—subsistence borrowing, accumulated taxes, unpaid fees, the arrears of people whose obligations grew faster than their harvest—got cancelled. Silver debts—commercial credit between traders, self-liquidating loans extended for specific ventures—didn’t. The distinction isn’t complicated. A farmer who owes more grain than his field can produce ends up a debt slave—and a debt slave doesn’t run a farm. A merchant who borrows to ship copper and repays from the proceeds is just doing business. The Babylonians could tell the difference. So can you.

The debtor class wasn’t a collection of irresponsible borrowers who’d overextended themselves on Bronze Age McMansions. Most of these obligations weren’t even loans in the modern sense—credit preceded cash by millennia. They were families whose arrears accumulated faster than the harvest could cover. Compound interest doesn’t care about drought years. Enough time, enough families, same result every time.

The scribes knew this. The clean slate wasn’t charity. It was scheduled maintenance.

The Capture Sequence

If this system worked—and the archaeological record suggests it did, across multiple civilizations and millennia—the obvious question is: what happened to it?

It was destroyed. Specifically, it was destroyed by the people it constrained, using techniques that haven’t changed much in four thousand years.

The sequence is documented repeatedly—in fragmented clay tablets and fragile papyrus scrolls and yellowed parchment—and it is remarkably consistent:

First, an institutional mechanism protects the broad population from excessive wealth concentration. The clean slate, the Jubilee, Roman agrarian laws—the specific form varies, the function doesn’t.

Second, the beneficiaries of concentration develop workarounds. Fictive adoptions to circumvent inheritance rules. Contractual waivers where debtors “voluntarily” surrender their protections. Temple-precinct pledging to move assets beyond the reach of royal decrees. The loopholes are always more creative than the rules.

Third, the mechanism’s enforcers are killed, captured, or politically neutralized. Sparta’s kings Agis and Cleomenes were murdered for attempting debt cancellation. The Gracchi brothers were killed for pursuing land reform in Rome. Caesar was assassinated by a Senate heavy with creditor interests.1 The pattern isn’t subtle.

Fourth—and this is where it gets elegant—the mechanism is formally maintained but substantively gutted. The letter of the law survives. The spirit doesn’t. The single best example is Hillel’s prosbul, invented in the first century BC. Faced with the Torah’s Jubilee requirement to cancel debts every seven years, Rabbi Hillel created a legal clause allowing borrowers to “voluntarily” waive their protections. Sign here, and the sacred debt cancellation no longer applies to you. The ancient equivalent of clicking “I agree” to the Terms of Service—coercion dressed as consent, with just enough procedural formality to satisfy anyone not looking too hard.

Two thousand years later, we still use the same technique. Every credit card agreement, every adjustable-rate mortgage, every arbitration clause buried in the fine print of a contract nobody reads is a direct descendant of the prosbul. The innovation wasn’t financial. It was legal. Make the surrender of protections look like a choice, and nobody has to take them by force.

Fifth, and finally, wealth concentrates to the point of systemic failure. Rome is the endgame that played out completely. The early Republic had land distribution laws, debt limits, and a citizen-soldier class whose economic independence was the foundation of military power. Over several centuries, the creditor oligarchy dismantled every constraint. Public land was absorbed into private latifundia, industrial-scale farms worked by slaves and owned by the elite. The citizen-soldier was replaced by a professional army loyal to individual oligarchic generals. The tax base hollowed out. Military capacity degraded. Political legitimacy collapsed into factional warfare. The barbarians at the gates hadn’t changed. Rome’s internal structure had.

What makes Rome instructive isn’t the fall—empires fall. It’s that the people with the power to implement the correction were precisely the people who benefited from not implementing it. The solution existed. The political will to deploy it had been captured.

Why the Fix Never Comes From Below

There’s a tempting populist reading of all this: if the elites won’t cancel debts, the people should demand it. But this misreads the entire historical record. The clean slate was never a populist achievement. It was always an elite correction—one faction of the elite constraining another.

The Babylonian king cancelled debts not because the masses demanded it, but because the palace needed a functioning tax base and a military drawn from free citizens, not debt slaves. The king’s interests diverged from the creditor class’s interests. That divergence was the mechanism. The masses didn’t need to understand compound interest. They needed a ruler whose survival depended on not letting creditors cannibalize the productive base.

This maps uncomfortably well onto today. The people who feel the squeeze of wealth concentration want to “make billionaires pay their fair share”—but the mechanisms through which concentration actually operates are opaque enough that the policies voters demand often mechanically worsen the problem they’re trying to solve. This isn’t a failure of character. It’s the same structural mismatch the Babylonians identified: expecting the barley class to diagnose a silver disease.

The uncomfortable implication: the correction for financial capture has never come from below. It comes from elite fragmentation—when a faction within the ruling class has both the understanding and the motivation to build counterweights. It’s never the Occupiers versus Wall Street. It’s the Babylonian palace versus the creditor class. The Roman populares versus the optimates. When the elite is unified in its interests, there is no correction.

The rest of us just feel the squeeze.

The Complexity Defense

Modern finance has made the barley-silver distinction functionally impossible. When a mortgage is sliced into tranches, bundled into a collateralized debt obligation, insured by a credit default swap, and held as Tier 1 capital by a bank that clears every transaction in the economy—and then issues credit cards—there is no clean seam to cut along. The Babylonian palace could see every debt because debts were recorded on clay tablets in temple archives. In modern finance, the debt is the money supply. Cancel the debt, cancel the money.

In 2008, the banks really were too interconnected to let fail—not because bankers are special, but because the system had evolved to make the correction mechanism and the correction target the same thing. The clean slate presupposes that you can distinguish what to cancel from what to preserve. Structured finance has made that distinction impossible by design.

And there’s a deeper tradeoff that honest advocates of debt relief have to confront: making the barley-silver distinction doesn’t just reset bad debt. It means less barley credit gets extended in the first place. A lender who knows subsistence debt gets cancelled every cycle will lend less to subsistence borrowers. The Babylonians accepted this—and the economy functioned. The modern system chose the opposite: maximize credit availability, socialize the downside, and call it financial inclusion. More people get loans. More people get buried. The banks get bailed out. We call it a safety net.

Whether the complexity itself is a feature of capture—the system made opaque specifically to prevent correction—or a genuine emergent property of financial innovation is a question worth sitting with. Both are probably true. Financial systems do grow more complex over time for legitimate reasons. And complex systems are harder to reform, which benefits the people who’d rather not be reformed. The two dynamics reinforce each other until you get a financial system that is simultaneously too intricate to understand and too interconnected to reset.

Who Bears the Risk

The standard economics curriculum treats debt as a neutral instrument—a contract between willing parties, priced by the market, cleared by the legal system. Default is treated as an individual failure of the borrower. But four thousand years of evidence suggests something simpler: most people aren’t going to end up on the right side of the compounding equation. They never have. The question is whether we design systems that expect them to, or systems that accept it.

The Babylonians didn’t solve household debt by teaching farmers compound interest. They solved it by making the system absorb the reality that farmers would never beat compound interest. The barley-silver distinction wasn’t taught to barley debtors. It was imposed on silver creditors. The lender bore the risk of extending credit that wouldn’t always be repaid, because the system would periodically cancel it. That didn’t punish lenders. It made them careful. A creditor who knows the slate gets cleaned prices accordingly—extends less, charges less, or doesn’t extend at all.

We’ve built the opposite. Capital One will extend you $30,000 in revolving credit at 27% APR—not despite knowing you can’t sustainably service it, but because the system ensures you can’t escape it. Bankruptcy reform in 2005 made credit card debt harder to discharge. Student loans are effectively permanent. The entire consumer credit apparatus is engineered around the principle that barley debtors must pay, no matter what. This isn’t lending. It’s farming.

The mechanism doesn’t have to be a Jubilee. It doesn’t have to be a specific law. But the standard it points toward is clear: the entity extending credit should bear the cost when that credit can’t be repaid. Not as punishment—as systemic maintenance. Make the silver side eat the barley risk, and the silver side will figure out very quickly how much barley credit the system can actually sustain. They’ll do the math. That’s what they’re good at.

Babylonian scribes used clay tablets and concluded that periodic debt correction was necessary infrastructure, like irrigation. We run the same numbers on supercomputers and arrive at the same conclusion—but we’ve decided the answer is to make the borrowers smarter rather than the lending safer. The math hasn’t changed. The politics has.

Good news. Regulations are easy to cut. They’re written on paper. You can find the ones that don’t work and rip them up. You don’t need to invent new technology or change human nature. You just need a committee with some spine, and the willingness to tell a bureaucracy that the form it’s been stamping for thirty years no longer needs stamping. You can even suggest where they shove the stamp. Difficult politically. Trivial mechanically. The deregulation problem is a solvable problem, and we should solve it.

But that’s not the hard problem.

Subscribe now

The Invisible Gap

The hard problem is the regulations we don’t have.

Not in the progressively anti-progress sense—the reflexive demand for more rules on whatever feels dangerous on Bluesky this week. The Left wants more regulation on firearms and less on abortion access. The Right wants more regulation on immigration and less on energy production. Neither side is “pro-regulation” or “anti-regulation” in any consistent way. They’re pro-regulation for things they find threatening and anti-regulation for things they find acceptable. The entire debate is downstream of political coalition-building, which is why it never produces anything resembling a coherent principle.

The question upstream of that debate is simpler and more uncomfortable: what domains are currently ungoverned where realistic failure modes exist, whose failures spill externalities onto the rest of us, and what’s the cost of finding out the hard way?

That’s not ideology. It’s closer to what an actuary does. Map the territory, estimate the exposure, price the risk. And unfortunately, nobody’s doing it—not because it’s impossible, but because regulatory attention is a finite resource, and we’ve allocated most of ours to legislating which sports leagues should permit testicles.

The Known Unknown

After 2008, we rebuilt the banking system’s guardrails. Banks now hold more capital in reserve, submit to stress tests, and operate under oversight designed to ensure that a wave of bad loans can’t cascade into a global meltdown ever again. Whatever you think of Dodd-Frank, the basic logic was sound: if your lending blows up, and taxpayers are on the hook for the fallout, then someone needs to be watching the books.

Which is why the banks moved their money off the books.

Private credit—a term most people have never heard, which is itself the problem—is what happens when instead of lending directly and keeping loans on their balance sheets, banks redirect them into private, third-party investment funds that don’t have to follow banking rules. It started small. Over the past fifteen years, it’s grown to roughly $2 trillion. To put that in perspective, that’s larger than the entire subprime mortgage market was in 2007.

The mechanics are simple enough. A company needs to borrow money. It’s too leveraged or too risky for a traditional bank, which would have to hold capital against the loan and justify it to regulators. So instead, a private investment fund makes the loan. But the fund isn’t just the high-risk play money of the rich. No, it raised its investment capital from pension funds, insurance companies, endowments—institutions managing the retirement savings of teachers and firefighters and state employees. You know, the pensions that you’re already on the hook for. The loan sits on no bank’s balance sheet. No regulator stress-tests it. No public filing discloses the terms. The company gets its money, the fund earns a fee, and the risk settles quietly into the retirement accounts of people who have no idea it’s there. And if it goes bad? Well, heads they win, tails you lose. Remember too big to fail?

This is by design. The money moved specifically to escape oversight, and the oversight didn’t follow. When the default cycle comes—and it will, because that’s what cycles do—the losses won’t surface at a bank, where regulators can see them coming. They’ll surface in pension funds and insurance reserves, and the people holding the bag will be the last to find out.

Everyone who was around in 2008 remembers the sickening discovery that nobody knew who owed what to whom. We’re building that exact opacity again, in a market most people don’t know exists, supervised by no one in particular. The reason nobody is regulating it isn’t that someone weighed the risks and decided it was fine. It’s that it grew up in a space where the rules weren’t, and the people shouting for regulations don’t know what’s actually risky.

The Unknown Unknown

Private credit is at least a risk you can point at. Somebody knows what it is, even if most people don’t. The harder category is the risks nobody’s identified yet—not because they’re unidentifiable, but because we don’t know what we don’t know.

Facebook ran internal studies showing that Instagram was damaging teenage mental health, especially for young girls. This is usually told as a scandal: they knew and they hid it. But step back and notice the stranger part. Nobody required them to look. No regulation mandated the study. Facebook did the research on its own, and when the findings leaked, the reward for having investigated their own product was a congressional hearing and a billion-dollar PR crisis.

Now imagine you run the next platform. The lesson is unmistakable: if you study your product’s effects and find something bad, you’ll be punished for knowing. If you never look, you never knew. Which means you can never have concealed anything. The rational move is to not ask the question. The public flogging goes much easier that way.

That’s where we are with most of the technologies reshaping daily life. The companies that could study their own effects have every incentive not to, and nobody outside those companies has the data to do it instead. Many of them still study their own products. These are good people, trying to build good things. But we’re increasingly incentivizing against good behavior.

The obvious alternative is mandatory disclosure—force companies to hand over their data so researchers can look for harm. But think about what that actually means. It’s someone showing up at your door and saying “give me access to everything—all your proprietary research, all your internal metrics, all your user data.” You ask what they’re looking for. They say: “We won’t know until we find it.” That’s not a workable basis for policy in any domain. It’s a fishing expedition, and everybody knows it. Which is why it never gets past the hearing stage.

So the unknown unknowns stay... unknown. Not because they can’t be found, but because the incentives encourage us not to look too closely. Social media’s impact on adolescent mental health wasn’t some deep mystery that required a scientific breakthrough. It required having a teenager. But, presumably, we could’ve started asking the question before a generation of children had already served as the experiment. Nobody asked. Not because they couldn’t. Because nobody’s job depended on it, nobody had the authority to demand the data, and the people who did have the data learned that revealing what they found was more dangerous to them than not looking.

The Response Gap

Even when problems do surface, the response time is measured in decades.

The systemic overprescription of opioids was visible in the data by the early 2000s. Meaningful federal action didn’t arrive until 2018. By then, half a million Americans were dead. The internal research on social media-driven psychological harm existed by 2017. Congress is still holding hearings. The pattern is always the same—a problem emerges, a decade of committee testimony follows, legislation eventually passes, calibrated to reliably solve the crisis from twenty years ago, instead of the one developing next. And just like with private credit, by the time the rules arrive, the industry has already migrated to whatever the rules don’t cover.

The lag isn’t a failure of will. It’s structural. Regulatory agencies are organized around industries that existed before those agencies were created. The SEC watches securities. The FDA watches drugs. The FCC watches communications. When something new doesn’t fit into any existing bucket—and the interesting things never do—it lands in a jurisdictional gap. Nobody has clear authority. Nobody has allocated budget. Nobody’s career depends on figuring it out. Prediction markets sit somewhere between gambling, securities, and opinion polling, which means several agencies could plausibly claim jurisdiction and none of them clearly do.

The result is an institutional system built to perfectly fight the last war, every time, a decade after it’s been lost.

The Actual Opportunity

AI lands in the middle of all of this, and the familiar arguments have already started. Regulate it now, before we know what it does. Leave it alone until we do. Both positions are exactly how every previous technology was handled, and there’s no reason to expect different results from the same script.

But AI has one structural property that none of the previous technologies did, and it’s worth taking seriously rather than waving at. A chemical plant can’t simulate its own explosion before it happens. A social network can’t predict what it’ll do to teenagers before the teenagers sign up. The risk assessment for every previous technology had to happen after deployment, performed by humans, slowly, with incomplete data. That’s why the lag exists. Not because regulators are lazy—because the problem can’t physically be understood until after the damage starts.

AI systems can, at least in principle, model their own failure modes before deployment. They can stress-test scenarios at a speed and scale that no congressional committee or regulatory agency could match. In fact, they could do adversarial testing against each other, which would align the incentives toward revealing findings that much more quickly, all while racing to repair similar flaws in their own products. That doesn’t mean they will. But it’s a genuine structural difference—the first technology in history that could help write its own safety manual.

The opportunity isn’t “regulate AI proactively” in the usual hand-wringing sense. It’s narrower and more interesting than that. The question is whether AI gives us a tool to compress the response gap—to shrink the decade between “we discovered a problem” and “we have a framework for it” down to something closer to the speed at which problems actually develop. Not writing rules for hypothetical risks. Building the institutional capacity to react in months instead of decades when the next unknown becomes known.

The Constraint

Building that capacity would require something the American regulatory system has never once demonstrated: the willingness to pay attention to problems that haven’t yet produced a political crisis. Every major regulatory body was created in response to a disaster. The SEC after the 1929 crash. The EPA after rivers literally caught fire. OSHA after enough workers died that inaction became more politically costly than action. The institutional DNA is reactive. We build the fire department after the fire.

The deregulation crowd is right that we have too many rules. They’re wrong that the answer is simply fewer. The answer is better-allocated attention—less time relitigating whether the 872^nd rocket launch needs eighteen months of environmental review, more time asking whether anyone is watching the $2 trillion lending market that exists specifically because no one is.

That reallocation is harder than either side’s bumper sticker. Cutting rules is popular, at least when they’re rules you’d like to break. Building new capacity for risks that haven’t blown up yet is not—the beneficiaries are invisible, the costs are immediate, and no politician ever won an election by preventing a crisis nobody noticed was averted. Every incentive in the system points toward waiting for the disaster, then acting shocked.

We’ve run this experiment before. We know how it ends. The question is whether the tools exist, for the first time, to change the ending—and whether the institutions that could use those tools will be built before the next obvious-in-retrospect catastrophe that everyone saw coming and nobody prevented.

The regulations we have too many of are a nuisance. The regulations we don’t have are a risk. One of those problems is expensive. The other is dangerous.

They’re not alone—our greatest futurists and science fiction writers get it wrong too.

There are really only two models for what a post-scarcity future looks like in the popular imagination. The Star Trek model: energy is free, replicators produce whatever you need, and people spend their time pursuing knowledge, art, and self-actualization because material want has been eliminated. Or, the Star Wars1 model: technology is so advanced it might as well be magic, yet people are still poor, still struggling, still trading in black markets on desert planets. Neither of these models is right, but one’s closer. It’s not the one we want.

Subscribe now

People feel pinched today, but the average American lives better than a Roman emperor ever did. We have jumbo jets and Netflix. Open-heart surgery and Amazon Prime. Central heating, GPS navigation, and antibiotics—luxuries that no amount of gold could have purchased even two centuries ago. By any absolute measure, we are already living in a post-scarcity world for most of the goods that defined scarcity for the previous ten thousand years of civilization.

And yet nobody feels post-scarce. That disconnect isn’t a failure of gratitude or a limitation of science. It’s the tell.

The standard economic framework has the causality backwards. We’re taught that scarcity is an external condition—a feature of the physical world—and that value emerges from it. Things are valuable because they’re scarce. But that’s not quite what’s happening.

Cancer cells are scarce, but I bet you don’t want any. Scarcity isn’t a condition that produces value. It’s the other way around. Value produces scarcity.

Wherever people want more control over their lives than they currently have—which is everywhere, always—we value whatever would give it to us. More money, sure. More power, more influence, more attention, more acclaim. What we crave is inherently “scarce,” whatever form it takes, because if it were plentiful we wouldn’t want it. When was the last time you thought about how much you want air to breathe?

But the second we get what we want, we don’t suddenly stop wanting. We just notice the next constraint down the stack. The word for this process is “progress.” The thing people keep expecting to eliminate scarcity is precisely what drives it.

The End of History

The history of human progress is, at bottom, a history of solving one constraint only to immediately discover the next. Each solution doesn’t eliminate scarcity—it reveals a deeper one that was always there, masked by the more urgent problem above it.

For most of human history, the binding constraint was calories. You spent your waking hours trying to acquire enough energy to not die. Farming solved that, more or less, which immediately surfaced the next constraint: land. When calories come from soil, arable territory becomes the thing you’ll kill for—and people did, enthusiastically, for the next several thousand years.

Industrialization broke the land constraint. Suddenly, value creation decoupled from acreage. A factory on a half-acre could out-produce a thousand-acre farm in economic terms. But this merely revealed that the true bottleneck had shifted to labor—specifically, higher-productivity labor capable of operating increasingly complex systems. The scramble for human capital replaced the scramble for territory, and the nation-states that industrialized first didn’t just get richer. They conquered the ones that hadn’t. More killing ensued.

Then came the information age, which ostensibly solved the labor constraint by making knowledge infinitely reproducible at zero marginal cost. Wikipedia alone would have been worth more than the Library of Alexandria and every medieval university combined, measured by information density. We now distribute the sum of human knowledge to anyone with a phone, for free. Problem solved.

Except it wasn’t. Because the moment information became abundant, the bottleneck shifted again—to attention. It turns out that an infinite supply of knowledge is worthless without an infinite mind capable of processing it. You can lead a civilization to Wikipedia, but you can’t make it think. The scarcest resource in the information age isn’t information; it’s the twenty minutes of uninterrupted focus required to do something useful with it.

Calories, land, labor, information, attention—each time we solved the top constraint, we merely established a new “top”.

The Next Abundance

Which brings us to the breathless promises of our current moment. AI, robotics, and fusion energy represent the most potent bundle of scarcity-killing technologies since agriculture itself.

The implications are real and genuinely transformative. Vertical farming fed by fusion power could make food production independent of arable land and weather patterns entirely. Desalination at scale—powered by energy so cheap it’s not worth the cost of the meter—eliminates water scarcity for every coastal civilization on Earth, which is most of them. Robotic construction could crash the cost of homebuilding the way industrialization crashed the cost of textiles. AI-driven drug discovery and diagnostics could make personalized medicine as routine as the antibiotics we already take for granted. Climate management becomes tractable when you have the energy budget to actually do something about atmospheric carbon other than just measuring it.

This isn’t science fiction. The component technologies exist. The engineering challenges are formidable but finite. Within a generation, possibly two, the material constraints that currently define middle-class anxiety—housing costs, healthcare costs, food costs, energy costs—could plausibly fall by an order of magnitude.

And it won’t matter. Not in the way people think.

The Scarcity Stack

Here’s where the techno-optimists lose the plot. Grant them everything—every breakthrough, every timeline, every breathless promise fulfilled. It still doesn’t get us where they think it does.

Maslow’s hierarchy is useful here, not because it’s a perfect model of human motivation, but because it reveals how far we’d have to go before material abundance even touches the constraints people actually care about. The base of the pyramid is physiological: air, water, food, clothing, shelter, warmth, sleep. AI and fusion and robotics could plausibly solve all of that, for everyone, within a generation or two. Genuinely miraculous. And entirely insufficient—because all you’ve done is clear the ground floor and force everyone to notice that the building has more stories.

One level up from physiology are safety and security—health, employment, property, personal stability. Even with unlimited cheap energy and robotic construction, a man whose industry just got automated isn’t comforted by having nothing useful to do. Material plenty doesn’t solve the constraints that actually keep people up at night. And that’s only the second floor. Health sits on this level, and its logical endpoint is mortality itself—you’d have to solve death just to finish the bottom half of the “needs” ladder. Everything that actually constitutes human flourishing lives above that line.

Besides which, there are plenty of people who can already buy all of the energy they want. People who can eat what they want, access the best healthcare, fly private. Why aren’t they living the post-scarcity utopia already? They’re not self-actualizing on the holodeck. They’re competing for status, chasing meaning, and getting divorced.

Elon Musk is the richest man on earth, and all that means is that now he wants Mars.

Because scarcity isn’t a quantity. It’s a value judgment. It’s the permanent condition of wanting more from life than you currently have—which is just another way of describing what it feels like to be alive.

The Structure of Desire

This is why the Star Wars model is closer. Not because its economics make sense—they don’t.2

But it captures what post-material-scarcity actually feels like: miraculous technology, and people still struggling. Star Trek assumed that solving material want would solve wanting.

The trust-fund kid with a cocaine habit and no direction isn’t suffering from a lack of resources. He’s suffering from a lack of constraints. Remove the relationship between effort and outcome and you don’t liberate desire—you make it aimless. He has everything except something to work toward.

AI and fusion and robotics may very well solve the next level of the material layer—and they should. But it won’t deliver the transformation that post-scarcity evangelists promise, any more than indoor plumbing delivered spiritual enlightenment. What it will deliver is a shift in what’s competed over. You can build more houses. You can’t build more Malibu. You can manufacture cheaper drugs. You can’t manufacture trust in the pharmaceutical industry. You can generate infinite content. You can’t generate the wisdom to decide what’s worth paying attention to.

The shape of scarcity in 2075 will be as unrecognizable to us as ours would be to a subsistence farmer in 1300. And whatever our grandchildren face, they’ll be convinced it’s uniquely urgent, uniquely unfair, and surely solvable with the right technology.

They’ll be wrong about the last part. But they’ll solve it anyway. And discover the next constraint just underneath.

Now consider your neighbor. Same $150,000 in income. But they also borrow $40,000 a year—home equity lines, credit cards, an underwater auto loan rolled into the next one. They spend $190,000. By every government measure of economic activity, this household is more productive. More transactions. More consumption. More GDP, if households had a GDP.

They’re also broke as shit. And getting broker.

Subscribe now

You already know which household is actually prospering. You don’t need a degree in economics to see it. You need a kitchen table to update your balance sheet at.

Nations don’t use balance sheets. They use GDP.

What GDP Was Built For

GDP deserves more credit than its critics give it. When Simon Kuznets developed national income accounting in the 1930s, he was solving an urgent, specific problem: how much economic capacity can the United States mobilize right now? With the Depression grinding on and war looming, that was exactly the right question. FDR needed to know how many tanks, planes, and uniforms the economy could produce. Kuznets delivered a tool to answer it.

He also warned—explicitly, in his 1934 report to Congress—that national income figures should not be treated as a measure of welfare. The tool measured throughput: how much activity is the economy generating this quarter, this year? It was a flow metric, designed to capture immediate capacity. And for its intended purpose, it worked brilliantly. We saved the world.

The problem is what happened next. A wartime throughput gauge became, by institutional inertia and political convenience, the default measure of national prosperity. GDP growth became synonymous with progress. Quarters of positive GDP meant the economy was “strong.” Negative quarters meant crisis. An entire vocabulary of economic health was built on top of a metric that measures activity the way a tachometer measures RPMs—accurately, but with no indication of whether the car is actually going anywhere.

Name another investment that’s compounded for seventy years with as little to show for it. American GDP has grown nearly every year since the 1950s. Real, inflation-adjusted growth. Exposed to the magic of compounding across seven decades. And yet: real wages for the median worker have barely budged since the early 1970s. Homeownership rates for young adults are at historic lows. Household savings rates have collapsed. The national debt has exploded from concerning to existential. Personal debt has followed the same trajectory.

Seventy years of uninterrupted “growth.” And the median household is running harder to stay in the same place.

Why Nobody Fixed It

The obvious question—if GDP is the wrong metric, why hasn’t it been replaced?—has a boring answer, which is usually a sign it’s the right one.

Path dependence. International comparability. Institutional inertia. Treaty obligations pegged to GDP thresholds. The IMF. Careers built on interpreting GDP data. Government agencies structured around reporting it. Entire academic disciplines organized around modeling it.

But ultimately, a metric that rewards government spending will always outcompete a metric that rewards fiscal restraint.

Which is why nobody owns the whole picture. The Bureau of Economic Analysis measures output. The Bureau of Labor Statistics tracks employment and prices. The Federal Reserve monitors financial conditions. Treasury manages debt. Each agency does its job competently within its lane. None is tasked with assembling the pieces into a coherent answer to the question that actually matters: after all this activity, did we end up with more than we started with?1

The economy, meanwhile, shifted underneath the measurement system without anyone updating the dashboard. In the 1950s, economic output was overwhelmingly physical: steel, automobiles, housing starts, bushels of grain. Measuring throughput in that world captured something real about productive capacity, because physical goods are harder to fake and harder to finance with leverage than financial abstractions.

Over the following decades, the economy transformed in ways that made throughput an increasingly poor proxy for progress. Manufacturing gave way to services. Savings gave way to leverage. Ownership gave way to financialization—a world where the same underlying asset could generate dozens of transactions, each one counted as “output,” without a single new thing being produced. A house built in 1985 gets sold, refinanced, securitized, tranched, insured, derivatized, and traded—every one of those transactions registers as economic activity. Meanwhile, the roof starts to leak.

The nature of “investment” itself has shifted. In 1960, when a corporation spent a billion dollars, it was probably building a factory. The output was visible, durable, and productive for decades. Today, a billion dollars of corporate spending might mean stock buybacks, financial engineering, or acquiring a competitor to eliminate competition rather than create capacity. All recorded identically in the national accounts. All very different in what they leave behind.

The economic measurements didn’t adjust for any of this.

It’s not a conspiracy. It’s not even really incompetence. It’s just the way institutions work. They accrete. They don’t refactor. Nobody wakes up in the morning and decides to perpetuate a flawed metric. They just cite the number everyone else is citing, they fill in the cell in the spreadsheet it’s their job to fill in, because building a better one isn’t anybody’s job.

The Accounting Asymmetry

Of course, accretion doesn’t work in a competitive setting. Which is why the private sector solved this decades ago.

Any publicly traded company is required, by law, to distinguish between revenue and profit. Between operating expenses and capital investment. Between assets and liabilities. Between cash flow from operations and cash flow from financing. A company that reported only its top-line revenue and called it “growth” would be laughed out of every analyst meeting on Wall Street and delisted before lunch.

We don’t just require companies to report income. We require them to report what they did with it. Did they reinvest in productive capacity? Did they maintain existing assets? Did they return capital to shareholders? Did they borrow to fund operations? These distinctions matter, because revenue without context is meaningless. A company that grows revenue 20% by borrowing 30% more is not growing. It’s dying, with panache.

Governments face no equivalent reporting discipline. GDP counts the spending. It does not count the bill. It records the activity without distinguishing between a dollar spent building a bridge, a dollar spent replacing a bridge that fell down, a dollar spent on interest payments for the debt that financed the original bridge, and a dollar that simply moved from one government account to another. All four are GDP. None are equivalent.

The entity managing the most money on Earth operates with less financial transparency than a regional car dealership. The dealership has to file audited financials. The government files a press release.

First Principles

Strip away the institutional history and the political noise, and the accounting problem is elementary.

Every dollar of economic output does one of three things. It replaces something that wore out—depreciation, maintenance, the cost of keeping the lights on. It builds something new—productive capacity that didn’t exist before. Or it gets consumed—eaten, burned, enjoyed, and gone.

GDP treats all three equally. A dollar is a dollar is a dollar. But a dollar spent patching a pothole is not the same as a dollar spent building a road is not the same as a dollar spent on cigarettes for the work break. One keeps you where you are. One takes you somewhere new. One goes up in smoke.

This is where debt enters the picture, and where the measurement failure becomes actively dangerous.

Debt is a timing mechanism. It lets you pull future consumption into the present or push present investment into the future. Used well, it accelerates the conversion of savings into productive assets—borrow to build a factory, generate returns, pay back the loan. At the end, net savings are higher than before, and you have a new factory to boot.

But used poorly, debt lets consumption impersonate investment. You can borrow a trillion dollars, spend it entirely on digging holes and filling them back in, and GDP will faithfully record a trillion dollars of “growth.”

The activity was real. The output was not.

This is why a society can grow GDP for seventy consecutive years and still end up with crumbling infrastructure, insolvent entitlement programs, and a median household that can’t afford a home, an education, or a retirement. GDP recorded every dollar of spending. It never asked whether the spending left anything behind.

The Question We Never Ask

The data needed to answer the real question already exists. We track gross domestic product. We track capital depreciation. We track net investment. We track government debt issuance. We track household leverage. We track infrastructure quality indices. Every input is measured, somewhere, by someone.

We simply never assemble them.

The question is: After paying for upkeep, after servicing the debt used to finance it, did this year’s economic activity leave us with more productive capacity than last year’s?

That’s it. That’s economic growth. Not “how much did we spend?” Not “how fast did activity grow?” But: did we actually build anything, net of what we consumed and the financing cost?

Call it whatever you want. Net Productive Growth. Adjusted Capital Formation. The National P&L. The label doesn’t matter. What matters is the discipline of asking, at the end of every fiscal year, whether the nation’s balance sheet—its productive assets minus its obligations—got stronger or weaker. The way every lemonade stand with a cigar box full of quarters already does.

The Back of the Envelope

Even a crude attempt at this calculation is revealing.

In 2025, US GDP was approximately $31 trillion. Impressive, until you start subtracting. Capital depreciation—the cost of replacing worn-out equipment, infrastructure, and structures—consumed a little over $5 trillion. Net interest on government debt ate another $1 trillion. Household, corporate, and municipal debt service absorbed about $1.5 trillion. State and local maintenance backlogs—the deferred repairs on roads, bridges, water systems—is over $9 trillion in total, so maybe $700 billion in annualized terms.

That’s $8.2 trillion before anyone builds anything new. Call it $8 trillion, rounding graciously.

Of the remaining $23 trillion, how much represents genuine net investment in productive capacity—new factories, new infrastructure, new technology, new human capital? Gross private domestic investment was roughly $5 trillion, but much of that is replacement investment counted as “gross.” Net private investment—the portion that actually expands capacity—was closer to $2 trillion. Government investment, net of maintenance spending, adds perhaps another $800 billion on a generous reading.

So: approximately $2.8 trillion in net new productive capacity, from $31 trillion in activity. Under ten percent. The economy runs $31 trillion through the meter and less than a dime on the dollar comes out the other side as something new that will still exist next year.

And that’s before asking how much of even that $2.8 trillion was financed by adding to the $40 trillion national debt—whether the “investment” was funded by genuine savings or by pulling forward future claims.

The federal government added approximately $2.3 trillion in new debt in 2025. If even half of that funded consumption rather than investment, the net productive growth number shrinks further—potentially to low single digits as a percentage of GDP, possibly to zero.

Seventy years of “growth.” Nine cents on the dollar actually building something. And most of that was borrowed.

This is a napkin calculation. The real number could be a little better or worse. But the order of magnitude is what matters, and it explains something that GDP alone cannot: why all this activity doesn’t feel like progress. It doesn’t feel like progress because most of it isn’t. It’s maintenance, consumption, debt service, and the statistical echo of money changing hands.

Why It Feels Like This

You’ve felt this, even if you’ve never run the numbers.

Every time a headline announces “strong GDP growth” and you look around and think, strong for whom?—this is why. The metric is reporting motion. You’re experiencing a lack of direction.

When your wages rise 3% but your rent rises 5% and your retirement account buys a smaller share of the economy every year despite faithfully contributing—this is why. The flow is positive. The stock is not. You’re on a treadmill that reports your heart rate but not your distance.

When politicians from both parties claim the economy is either booming or on the verge of collapse and neither description matches your lived reality—this is why. They’re reading a tachometer and telling you about speed. One side says the engine is running hot and everything’s great. The other says it’s about to blow. Neither bothers to check whether the car has moved.

Asset prices are at all-time highs, and so is financial anxiety. Both things are true simultaneously, and they’re not contradictory—they’re complementary. Asset prices rise because money is chasing stores of value. Financial anxiety rises because wages can’t keep pace with the assets people need to acquire in order to achieve basic security. The economy is generating enormous activity. The activity isn’t generating prosperity.

This isn’t a populist complaint about elites cooking the books. The books aren’t cooked. They’re just reliably measuring precisely the wrong thing. GDP is an honest answer to a question nobody should be asking outside of a war-planning room.

The dissonance between reported economic performance and felt economic reality isn’t irrational. It isn’t ingratitude. It isn’t economic illiteracy. It’s the mathematically predictable result of chasing revenue at the expense of profitability. A society that measures activity instead of accumulation will always feel like it’s running in place—it is.

What Would Change

If we measured what matters—net productive growth rather than gross economic activity—several things would shift, and none of them are ideological.

Debt-financed consumption would show up as what it is: a drawdown on future capacity, not “stimulus.” A trillion-dollar spending bill that builds high-speed rail would look very different from a trillion-dollar spending bill that funds temporary transfer payments—even though GDP treats them identically today. Government spending that creates lasting infrastructure would be visibly distinguished from spending that simply moves money between accounts and calls it growth.

The perverse incentive to maximize throughput at the expense of durability—building cheap roads that need replacement in ten years instead of good roads that last forty—would become visible and, presumably, embarrassing. Under current measurement, the cheap road is better for GDP: you get to count the construction spending twice in twenty years instead of once in forty. A net productive growth metric would expose this as the accounting absurdity it is.

Most importantly, the political conversation would change. “GDP grew 1% this quarter” would become an input, not a conclusion. The follow-up questions—how much of that was maintenance? how much was borrowed? how much actually expanded our productive base?—would be inescapable once the framework existed to ask them.

We wouldn’t need to agree on what to do about the answers. We’d just need to start asking the questions. A nation that discovers it’s spending $31 trillion a year to generate $2.8 trillion in bottom line progress will have a very different policy debate than one that simply celebrates the headline.

The Point of Growth

The point of growth was never the growth itself.

Nobody wakes up in the morning wanting to maximize GDP. They wake up wanting to build something—a business, a home, a family, a life that’s a little more secure than the one they were born into. They want their kids to start farther ahead than they did. To inherit not just money but infrastructure, institutions, and productive capacity that compound across generations. To stand on the shoulders of giants, and to be those giants for whoever comes next.

That’s what an economy is for. Not activity. Not throughput. Not the frenetic circulation of dollars through a system that tallies transactions without asking what any of it built. An economy exists to convert human effort into durable prosperity—the kind that survives the people who created it.

When all of the numbers go up and to the right, but it now takes two full-time incomes to secure the lifestyle that our grandparents earned with one, the problem isn’t the lifestyle. It’s the numbers. You improve what you measure, and we’re not measuring prosperity.

Kids used to put themselves through college with summer jobs. Now they’re paying off student loans in their golden years. We once led the world in production, it’s how we won the Great War. Now we’re importing Temu plastic from our adversaries. A generation ago, a thirty-year-old could buy a house. Today they’re buying Labubus on installment plans.

This isn’t nostalgia for a time gone by. It’s an audit. For all of our effort, for all of the progress, for all the trillions flowing through the meter, are we better off?

A nation that can’t answer that question about itself isn’t being governed. It’s being managed—quarter to quarter, headline to headline, with no one keeping score on the only results that matter across generations.

We measure GDP, but we don’t keep it. It slips through our fingers, year after year. Real prosperity is built from what remains after our labor. And it’s the whole point.

The skepticism shows up in the data, too. Executives are bullish—AI is transformative, a paradigm shift, a major investment priority. Meanwhile, most employees report minimal impact on their actual work. This disconnect isn’t just corporate myopia. It reveals something important about what AI actually does well versus what we expect it should do well.

Subscribe now

What AI Actually Does

LLMs excel at executive processing tasks: deciding which approach to try, switching between contexts, weighing multiple solution paths simultaneously. These are activities that scale with scope, not depth—the work that benefits from being able to consider many possibilities simultaneously rather than mastering one thing through repeated iteration.

What LLMs don’t do is learn from experience and refine through practice. They can’t discover a lesson from iteration five and apply it systematically to iterations six, and onward. They don’t build expertise through repetition. They aren’t a “they” at all. Every response is essentially independent, generated from the same static training distribution with, at most, a modicum of extra context within a single thread.

The only way for them to get better at a specific task is for the commercial operator to load it into the data in advance. And there’s one big domain where they’re doing just that: writing code.

AI is good at writing software. Very good, and getting much, much better. Three years ago, GitHub Copilot was a glorified autocomplete. You’d accept maybe a third of its suggestions, correct another third, and reject the rest as useless nonsense. Functional, occasionally impressive, but wrong frequently enough to establish a baseline of reasonable skepticism in many developer circles.

Now? With sophisticated workflows, an experienced AI engineer can implement more in a day than they could in a week two years ago. Not incrementally better—categorically different productivity. Multiple agents running in parallel, one reviewing the code another wrote, a third refactoring for performance, a fourth auditing security, a fifth writing tests. Developers are orchestrating these systems the way an editor manages a publication, steering the direction rather than typing every character themselves.

Why does coding work so well, while other tasks don’t? Fundamentally, it’s because code has a tight feedback loop that doesn’t require iterative learning, just contextual decision making. The software either runs or it doesn’t. The tests pass or fail. The code compiles, or the build breaks. You don’t need to master any new lessons from previous attempts—you need to generate syntactically valid, logically correct text that satisfies formal constraints. That’s pure pattern matching at massive scale, which is exactly what LLMs are built for.

But that massive scale comes with a price tag.

The Infrastructure Toll

Here’s where it gets expensive.

To achieve that 5x productivity gain, which elite practitioners are already surpassing, you need to consume a lot of LLM usage. This isn’t occasional queries to answer a question. It’s dozens of agents running in parallel for hours, processing entire codebases, generating thousands of lines of suggestions, grinding through complex implementations.

Current usage limits make it impossible to maintain peak productivity for a full forty-hour week, even with top-tier subscriptions, so developers overflow into extra usage fees. Claude Code’s Max subscription runs $200 monthly, plus overages when rate limits are hit. And that’s just one tool. Sophisticated developers are using Cursor, Claude, GitHub Copilot, Codex, Grok, and specialized models for different tasks. Prosumer usage easily reaches $300-500 monthly, and often significantly more for the most intensive workflows.

For an engineer earning $200,000 annually, it’s worth it. To the company paying their salary, it’s an obvious investment—why wouldn’t you spend $500 monthly to get 5x output from a $200k employee?

But for the past decade, we’ve sold ‘learn to code’ as the great equalizer—the one skill that could lift anyone into the middle class regardless of background or credentials. And now we’re breaking our promise.

Learning that skill is no longer enough. Now it’s learn to code, plus pay $300-500+ monthly just to stay competitive, or get destroyed by peers who can.

The Democratization Inversion

The most excited, accelerationist AI advocates claim that soon anyone will be able to build software. They’re half right at best. The engineering skill requirement might become optional—it hasn’t yet. But the promised democratization isn’t coming.

The old model was straightforward: high barrier to entry, low barrier to compete. Learning to code was hard—really hard—but once you learned it, you learned it. The knowledge was yours. You could practice for free, build for free, compete for free. A kid in Bolivia with a $300 laptop and a spotty 3g connection could learn Python from open-source tutorials, build a portfolio on a free GitHub account, and compete for remote contracts on genuinely equal technical footing with developers in San Francisco. Zero marginal cost to deploy your skill. No monthly fees. No recurring expenses. Just you, your knowledge, and what you could build.

The new model inverts this completely: low barrier to entry, high barrier to compete. Even if they’re right, that anyone can build software with AI, no training required, they’re missing the point. For anyone trying to build a career or business in software, AI hasn’t removed the need for coding skills. It’s added an LLM tollbooth.

The person who “doesn’t need to code” because AI does it for them isn’t competitive with the developer who is orchestrating half a dozen AI agents in parallel. The gap isn’t small. It’s 5:1 productivity, minimum, and widening. We’ve traded a one-time knowledge investment for a permanent subscription tax, priced in Silicon Valley margins.

The Compounding Problem

This isn’t like previous technological transitions. When CAD software emerged, it became a new requirement—but you could meet it. The productivity gain plateaued once you got it. Learn SolidWorks, get a license: you’re competitive. The tool didn’t get dramatically better every six months.

LLMs are different. The tools improve with each release. The workflows evolve as developers discover that agents can review other agents, that sophisticated orchestration multiplies productivity beyond what any single tool provides. Developers keep discovering new prompting tricks and contextual techniques that improve accuracy and performance. The skill gap isn’t just can you use AI, but rather: how sophisticated is your AI workflow, and how much can you scale it with API credits, subscription fees, and the latest release version?

As each new model is trained with more data, on more GPUs, with more power, the costs keep rising. Commoditization might eventually bring prices down. They haven’t yet. This isn’t a race, it’s a treadmill.

The developer who can’t afford the tools isn’t just slower. They’re unemployed. Once management sees what’s possible with AI augmentation, that becomes the new baseline. Everyone else is irrelevant. Software development has always been competitive—a marketplace of skills. But the competition used to be who can solve this problem better. Now it’s who can afford to solve this problem at competitive speed.

The Bolivian Problem

In 2015, an aspiring programmer in La Paz could learn for free, build on outdated, modest equipment, and compete on equal technical footing. Their lower cost of living meant they could even undercut US developers on price while maintaining great margins. Access to a $300 laptop wasn’t trivial—it represented a real barrier to entry—but it was the only one. Solve it and you’ve got a real shot.

Now, they still need the laptop, plus reliable internet for their entire working day, which remains spotty or expensive in much of the world.

They need $200 monthly in subscriptions, minimum—$2,400 annually, eight times the cost of that one-time $300 laptop. And they need it every year. Top end usage is equivalent to two of those barrier-to-entry-laptops, a month.

That subscription cost might be 30-50% of gross income in many developing economies. Every month, forever, or you fall behind. It only looks democratic—anyone can talk to AI. But the barrier to performance has skyrocketed. “Talking to AI” isn’t the same as your competitors orchestrating multiple, parallelized AI agents, while you’re asking ChatGPT to debug your for-loop.

The skills are democratized. Free learning resources exist everywhere, better than ever. But the infrastructure is privatized. Expensive, recurring subscriptions that represent trivial costs for established professionals and prohibitive barriers for everyone else.

The Pattern We Keep Missing

This is the same playbook we’ve seen before, dressed up in the latest fashion. We told everyone they should own a home, then turned homes into tax vehicles and watched prices decouple from wages. We told everyone they should go to college, made degrees all-but-mandatory, and watched costs explode while relative value cratered. Now we’re telling everyone they should learn to code while simultaneously making it impossible to compete without paying monthly rent to AI megacorps.

The democratization rhetoric always obscures the infrastructure capture. We confuse access to training with access to opportunity. Anyone can learn Python for free on YouTube. But can they compete without the tools that multiply productivity 5x? The barrier isn’t knowledge anymore—it’s subscription fees.

What makes this particularly insidious is that in previous cases, you could at least finish. Get the degree, buy the house, learn the language. You paid the price and you were done. With AI subscriptions, you never finish. The treadmill never stops. Your competition will use the newest model, whatever the cost. Which means you will too, or you’ve already lost.

The Uncomfortable Economics

The structure is almost elegant. Free skill acquisition plus expensive skill deployment equals perfect conditions for wealth extraction. They’ve figured out how to let you build human capital for free, then charge rent on the ability to use it competitively. You can spend a thousand hours learning Python and JavaScript without spending a dollar. But to deploy those skills at professional levels? That’ll be hundreds of dollars a month in service fees, for the rest of your career.

This isn’t to say AI is bad. And it certainly won’t be an economic, or even employment, apocalypse. That’s always the prediction. And, predictably, it’s always wrong. Tractors didn’t end employment, they ended hunger. The old family farms got gobbled up, but the former farmers got cars and factory jobs. If people don’t have income, they won’t pay the AI companies, or buy the software that AI engineers create. The transition will be messy. Real people will struggle. But history rhymes, and it sounds like progress.

The risk isn’t whether AI will destroy everyone’s livelihoods. Some people are already 5x more productive with it. That productivity shows up somewhere—in higher pay for AI users, in corporate profits, in cheaper software, in all of the above. Someone benefits. The only question is who.

But what we should be skeptical of is the democratization. Because what we’re actually getting is stratification, between those who can pay the toll versus those who can’t. Not based on skill. Not based on merit. Based on whether you can afford the monthly fee to make your skills competitive in the marketplace. The end of the artisan engineer, and the dawn of factory farm coding. That’s what the real AI economy looks like.

Turns out, lots of places. And they’re going there. We have planes now.

Subscribe now

The fundamental miscalculation is a simple one that should be obvious to anyone with a laptop. If you can work from a couch in California, you can work from one in Brazil, just as easily. And this is a problem for anyone who’s still planning on “taxing the rich” in 2026, using a 1936 tax regime.

In 1936, when marginal rates hit 79%, you could actually enforce that—not because people loved paying taxes, but because wealth was physical and escape was hard. Your factory was bolted to the ground in Detroit. Your farmland wasn’t relocating to Panama. Moving gold across the Atlantic meant booking passage on a ship, a four or five day journey at best. Languages fragmented the world into distinct economic zones with very little mobility. The wealthy were, practically speaking, stuck.

Today’s wealth is a different species entirely. Equity portfolios transfer across borders in milliseconds. Intellectual property domiciles wherever the paperwork says it does. Cryptocurrency exists in a jurisdiction called “everywhere and nowhere.” A software consultant in Bali can serve the same clients as one in Boston—the work is identical, but the tax bill is not.

The kicker? Legally reducing taxes isn’t even particularly hard to execute. You don’t need to be a billionaire with Cayman Islands shell companies and Swiss banking secrecy. Senior software developers, social media consultants, and freelance designers are quickly learning that the same work product, for the same clients, can legally generate an 85% reduction in tax liability with nothing more exotic than a plane ticket and an afternoon of paperwork.

The American Disadvantage

Let’s start with the hardest case: American citizens. The US is one of only two countries on Earth that taxes based on citizenship rather than residence. Eritrea is the other, which should tell you something. This means Americans owe US taxes on their worldwide income regardless of where they live or work. Sounds airtight. It’s not.

The mechanics of American tax reduction are straightforward. Establish bona fide residence in a foreign country—meaning you live there, not just visit—and the Foreign Earned Income Exclusion shelters your first $130,0001. That’s not a deduction or a credit; it’s excluded entirely from taxable income.

But wait, you also get a housing allowance. Live somewhere expensive enough and that could be another $40,000 excluded. You haven’t done anything sophisticated yet—you’ve just moved and filled out Form 2555.

Self employed with a high income? Tack on a solo 401(k). You can contribute up to $72,000 pre-tax, in 2026. Unless you’re over the age of 50, in which case it’s a full $80,000 pre-tax deduction. High earners with more sophisticated setups can add “profit-sharing” plans to scale the deductions even higher.

Run the math: $130,000 (FEIE) + $40,000 (housing) + $72,000 (solo 401k) = $244,000 shielded from federal income tax, legally. And while still subject to Social Security and Self Employment taxes (the employer-side Social Security and Medicare matching), any decent CPA will likely get that portion down substantially by making you an employee of yourself, paying yourself a modest salary, and taking the rest as “business profits” instead of earned income.

From a beach in Bali, a consultant earning $300,000 could now pay US federal income tax on only $56,000. At a 32-35% marginal rate, that’s roughly $18,000 in US income taxes and maybe as little as $9,500 in Social Security and Self Employment taxes. On $300,000 in income. A 9% effective rate, in a country with postcard-perfect beaches, and where a penthouse apartment is $1,500 a month.

Compare that to staying in California: you’re paying 37% federal on income over $191k, plus 9.3-13.3% California state tax, plus payroll taxes, Self Employment taxes, sales taxes, and property taxes. Your effective rate on $300k is pushing 50%, or more. That’s half of your money, right off the top, for the privilege of sitting in Los Angeles traffic while burning gas at $7 a gallon.

For Americans, the system is designed to be inescapable. But inescapable is a spectrum, and the spectrum runs from 50% to 5%. That’s not a rounding error. That’s a different life.

The Everyone-Else Advantage

If you’re not American, it gets really absurd.

Most countries tax based on residence, not citizenship. Establish residence nowhere, pay taxes nowhere. This is not a loophole—it’s the explicit design of territorial tax systems encountering the reality of digital work.

The perpetual traveler strategy is beautifully simple: never stay in any country long enough to trigger tax residency. Most jurisdictions use 183 days as the threshold. Stay 180 days in Thailand, 120 days in Mexico, 65 days bouncing around the EU. You’re a tax resident of nowhere. You owe income tax to no one. Want to be really safe? Get a legal residency in an ultra-low tax jurisdiction and establish official tax residency there.

“But surely that’s illegal!” No, it’s just... how income taxes work. Countries designed their tax systems around the assumption that people live somewhere. Digital nomads do live somewhere—they live everywhere—they just don’t live there long. The systems never contemplated someone whose permanent residence fits in a laptop bag.

Your Estonian software company—which took 3 hours and €200 to incorporate online through e-Residency—bills your American clients. The revenue sits in a Wise business account, accessible from anywhere, convertible to any currency. Or register a C-Corp in Panama, where you only pay taxes on work actually conducted in Panama. So you just don’t go there. Panamanian banks are just as connected to the global banking system as any other bank.

This isn’t theory. Digital nomad communities have turned this into paint-by-numbers. Reddit forums are full of ordinary people comparing notes on visa runs, optimal country rotations, and which coworking spaces have the fastest internet. They’re not criminals. They’re responding rationally to a system that still thinks “where you work” and “where you live” are redundant questions.

The Corporate Shell Game

For those with more substantial operations, the optimization becomes trivial. Incorporate in Singapore (17% corporate tax, territorial system, global business hub). Or Ireland (12.5% corporate tax rate). Or the UAE (0% corporate tax for most activities). Or Estonia (0% on retained earnings). Or, ironically for non-Americans, Wyoming (no corporate income tax, no annual report requirements, full anonymity).

These aren’t exotic frontier zones. They’re legitimate jurisdictions with robust legal systems, actively competing for your business registration. And why wouldn’t they? Corporation registration fees, local employment, registered agent services, office leases—these generate revenue and boost local economies without requiring the corporation to pay tax on global operations or inflating the costs of local housing stock.

Your US clients pay your Irish company. Your developers work remotely from Portugal, Argentina, and the Philippines. Your servers are in AWS data centers spread across continents. The company pays Irish corporate tax on Irish-sourced income—which is minimal because the value creation happens elsewhere. The actual profits? They sit in the corporate account, undistributed, growing or reinvested back into the business. Eventually you move to somewhere low tax, establish residence, and take distributions for the rest of your life.

Every step is legal. Every step is well-documented on government websites. Every step is exactly what the respective jurisdictions intended. It’s just that no single jurisdiction designed their rules expecting them to be chained together quite this efficiently.

The Enforcement Asymmetry

The standard objection: “Can’t governments just crack down on this?”

On what, exactly? Following the law? The problem isn’t rule-breaking—it’s that the rules were written for a world where people and capital couldn’t move freely, and now they can.

Enforcement scales nonlinearly with global mobility. When wealth was a factory, you just walked in and counted the machines. When wealth is a portfolio of global equities, a crypto wallet, and IP rights to software, what exactly are you enforcing? Every enforcement mechanism requires international cooperation, and international cooperation requires every country to act against their own interest in attracting that tax base.

The OECD’s Base Erosion and Profit Shifting (BEPS) initiative and global minimum tax proposals are attempts at coordination. They’re failing for the obvious reason: coordination only works when defection isn’t profitable. Ireland isn’t going to torch its competitive advantage. Estonia isn’t going to close e-Residency. Dubai isn’t going to stop attracting digital businesses. They benefit from the capital flight, and capital will fly to whoever doesn’t coordinate on higher taxes. It only takes a few holdouts.

Meanwhile, the costs of implementing these tax optimization strategies are falling toward zero. Stripe Atlas will incorporate your company in Delaware for $500. E-Residency in Estonia is €200 and takes 72 hours. The Bank of Georgia will gladly accept your deposit and open your account online. International wire transfers are free or near-free through digital banks. Legal templates are available online. The infrastructure for optimization is now commoditized.

You’re asking governments to fight an asymmetric war against their own interests, where the attack surface is essentially infinite and defense costs scale exponentially.

The Marginal Taxpayer Problem

The one counterargument, that most people won’t move for tax reasons, is mostly true. And largely irrelevant.

Sure, people have family, friends, roots. But a flight costs a few hundred dollars, and taxes at these levels cost a hundred thousand. At a certain point, it’s cheaper to fly them to you. Or fly to them. Hell, fly private. You can afford it with all that extra money you have.

The point is, the problem is real even if most people don’t move. You only need the marginal taxpayer to move. And high earners are, definitionally, marginal in the distribution. The top 10% of households pay over 70% of the federal income taxes in the US. If even a fraction of them optimize internationally—and they represent the top-end distribution of that revenue—you’ve just blown a $300+ billion crater in the federal budget.

And here’s the thing about human behavior: people don’t move for a 2% difference. They move for a 20% difference. Tax rates operate as a step function in behavior, not a smooth curve. Most people absorb modest rate increases with grumbling. But when the gap between “stay and pay 50%” and “move and pay 5%” gets large enough, the calculus flips. And once it flips for your peers, the social cost of relocation drops to approximately zero. Your reference group isn’t the neighbors who stayed—it’s the network who left. When enough of your friends are drinking beers in Thailand, you might just find yourself transplanting your roots.

There’s a tipping point. And we’re approaching it faster than anyone in Washington, London, or Sacramento seems to realize.

The Incoherent System

Free people aren’t just captive revenue sources for the government. Short of a Berlin wall across Niagara Falls, and F-35s shooting down private jets, people are going to go where they want. If you want their tax revenue, you have to actually be the place where they want to go. That should be easy. For decades, millions of people have gladly lived in California and paid comparatively exorbitant state taxes to do so. They could have walked across the border to Nevada any time they wanted to. They didn’t, because California has a lot to offer.

The US has a lot to offer. As does the UK and the European Union. But quality of life isn’t a certification you get once that lasts forever. It’s something you have to deliver day after day, and year after year.

When people couldn’t easily up and leave, the pressure to innovate and compete on tax policy was modest. It’s not any more. Leaving is trivial now. The 20th-century model of high marginal rates funding expansive government assumes a captive tax base, or broad political consensus. Those assumptions are obsolete. The infrastructure for legal tax optimization is built, documented, and accessible to anyone with internet access and a plane ticket.

We can continue to insist that the wealthy should pay their “fair share.” We can pass laws raising rates to whatever feels satisfying. But the wealthy aren’t blindly accepting any one tax code anymore—they’re comparing Portugal’s and Panama’s. And the gap between what governments think they can extract and what they’ll actually collect is widening every year.

The tax competition isn’t some theoretical game we should start playing. We’re already in it. And we’re losing. Countries are competing for each other’s tax base whether they realize it or not. Some—Dubai, Estonia, Portugal—are competing intentionally. Others are watching capital flee and their tax bases shrink while chanting “tax the rich” at self-congratulation rallies.

10% of Americans pay 70% of the taxes. And they have passports. We might want to start thinking about what they want, and offering it to them. If we don’t, it’s the other 90% who will pay.

Either way, IQ plainly isn’t a universal score for human worth. It doesn’t measure creativity, artistry, charm, empathy, social grace, political instincts, discipline, wisdom, morality, or the ability to avoid lying to yourself. Entire categories of human excellence, and fields of professional achievement, depend on those traits, not pattern-matching under time pressure. Plenty of people with average IQ do extraordinary things; plenty of high-IQ people do nothing with it.

Subscribe now

But there are domains where IQ is brutally predictive. High scorers are more likely to earn advanced degrees, out-earn peers even controlling for education, and receive stronger on-the-job performance ratings from supervisors. They are also disproportionately responsible for patented invention. At the country level, Gelade (2008) reports a correlation of about r = 0.51 between national mean IQ and patents per million, rising to roughly r ≈ 0.64 when focusing on the top 5% alone. Whatever IQ is measuring, it maps tightly onto economically consequential output at a national scale.

Externalities

IQ is, in large part, a lottery. There’s a genetic component, for members of the lucky sperm and egg club, expressed like other complex traits: regression to the mean is real, siblings vary, and high-IQ parents can absolutely have average kids and vice versa. But all things considered, IQ does tend to run in the family like height or the size of your chin. Environment matters too—nutrition, toxins, illness, stress, and early development all have an impact. And sometimes it’s just bad luck: injuries, developmental anomalies, random setbacks. Nature is unevenly distributed.

By early adolescence, IQ is fairly stable. We can improve test performance through familiarity, coaching, and reduced anxiety, but the evidence for large, durable shifts in underlying capability is thin. In other words: we can’t count on “teaching IQ upward” at scale.

So the rational goal of public education isn’t to pretend everyone will end up equally capable. It’s to maximize real-world competence across the whole distribution, and to fully develop the high-end tail that produces outsized spillovers. These externalities—growth, invention, tax base, civic stability—are why society pays the bill.

But if that’s our goal, we’re doing a lousy job of it.

Our Education Priorities

America spends an enormous amount on K–12 public education—on the order of a trillion a year. In 2020–21, total US public elementary and secondary school expenditures were about $927B1. But “total spending” is the wrong headline, because the distribution is wildly uneven.

Take special education. Using district finance data from 24 states (covering ~41% of special-needs students), Bellwether found districts spent $38.8B on special education in FY20202. Scale that to the whole country and you’re quickly in the $90–100B neighborhood. And this is not discretionary: federal law (IDEA) requires schools to provide a “free, appropriate public education” to all students with disabilities, whatever the cost. In this case, “whatever the cost” is about $13,127 in dedicated spending per identified student.

Compare that to gifted education—the pipeline for the future engineers, founders, inventors, and scientists that our current social contract depends upon. There is no federal mandate to identify gifted students or serve them consistently; definitions and access vary by state. Worse, exactly zero federal agencies comprehensively track gifted-education expenditures, which itself hints at the lower priority and funding. At the federal level, the sole dedicated program is the Jacob K. Javits Gifted and Talented program—funded at just $16.5 million per year, nationwide. This minuscule amount (roughly 0.02% of the U.S. K‑12 federal education budget) allocates only $2.50–$4.00 for each gifted student in the country. Would you like fries with that?

Major Delay

When most people criticize the out-of-control costs of higher education, they’re focusing on the tuition, the student loan interest, and the depressingly low ROI between the cost of most degrees and the relative impact on career earnings. College is expensive. But the real cost is time.

The average student actually completes their four-year degree in five. That’s five years of forgone earnings. Five years of missed promotions, raises, and job experience. It’s five years of accumulated debt, instead of 401(k) contributions. And the compounding is immense.

They’re not giving up the first five years of income, they’ll still have to go through those. It’s the loss of the last five that they’re sacrificing. The highest paid, most senior, and most experienced years of a career. And since high-IQ students are most likely to obtain advanced degrees, often over as many as 8-12 years of post-secondary education, it’s also our most innovative and economically productive students whose productive years we’re maximally sacrificing at the altar of one-speed schools.

What if our most academically gifted students, the top 5-10%, were offered accelerated classes, instead. Classes whose pace was specifically tuned to their learning rate. If course material was covered at even a 25% faster rate, 1.25 grade levels per year, these students would complete their K-12 education by age 15. They’d complete an undergraduate degree in about the same amount of time as it takes for a traditional high school diploma, all with public funding. And it’s quite likely that the engagement rates would be markedly higher. Believe it or not, it’s painfully boring to sit in a public school classroom and listen to a teacher patiently repeat things that you already know to kids who aren’t interested anyway.

Most importantly, they’d enter their careers years earlier, and with far less debt. This isn’t just a personal boon for the already-privileged. These are the students who are most likely to cure cancer, send space probes to Proxima Centauri, develop breakthroughs in clean energy, in longevity, and in climate management. Whatever Herculean task you see before the human race, there’s a very high likelihood that the people who solve it will come from this particular subset of the population. And, these are the ones disproportionately bearing the tax load and funding the social programs we use to bridge the gap until we solve our biggest societal challenges.

Graduating Class

Education isn’t only an economic machine. Schools socialize kids. They function as childcare for working parents. For many students they’re a reliable source of meals, structure, and stable adult relationships—including mandated reporters when things go wrong at home. All true.

But none of that makes it irrational to differentiate instruction for high-aptitude students. We already accept differentiated schooling as normal when it’s framed as a need: special education, IEPs, individualized accommodations, specialized staffing. The principle is established. The controversy is selective.

In practice, “equity” objections often backfire. The kids who most need school as a ladder—capable students from low-income families—are the ones most likely to be left in a one-speed classroom. Affluent families simply opt out: private schools, tutors, test prep, enriched environments, and the Ivy League conveyor belt. If you want accelerated education to be less elitist, you provide it publicly and at scale.

And it doesn’t require doubling budgets. Most acceleration is pacing and grouping, not exotic content—at least until later in high school. Magnet schools already prove the model works; most states already have gifted programs, just inconsistent, under-resourced, and politically timid. You don’t need a new system. You need permission to take mastery seriously.

The bigger point is macro: we can’t math our way out of debt and entitlement pressure by redistributing a stagnant pie. We need faster growth. That means either more labor input—more hours, higher taxes—or more output per hour via innovation. If public education is justified by its externalities, then starving the cohort most likely to generate those externalities is self-sabotage and malinvestment. And if individualized education is a moral necessity for one tail of the distribution, it’s hard to argue it’s immoral for the other.