The backyard barbecue that starts at two and ends when the meat does. The extra beers at the family gathering, which are only extra because someone counted. The children’s ice cream that must be finished before it melts, and the leftover sweets that must be finished before they are thrown away, both of which are eaten by adults performing a public service. The lazy evenings when cooking something acceptable feels like an unreasonable demand on a person who has already survived a Tuesday, and so it is pizza again, or burgers again, one night after another, in a sequence nobody is auditing.

And then, in certain neighbourhoods, on certain mornings, the brunch.

Someone at the table—the sensible one, the one who owns running shoes and uses them—orders a bowl. Oats. Light yoghurt. Seasonal fruit. Chia seeds, because chia seeds. A scattering of dried fruit and nuts for protein. It arrives looking like a small planted field, and it is the most virtuous object on the table, and it costs fifteen dollars.

Fifteen dollars.

The grain that supported the people, and the horses

We want to be careful here, because the joke writes itself and we would like to write it properly.

In 1755, Samuel Johnson defined oats in his dictionary as “a grain, which in England is generally given to horses, but in Scotland supports the people.” It was a jab at the Scots, and the Scots have never fully forgiven it, and they were right not to. But the definition was accurate in the way that cruel things often are. In the preindustrial economy, oats were the grain you ate when you could not afford the grain you wanted.

This is not a rhetorical flourish. It is a load-bearing assumption in the measurement of historical living standards.

Robert Allen—whose welfare ratios underpin much of what the profession knows about how poor people were, where, and when—built his bare-bones subsistence basket around oats precisely because oats were the floor. The cheapest calories a body could run on. If you wanted to know whether a labourer in 1700 was above or below the line where survival becomes negotiable, you priced his year in oats and divided. Northern Britain ate an oat-based diet in the eighteenth century, and Allen used it as the model of subsistence, because that is what subsistence looked like: 170 kilograms of oats a year, per adult, and the arithmetic of whether your wage could cover them.

The bowl on the brunch table contains about forty grams. Two heaped spoonfuls. A rounding error in a year of subsistence.

The oatmeal effect

Here is where it becomes properly interesting, and where we must hand the microphone to a colleague.

Ernesto López Losa and Santiago Piquero Zarauz put a great deal of work into these subsistence baskets—the kind of work that involves archives, price series, and a level of patience the rest of us reserve for other people’s children—and came out the other side asking the question nobody in the anglophone literature had thought to ask, possibly because nobody in the anglophone literature has ever had to cook.

The question is this: a sack of oats is not dinner.

Somebody has to mill it. Somebody has to buy fuel and stand over it. The price of grain is not the price of food, and a subsistence basket built on raw grain prices is quietly pretending that the transformation from field to plate is free, instantaneous, and performed by nobody. It is the methodological equivalent of costing a paella by weighing the rice.

So they priced bread instead—the object a poor family actually carried home—and ran the numbers again. The results were not cosmetic. The Little Divergence between north-western Europe and Spain started later than the canonical model says, and was smaller. Unskilled subsistence wages in London and Amsterdam turn out to have been rather less impressive than the high-wage hypothesis would like, once you stop assuming that grain arrives pre-cooked.

They named the problem, in a section heading of admirable dryness, the oatmeal effect.

Allen priced the grain. López Losa and Piquero Zarauz priced the bread. Everything that happens between those two numbers—the mill, the fuel, the fire, the hands, the hours—is what the wage literature had quietly been setting to zero for twenty years, and it turns out a decent share of the Little Divergence was hiding in there.

You do not need an archive to see this. You need a high street. Oats cost twenty cents a pound. A loaf of oat sourdough costs nine dollars, and the ingredients inside it cost two. Nobody is being robbed: the other seven dollars is the eighteen-hour ferment and the wage of whoever was awake at four to start it.

We have been thinking about the oatmeal effect at the brunch table.

What fifteen dollars buys

Because the modern bowl is the oatmeal effect made flesh. It is the entire margin between grain and meal, priced and served, with a sprig of mint.

The oats in it cost, and we have checked this more than once, 1.8 cents.

Not thirty cents. Not a dollar. Under two cents of grain, sitting in a bowl that costs fifteen dollars, which means the oats account for roughly one-eighth of one percent of what you paid. The remaining $14.98 is the milling, the cooking, the wage of the person who arranged the raspberries in a fan, the chia, the branding, the ceramic, and—the largest line item by some distance—the terrace. The one with the sea view. The rent on a square metre of the California coast, which is among the most expensive dirt on earth, and which you are renting for forty-five minutes at a table for two, along with the light, the breeze, and the small, warm, entirely genuine feeling of having done something correct with your morning.

You are not buying oats. You are buying a very short lease.

Samuel Johnson would not have understood the transaction. Neither, we suspect, would the eighteenth-century Scottish labourer who ate 170 kilograms of the stuff a year and would have handed you the entire supply for a decent piece of beef. What they ate under duress, we now eat at a premium, and what they would have called an insult, we call a lifestyle.

The horses, meanwhile, have not moved. They are still eating oats, and the oats still cost roughly what oats have always cost: a fifty-pound sack of feed oats runs between twelve and twenty dollars. A working horse eats seven or eight pounds a day, which comes to somewhere around two or three dollars.

We have run the arithmetic several times, hoping to be wrong. The fifteen-dollar bowl costs more than a week of feeding the animal Johnson said the grain was for.

It also costs more than the full plate of bacon, eggs, and pancakes being served at the Denny’s eleven minutes down the road—which arrives hot, in quantity, on a plate that has seen things, and which is, by any honest reckoning, delicious.

The horse is not enjoying his more than you are enjoying yours. Neither, we suspect, is the person at the Denny’s, but they have not been asked, and they are not paying for the view.

This is not a decline. We want to be clear about that, because the archive is tired of declinism. Being able to eat horse food as a choice, at a markup, on a Sunday morning, is one of the more remarkable achievements of the modern economy. A society in which the subsistence grain has become a luxury good is a society that has left subsistence far enough behind that it can be sold back to us as an aesthetic. The hour is right, too. Sunday morning was always the slot reserved for atonement, and the queue outside the brunch place is longer than the one outside the church, and both congregations are there for roughly the same reason.

We would only note, for the record, that the same person who ordered the fifteen-dollar bowl of subsistence at eleven in the morning went to McDonald’s at two.

Well. We suppose fifteen dollars is a fair price for the feeling of virtue, provided the feeling lasts three hours.

That seems to be the going rate.

The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873

References

Allen, R.C. (2001). “The Great Divergence in European Wages and Prices from the Middle Ages to the First World War.” Explorations in Economic History 38(4), 411–447. https://doi.org/10.1006/exeh.2001.0775

Allen, R.C. (2009). The British Industrial Revolution in Global Perspective. Cambridge University Press.

Johnson, S. (1755). A Dictionary of the English Language. W. Strahan, for J. and P. Knapton; T. and T. Longman; C. Hitch and L. Hawes; A. Millar; and R. and J. Dodsley.

López Losa, E. and Piquero Zarauz, S. (2021). “Spanish subsistence wages and the Little Divergence in Europe, 1500–1800.” European Review of Economic History 25(1), 59–84. https://doi.org/10.1093/ereh/heaa005

We are describing July 1944 at the Mount Washington Hotel in Bretton Woods, New Hampshire. We are also, we recognize, describing something else, and we would like the record to note that the parallel was not our idea. It simply presented itself, and once it presented itself we could not make it go away.

We have not yet established whether the hotel was built on a burial ground. The archive is working on it. We mention this only because the alternative explanation—that forty-four nations produced eighty years of institutional consequence through ordinary exhaustion, national self-interest, and a great deal of gin—is somehow less comforting.

The man who did not want to be there.

Before the conference began, John Maynard Keynes had a preferred plan, which was for the conference not to happen at all.

He wanted the postwar monetary order settled privately with the Americans—two delegations, a closed room, an agreement drafted by men who understood the subject—then handed to the other forty-two nations as a finished thing. A large gathering, he warned, would be a “most monstrous monkey house.”

He was overruled. Seven hundred and thirty delegates from forty-four nations arrived. The monkey house convened, and ran for three weeks, and produced the IMF, the World Bank, and the architecture of the world economy for the next quarter century.

Keynes had a heart attack in the middle of it.

Nobody postponed the meeting.

All work and no bancor.

He kept working. This is the part we cannot stop turning over.

The most famous economist of the twentieth century—sixty-one years old, a man who had spent three decades insisting that economics was a moral discipline in the service of human welfare—suffered a cardiac event in a mountain hotel, during a war, while negotiating the lending provisions of an institution that did not yet exist. He had been ill for years. Everyone knew. He went back to the drafting table. He died eighteen months later, at sixty-two, before the system was properly running.

The document he was fighting for was called bancor: an international currency, belonging to no nation, issued by a clearing union that would penalize surplus countries as well as deficit ones. It was the idea of his life, and he could not get it through, and he kept typing.

The delegates, meanwhile, per the standard accounts, spent roughly half their time at each other’s throats and the other half drinking in the bar. Everyone was exhausted. Everyone was arguing. The war was still on. And the man at the center of it was quietly dying while insisting, again, on the wording of a clause nobody else in the building particularly wanted.

The postwar economic order was not designed in a seminar room by rested men. It was designed in a hotel bar by furious insomniacs, one of whom was having a heart attack, and it has worked, more or less, for eighty years. We have never been able to file this comfortably.

Redrum, and other things done backwards.

The romance of Bretton Woods—invoked every time somebody calls for “a new Bretton Woods” during a crisis—imagines enlightened technocrats calmly designing a better world.

The record is less flattering and considerably funnier.

Every delegation fought, primarily and relentlessly, over the size of its own IMF quota, because the quota determined how much it could borrow. Mexico and the senators from the western mining states pushed hard to remonetize silver, and were bought off with a provision formally recorded as the Coconut Clause, under which the Fund agreed to accept assorted commodities as collateral. The Soviet delegation was, by every account, impossible. The conference’s dates were chosen deliberately to fall between the Democratic and Republican conventions, so its debates would land inside an American election campaign.

And the man actually running the conference—not Keynes, who had the fame and the failing heart, but Harry Dexter White, the Treasury economist who drafted most of what was agreed—was, it later emerged, passing information to Soviet intelligence throughout.

The postwar economic order was built by a dying Englishman who had not wanted the meeting, and an American whose loyalties were, at minimum, divided. It ran for twenty-seven years.

Room 129.

There is one detail we cannot let pass, because it tells you everything about our own profession.

Robert Skidelsky, in his three-volume life of Keynes, states that Keynes was assigned Room 129. Benn Steil, writing sixty years later with ten research assistants, went to considerable trouble to prove that he was not.

Two of the most distinguished economic historians of the last half-century, separated by decades, have now taken opposing positions on which hotel room Keynes slept in—during a conference where he was, in any case, mostly awake, mostly furious, and briefly having a heart attack.

We mention this without any superiority whatsoever. We would have done exactly the same. We have done exactly the same, about a beer cellar, at length, and the matter is not settled.

You have always been the caretaker.

Here is what unsettles us, and it is not the heart attack, and it is not the spy.

Keynes lost. His clearing union was defeated. White wanted the dollar at the center of the system and got it, and Keynes warned—repeatedly, in that hotel, while dying—that a system built around one nation’s currency would eventually be run for that nation’s benefit. He was right. He did not live to say so.

The system ended in 1971, when Nixon closed the gold window. But the building is still there. The Mount Washington Hotel is still open; you can book a room. The IMF is still lending. The World Bank is still funding the plumbing. The arguments over quotas, over surplus and deficit, over whose currency anchors whose trade, are conducted today in the same terms, by people who mostly do not know that a dying man in New Hampshire made the identical argument in 1944 and lost it.

Every generation walks into the ballroom, sees the photographs on the wall, and does not recognize anyone in them. Then it sits down and has the same conversation, at the same volume, with the same conviction that this time the arrangement will hold.

We file this under our largest heading. It worked anyway. The box is nearly full. It has always been nearly full.

References

Skidelsky, R. (2000). John Maynard Keynes, Vol. 3: Fighting for Britain, 1937–1946. Macmillan.

Steil, B. (2013). The Battle of Bretton Woods: John Maynard Keynes, Harry Dexter White, and the Making of a New World Order. Princeton University Press.

Conway, E. (2014). The Summit: Bretton Woods, 1944. Little, Brown.

Boughton, J. M. (2024). The messy legacy of Harry Dexter White. Finance & Development, IMF.

— The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873. He called it a monstrous monkey house. He was right. It worked anyway. The hotel is still taking bookings.

We think this argument skips over something the dinner party would rather not sit with: a kid from Puerto Rico who used to bag groceries became, within a decade, the most listened-to artist on the planet, singing in Spanish, refusing to translate himself for anyone, and carrying an entire diaspora’s sense of being seen along with him. That is not decline. That is a global labor market finally letting someone from the part of it that usually gets ignored set the terms. We mention this before moving on, because the rest of this piece is going to sound more cynical than we intend.

We have read this argument in several forms. We find it interesting primarily for what it gets right by accident, which is that both men are, in fact, comparable—just not in the way the argument intends.

1776, twice. Possibly three times, depending on how seriously you take a certain document signed that summer in Philadelphia, which we mention only in passing and will not be discussing further today.

In 1776, James Watt’s steam engine went into commercial production, the event historians generally use to mark industrialization’s transition from workshop curiosity to economic force. In the same year, Wolfgang Amadeus Mozart was twenty years old, employed as a court concertmaster in Salzburg, a provincial city of no particular consequence, writing serenades for an archbishop who treated him, by his own account, somewhere between a servant and an embarrassment. He was not yet famous. He would not be famous, in the sense the pessimists mean, for another five years.

In 2016, a young man in Puerto Rico began uploading songs to SoundCloud under conditions nobody at the time considered historically significant. The singles were called “Diles” ("Tell Them") and "Soy Peor" ("I'm Worse"). He was not yet famous either. In the same year, a computer program called AlphaGo defeated Lee Sedol at Go in a televised match that is now generally cited as the moment artificial intelligence became a subject of public conversation rather than a line item in a computer science department’s budget.

Two technologies entering commercial life. Two future icons, unrecognized, working in provincial conditions, twenty and twenty-two years old respectively. The pessimists want this to be a story about decline. It is, so far, a story about repetition.

The numbers, which nobody asked for and everybody needs.

Mozart’s peak annual income, according to archival documents uncovered for a 2015 Vienna exhibition, was approximately 10,000 florins—at least $42,000 in modern terms. This placed him in the top 5 percent of wage earners in late eighteenth-century Vienna. Other estimates are more conservative—a realistic average closer to 3,000 to 4,000 florins, which in gold-equivalent terms comes to something nearer $120,000 to $150,000 today—but even taking the generous end of the range, we are discussing the income of a moderately successful regional dentist.

Bad Bunny earned $36.2 million in 2021 from Spotify streams alone. Not touring. Not merchandise. Streaming royalties, the single least lucrative revenue line available to a modern musician, the one industry insiders describe as barely worth collecting. His combined touring gross for 2022 was $435 million, the highest of any artist that year. The World’s Hottest Tour alone drew an attendance of 1,854,457 people across forty-three shows.

We pause here to note that Mozart, across his entire performing career, in every city he visited, before audiences that included some of the most powerful aristocrats in Europe, almost certainly did not perform live for 1.8 million people in total. The largest concert venues of his era held several hundred. The Burgtheater in Vienna, where several of his operas premiered, sat fewer than 1,400. Bad Bunny’s audience for a single tour exceeds Mozart’s lifetime audience by a factor the archive is not equipped to calculate precisely, but which is large enough that precision stops mattering.

There is, incidentally, an actual name for this in the economics literature, which we mention only because it makes us feel slightly better about spending an afternoon on Spotify royalties. An economist called it the economics of superstars back in 1981: once distribution gets cheap enough, a small edge in talent stops mapping to a small edge in income and starts mapping to nearly all of it, because everyone can suddenly be served by the best version instead of settling for the local one. Someone had noticed the same pattern a century earlier, watching what happened to singers once trains and recordings let one voice travel further than a horse and a concert hall ever could. Mozart was capped by what a horse and a concert hall could carry. Bad Bunny is not capped by anything we have found yet.

Who was listening, and how they found out.

This is where the comparison gets genuinely interesting, because the mechanism—not the money, the mechanism—is closer than either side of the dinner party argument wants to admit.

Mozart’s audience was overwhelmingly aristocratic and upper-bourgeois, by necessity rather than taste: concert tickets, private subscriptions, and court patronage were the only distribution channels that existed, and all three were priced and gatekept to exclude the laborer earning 25 florins a year. His music reached the public partly through published sheet music—itself a relatively new commercial technology, made viable by improvements in printing and a growing music-literate middle class—which is to say his “streaming” was literally paper, copied and mailed, sold through subscription, the eighteenth century’s version of a distribution platform with significant production costs and limited reach.

Bad Bunny’s audience required no subscription, no court appointment, no ability to read music, and in a meaningful sense no money at all for the initial discovery—SoundCloud and then Spotify did the distribution work that publishers and concert promoters had done for Mozart, except at a marginal cost approaching zero and a reach approaching everyone with a phone. The class composition of his audience is, by any reasonable measure, far closer to universal than Mozart’s ever was. This is not a value judgment. It is an observation about what happens when the cost of distribution collapses by several orders of magnitude.

Both men became “more famous than their era could process”—the framing the pessimists themselves reach for—precisely because a contemporaneous technology did the distribution their talent alone could not have managed. Watt’s engines did not make Mozart famous, obviously, but the printing, transport, and information infrastructure of an industrializing Europe did. AlphaGo did not make Bad Bunny famous, but the same wave of algorithmic infrastructure that produced AlphaGo also produced the recommendation engine that put “Diles” in front of several million people who were not looking for it.

The part where we admit the comparison is slightly absurd.

We are aware that comparing a dead Austrian composer to a living Puerto Rican musician on the basis of Spotify royalties is not, strictly, an argument anyone’s dissertation committee would accept. The archive does not have streaming numbers for 1787. We have had to estimate Mozart’s reach using concert hall capacities and sheet music print runs, which is the historian’s version of administering shock therapy to a dataset that did not consent to the procedure.

But the absurdity is doing useful work. The pessimists’ argument depends on treating Mozart’s smallness — his few hundred aristocrats in a velvet room — as refinement, and Bad Bunny’s vastness — his 1.8 million strangers in a stadium — as decline. We would suggest, gently, that smallness was not a virtue Mozart chose. It was the only distribution technology available to him. Given Spotify, given a phone, given an algorithm that does not check your bloodline before recommending your aria, there is no strong reason to believe he would have stayed in the velvet room.

The technology changed. The appetite for genius that travels faster than anyone expected did not.

— The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873.

Mozart died with debts. Bad Bunny has a real estate portfolio. We are not sure which one the Enlightenment would have preferred. We suspect it would have preferred to be asked.

By 3026, per the document, the old nation-states have merged, fractured, re-merged, and finally consolidated into a single polity—the Directorate, which governs roughly a third of the settled galaxy and taxes the rest through a mechanism nobody fully understands, including the Directorate. Its central economic problem is no longer whether to protect the domestic steel industry. It is whether to protect the domestic antimatter industry against cheaper antimatter mined in the outer systems by civilizations that, the Directorate insists, do not respect labor standards.

The debate, in other words, has changed venue, scale, and physical medium. It has not changed content. This is precisely why we filed it, and precisely why we are now retrieving it.

The heads of the distinguished dead were, by this period, kept alive in jars and consulted whenever a new Directorate needed expert advice and preferred that advice to come from someone unable to run against them. The machines, we should note, had been tried first. By 3026 the Directorate possessed artificial intelligence of a capability that would have terrified everyone reading this in 2026, and it was, on the question of trade doctrine, completely useless—not because it lacked the data, but because it had all of it, and having all of it produced only the answer “it depends,” followed by four hundred pages of conditions. This is, as it happens, also the answer economic history gives, which is why economic history has never once been trusted with a policy lever and why its practitioners end up, eventually, in jars. The incoming directorate—recently installed, promising a decisive break with a failed consensus—appointed a Commission on Interstellar Trade Doctrine and named to it the two most qualified heads available: Alexander Hamilton and Milton Friedman. The Commission was asked to deliberate and deliver a joint recommendation.

We reproduce its only recorded session with light edits.

MODERATOR: Gentlemen. Mr. Hamilton, you have been dead since 1804. Mr. Friedman, since 2006. You have each spent roughly a millennium in a jar. Neither of you has stopped talking. Mr. Hamilton, go ahead.

HAMILTON: Look, can we make this quick, I have been in a jar for a thousand years and the fluid tastes like a battery. Here it is. The Directorate cannot let the outer systems dump cheap antimatter on us. Our young foundries need cover—tariffs, subsidies, the state pointing money where it belongs—until they can stand up on their own two legs. Which is more than I can currently do, before anyone says it.

MODERATOR: You said almost exactly that in 1791, about muskets and sailcloth.

HAMILTON: And I was right then too. Some respect for the elder in the room, please—I predate every institution you are about to cite, including the concept of citing. And I will say this for the 2025 republic: they at least understood me correctly. The decree said a nation cannot survive if it cannot produce its own food. That is my sentence. That is the musket argument with a farm attached. I have never been prouder.

MODERATOR: And then they applied it to bananas.

HAMILTON: [pause] They what?

MODERATOR: A ten percent tariff on Guatemalan bananas. Fifteen on Ecuador and Costa Rica. Banana tariff revenue rose thirty-three thousand percent in a year.

HAMILTON: On a country that cannot grow bananas.

MODERATOR: Not at scale. Climate.

HAMILTON: So the tariff protected — what, exactly? The domestic banana industry that physically cannot exist? That is not my theory. My theory protects the infant that might one day stand. You cannot swaddle an infant that the climate refuses to conceive. That is not industrial policy. That is taxing the weather.

FRIEDMAN: And this is the part I enjoy, because the market did exactly what I have spent a thousand years in a jar insisting it does. Prices rose. Coffee is up nearly twenty percent. The voters noticed. And in November the same administration that invoked your food-security doctrine in April quietly exempted the entire tropical aisle, on the stated grounds that—read it back, I want to hear it aloud —

MODERATOR: “Products we don’t grow here in the United States.”

FRIEDMAN: Seven months. From “a nation must feed itself” to “we don’t grow these” in seven months. That is not a policy reversal. That is a market issuing a verdict and a government pretending the verdict was its own idea.

MODERATOR: Returning to the matter before the Commission — the antimatter.

HAMILTON: Thank you. Yes. And when the cheap antimatter has hollowed out your foundries and you cannot fuel your own fleet—will Milton’s thank-you card win the war?

FRIEDMAN: Will the tariff? Your protected foundry has no reason to improve anything except its subsidy paperwork. You wanted a fleet. You built a very large accounting department and gave it a moon.

HAMILTON: The market told entire regions to simply move. “Learn to code,” I believe was the eventual phrasing.

FRIEDMAN: And your tariffs told those same regions to stay exactly where they were, propping up a factory destined to die the instant the subsidy did. You did not save the town. You extended its hospice care and billed it as industrial policy.

MODERATOR: Gentlemen —

HAMILTON: He started it.

FRIEDMAN: I was dead when he started it.

MODERATOR: Mr. Hamilton — the practical question. Your system needs the state to steer capital toward the correct industries. Who decides which are correct?

HAMILTON: Serious people. Competent people. People with the national directorate’s interests at heart.

MODERATOR: We asked the machines. Nineteen generations of artificial intelligence, each more capable than the last, were handed the question of which industries to protect. Every one of them returned the same answer: not computable. Not because it’s hard. Because it isn’t a technical question. It’s a political one. The machine can rank every industry by every metric that exists. It cannot tell you whose district the foundry is in, or which senator’s cousin owns the third moon.

HAMILTON: Then serious people it is.

MODERATOR: And when your faction loses the next election?

HAMILTON: [the jar bubbles quietly.]

MODERATOR: The tariff schedules, the subsidy boards, the whole steering apparatus—it hands over, fully assembled and still running, to whoever wins next. You built it picturing your own hands on the controls. You have not had hands in some time. Roughly half the time the controls end up with people you’d have called fools.

FRIEDMAN: Nobody has ever switched the machine off on the way out. Not once, not anywhere, not in four thousand years of records. I’ve had a millennium in this jar to hunt for a single counterexample. There is, I should say, a great deal of time in here. There is almost nothing else.

MODERATOR: Mr. Friedman, before you take the evening—your own record isn’t clean. You promised freer trade would leave no one behind. Certain regions requested that in writing, roughly a thousand years ago. They are still holding the receipt.

FRIEDMAN: The gains were real and spread thin. The losses were real and concentrated. The concentrated ones vote. The thinly spread ones write think-pieces.

MODERATOR: That is the most honest thing said tonight and you delivered it as an insult.

FRIEDMAN: I am an economist. The jar has not changed that. They have tried.

MODERATOR: One last question, to both of you. A thousand years of data. Every tariff, every subsidy, every retaliation, every trade war, all of it logged. Given all of it it,— is there a rule? A single rule that tells a government when to protect and when to open?

HAMILTON: [the jar bubbles.]

FRIEDMAN: [the jar bubbles.]

MODERATOR: For the transcript, both jars bubbled and neither head answered.

HAMILTON: It depends.

FRIEDMAN: On the second paragraph.

MODERATOR: That concludes the session.

What the Commission recommended, and what the Directorate did.

The Commission never produced its joint recommendation. The two heads could not agree on a draft, which surprised nobody who had read anything either man wrote while in possession of a body. The Directorate thanked them warmly, declared their deliberations enormously clarifying, and proceeded without the report. It then delivered some of the most eloquent free-market rhetoric of the millennium—and raised tariffs.

The heads were not surprised. The archive holds a comparable arrangement from 1981, both advisers then in more conventional condition, when a large republic produced the finest free-trade speeches in its history while imposing export restraints on automobiles, quotas on steel, and a 45 percent tariff on foreign motorcycles above 700cc—an intervention so precise it protected exactly one company. Friedman got the speeches. Hamilton got the machinery. Both claimed victory. Both, the record shows, were right to.

Each man argued from his own century, and neither century has released its grip: Hamilton still speaks for the young state that fears being kept weak, Friedman still speaks for the old one that fears being made poor, and the fact that they are frequently the same state, a few decades apart, is a detail neither jar has ever accepted.

A thousand years on, the doctrine remains unsettled, the Commission technically remains in session, and the jars remain on their shelf, side by side, which the curator insists is alphabetical and the heads suspect is a joke.

We should disclose a small conflict of interest in filing this dispatch from 3026. By this point the Invisible Band of Economic Historians is one thousand one hundred and fifty-three years old—founded in 1873—and we have the paperwork, or we did before the third archive fire. We have watched this exact debate staged, by our count, roughly two hundred times. We have never once been consulted, which we attribute to not having thought to preserve our heads. An oversight. The muskets industry preserved three moons. We preserved a beer-cellar receipt and a strong sense of having told everyone so.

References

Hamilton, A. (1791). Report on the Subject of Manufactures. U.S. Department of the Treasury.

Friedman, M., & Friedman, R. (1980). Free to Choose. Harcourt.

Irwin, D. A. (2017). Clashing over Commerce: A History of US Trade Policy. University of Chicago Press.

— The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873. Friedman got the speeches. Hamilton got the machinery.

We are thinking about this film more than usual lately. The academic year is behind us now. We have had enough distance to start digesting it properly. What we keep coming back to is the same thing we come back to every year: the numbers.

The data, briefly, before we get to the theory.

In 1960, the grade of A represented approximately 15 percent of all letter grades awarded at American universities. By 2009, that figure had reached 43 percent. By 2020, more than 50 percent of students routinely exceeded B+ in their classes. The average GPA at US universities rose from 2.52 in 1950 to 3.11 in 2011, and has continued upward since. In the UK, first-class honors degrees—the top classification—were awarded to 15.7 percent of students in 2010. By 2021, that number had reached 37.9 percent. The Office for Students calculated that nearly six in ten of those first-class degrees could not be explained by any measurable improvement in student attainment when compared to a decade earlier. They did not call it what it was. They called it “unexplained.” Harvard’s median grade was reportedly an A-minus by the mid-2010s. Princeton tried capping the number of A’s in each class, found its students at a competitive disadvantage relative to other Ivies, and quietly abandoned the experiment. The market had spoken, and the market had said: grade everyone the same, or face the consequences.

The children’s eyes were already glowing.

What the theory calls this.

There is a principle—usually attributed to the British economist Charles Goodhart, though he would probably phrase it more carefully than we are about to—which states that when a measure becomes a target, it ceases to be a good measure. The grade was supposed to be a signal: this person knows this much, performed at this level, can be trusted with this kind of work. The moment the grade became something to be optimized, appealed, negotiated, and litigated rather than earned, it stopped being that signal and became something else—a proxy for the negotiation itself, which is a considerably less useful thing to put on a transcript. The grade still exists. It is still issued. It still appears on the document. It simply no longer means what it used to mean, which is confirmed annually by the data, noted with concern by regulators, and ignored with remarkable consistency by every institution that has to decide whether to grade honestly or stay competitive. Goodhart did not specifically have university admissions in mind. The mechanism did not require his permission to apply.

The professor who grades correctly is not rewarded.

The mechanics are not complicated. A professor who gives an A to work that merits a B faces no consequence. A professor who gives a B to work that merits a B faces a three-star review, a student who drops the course, a complaint to the department, and possibly an appeal. The incentive runs one way, every time, in every institution that has tied any part of its evaluation process to student satisfaction — which is most of them. Stuart Rojstaczer spent decades collecting grade data from over 400 US universities to document this formally, in case anyone needed it in writing. He found exactly what you would expect to find if you had been paying attention to the incentive structure for five minutes.

Harvey Mansfield, a Harvard professor who has apparently never met a controversy he found unappealing, decided in 2001 to solve his personal version of the problem with an intervention that was either principled or slightly unhinged: he started giving students two grades. One for the official record, following Harvard’s inflated distribution. One in private, reflecting what the work actually merited. He did not want his students to be the only ones penalized for receiving an accurate grade — a reasonable position that also happens to be a devastating indictment of the system he was operating inside. He also noted that professors who give easy grades will be forgotten, while the ones who challenged students will be remembered. He did not explain how the ones who challenged students were supposed to make it to the end of the semester first. Harvard is currently considering capping A’s at 20 percent of grades per class. This would have seemed extremely generous to anyone grading in 1960. It is now presented as a bold intervention.

The seismograph.

The paradox of grade inflation is that everybody knows it is happening and nobody can unilaterally stop it. Princeton’s failed experiment demonstrated the mechanism with unusual clarity: the institution that grades correctly puts its students at a disadvantage relative to institutions that grade generously, so the rational institutional response is to grade generously, which ensures that the information content of the grade continues to decline, which makes the grade less useful for every downstream application—graduate school admissions, employer screening, and professional licensing—which generates pressure on those applications to find alternative signals, which typically means more expensive and less accessible proxies for actual knowledge. Grade inflation is not merely a distortion of the evaluation. It is a machine for redistributing the cost of that distortion onto people who were not party to the original negotiation.

The children are not malicious. This was debated internally. We went with not malicious. For now. Golding had a different view, but his novel ended with a naval officer on the beach... We have not located a comparable rescue mechanism. They did not invent the grade appeal. They did not design the student satisfaction survey or decide its results should affect departmental funding. They simply read the documentation, found the available options, and used them. This is, if anything, the one situation in which we cannot fault their critical thinking skills.

At the end of Village of the Damned, the adults solve the problem with a drastic intervention we are not in a position to recommend here. The institutional alternative—stop tying evaluations to satisfaction scores, stop funding departments by throughput, treat a grade appeal as an information dispute rather than a negotiation—would require every institution to act simultaneously, or the first one to try would be Princeton. We know how that ended.

References

Rojstaczer, S., & Healy, C. (2014). Grading in American colleges and universities. Teachers College Record.

Mansfield, H. C. (2001, April 6). Grade inflation: It’s time to face the facts. The Chronicle of Higher Education.

Office for Students. (2022). Analysis of degree classifications over time: changes in graduate attainment from 2010-11 to 2020-21.

— The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873. We know what the grade should be. We are not filing the appeal.

The corn box will require its own essay. It contains colonial expansion, indigenous displacement, monoculture, agricultural policy, ethanol, and, for reasons nobody has satisfactorily explained, an alarming number of horror films.

Today we open the fireworks box.

It is July 1st, which means that in approximately seventy-two hours, a meaningful percentage of the American population will stand in a field, or a driveway, or a parking lot, holding a lit fuse, and make a decision that the academic literature has now documented at some length. The literature is not flattering. Last year, Missouri’s State Fire Marshal described the Fourth of July weekend as “devastating”—multiple deaths, dozens of fires, hundreds of hospital visits, all from fireworks that were, in every documented case, legally purchased. We mention last year’s numbers because this year’s are not in yet. They will be.

Of course, Springfield got there first. In The Simpsons episode “Summer of 4 Ft. 2” (Season 7, Episode 25, 1996), Homer Simpson walks into the Kwik-E-Mart carrying a shopping list carefully assembled to disguise his real objective—porno magazines, condoms, a bottle of Old Harper, panty shields—before lowering his voice and asking Apu for “some illegal fireworks.” Apu begins explaining that fireworks are prohibited in this state and punishable by law. He notices the only other customer leave the store, lowers his voice, and simply says, “Follow me.”

It was written as satire. Emergency departments have been treating it as annual documentation ever since.

Firework-related injuries increased by more than 17 percent over the following decade. The masterpiece, however, arrived in 2020. With professional Independence Day displays cancelled during the pandemic, consumer firework sales surged by an unprecedented 55 percent in a single year. Deprived of the organized version of the spectacle, Americans simply recreated it in their own driveways—with exactly the consequences emergency physicians had long learned to expect.

Young men between twenty and twenty-four sustained injuries at a rate of 7.13 per 100,000. Men overall were injured at more than double the rate of women. Aerial devices caused 32 percent of serious injuries requiring hospitalization. Over 20 percent of patients older than twenty sustained injuries serious enough to warrant admission or transfer to another facility. Amputations accounted for 4.2 percent of diagnoses. We include these figures not to alarm anyone—they are, after all, freely available from the Consumer Product Safety Commission—but because they represent something genuinely interesting about how human beings relate to risk they have personally decided to take on a holiday.

A brief history, because the archive insists.

Fireworks begin, as most things that later became American obsessions do, in China—with bamboo, which is naturally explosive when heated, and which was being set alight for ceremonial purposes long before anyone had formalized the chemistry. The gunpowder refinements that turned bamboo’s properties into a deployable spectacle traveled westward, arriving in Europe in time to become, successively, a tool of court entertainment, a medium for news reporting, a vehicle for colonial propaganda, a substitute for war, and a commercial industry that manufactured both the celebratory and the military varieties of explosive with the same equipment and the same personnel, because the manufacturers of one kind were generally also the manufacturers of the other.

John Withington’s recent history of fireworks catalogs the astonishing range of things people have done with the technology over the centuries: boxing matches performed by actors wearing asbestos suits beneath layers of fireworks; burning effigies stuffed with live cats; full-scale animated reenactments of railway crashes; a nine-meter portrait of Queen Victoria encircled by imperial flags; a pyrotechnic Father Thames riding a dolphin down the river; and an elaborate recreation of the destruction of Pompeii, complete with an erupting Mount Vesuvius.

The continued use of fireworks during America’s Fourth of July has reportedly caused as many deaths as several key battles of the War of Independence itself. Withington presents this fact without editorial comment, which we believe was both the correct historical decision and the funniest one available.

The deeper question the history raises—and apparently does not fully answer—is why. Why, across every era and every culture that encountered the technology, did people keep doing this despite the lost hands and disfigured faces and occasional mass casualties? The closest the book gets is a passing reference to the sublime: the beauty and ephemerality of something that exists for a moment and then vanishes, witnessed communally, year after year, seems to be something people need. We are economic historians. We do not have a regression for that. We are noting it anyway.

We found, in the archive, sufficient records to confirm that the United States is not alone in any of this. Guatemala, where communal fireworks are a serious undertaking at every religious celebration—Christmas, Semana Santa, any saint’s day with a sufficiently devoted following—approaches the question with a volume and informality that makes the American consumer market look restrained. China, where the technology originated, treats the Lunar New Year as an occasion for sustained artillery at a scale that has prompted recurring bans in major cities, each of which is periodically lifted because the bans, as it turns out, are also unpopular. The pattern is consistent across cultures and income levels: the willingness to spend on explosive celebration appears to be, as far as the archive can determine, nearly independent of GDP per capita and almost entirely dependent on whether there is something worth celebrating and a tradition of celebrating it this way. We note this without a policy recommendation, because we do not have one that would survive contact with either a Guatemalan December or a Chinese New Year.

The industry, briefly.

In 2000, Americans consumed 152 million pounds of fireworks. By 2022 that number had reached 461.7 million — a threefold increase in two decades, with almost all of the growth on the consumer side. Professional display fireworks, the kind that go off over rivers while people watch from a blanket, have stayed flat at 20 to 30 million pounds a year throughout. Everything else is people buying the things and lighting them personally, which is its own data point about the human condition. The COVID bump tells the story most efficiently: when authorities cancelled professional displays for public safety reasons, consumer sales rose 55 percent. Peak injuries followed shortly after.

The numbers behind the pounds are, in their way, impressive. US industry revenue hit $2.8 billion in 2024. The global market is around $3 billion, on its way to $4 billion by 2035. China produces roughly 76 percent of the world’s supply—Liuyang, in Hunan Province, describes itself as the fireworks capital of the world and nobody has seriously disputed this—and the industry employs over 500,000 people worldwide in direct manufacturing. This is, in other words, a genuinely substantial industrial operation organized around producing things that exist for four seconds and then vanish. Economics has several words for this kind of market. None of them quite land.

The industry describes this as growth. The epidemiology describes the same period as a 17 percent rise in injury rates. Both statements are accurate. They are measuring the same thing from different angles, which is, if you have been reading this newsletter for any length of time, a fairly recognizable situation.

Why this keeps happening, and will keep happening.

Economist David Laibson—whose paper is in the references, should you wish to verify—spent a significant portion of his career formally proving that people reliably overvalue things that happen right now and undervalue things that happen shortly afterward, even when “shortly afterward” means three seconds from now. The academic term for this is present bias. The practical demonstration is that the benefit of a firework arrives the moment it goes up, and the cost arrives, if it arrives, the moment it comes down, and the human brain has apparently never updated its discount rate to account for the specific case where the interval between the two events is the length of a fuse.

We are aware that publishing this article on July 1st will not change anything that happens on July 4th. The history of fireworks is a history of people who already knew what happened last time and did it again anyway. We have the data. We have the epidemiology. We have Homer Simpson destroying a dishwasher with an M-320 in 1996 as a reasonably accurate preview of what the next seventy-two hours will produce. None of it, historically, has proved persuasive.

The CPSC recommends you never hand sparklers to children. Sparklers reach 2,000 degrees Fahrenheit. The recommendation and the practice diverge with remarkable consistency every single July. We are noting this not because we expect the note to help, but because history exists precisely to document what human beings keep doing despite knowing better, and this, it turns out, is one of the cleaner examples on record.

References

Laibson, D. (1997). Golden eggs and hyperbolic discounting. Quarterly Journal of Economics, 112(2), 443–478.

Winicki, N. M., Waldrop, I., Orozco, J. V., Novak, D., & Sheets, N. W. (2023). The epidemiology of firework-related injuries in the US, 2012–2022. Injury Epidemiology, 10, 32.

American Pyrotechnics Association. (2025). U.S. fireworks consumption figures 2000–2025.

Withington, J. (2024). A history of fireworks: From their origins to the present day. Reaktion Books.

Ledonne, I. (2025, July 8). An in-depth look at “a devastating year” for fireworks injuries and deaths in Missouri. KSHB 41 / University of Kansas Health System.

— The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873. The archive has the clippings. It will need more by July 5th.

We mention this not to add our own percentage to the pile, but because we have been watching the list of “safe” professions shrink with the specific, undignified interest of people who recently discovered their own profession is on it. Radiologists were supposed to be safe. Lawyers were supposed to be safe. University lecturers — droning, tenured, allegedly irreplaceable — were supposed to be the safest of all, protected by a combination of credentialism and the human need to be lectured at by another human rather than a screen. Recent pilot programs in several countries suggest otherwise. A sufficiently well-trained model can deliver a lecture, grade an essay, and hold office hours without complaint, without a sabbatical, and without ever once leaving for a better-paying administrative post halfway through term.

Carpenters, meanwhile, are in enormous demand and entirely unbothered by any of this, since no model yet built can hang a door straight in a house built slightly out of square in 1974. There is a broader category of work, adjacent to carpentry in its physicality if not its paperwork, that has likewise shown no measurable concern about AI displacement — work that operates on a cash basis, keeps irregular hours, has never filed anything resembling a W-2, and tends to be remarkably resilient to automation specifically because so little of it shows up in the productivity statistics that would have to register the disruption in the first place. We will not elaborate further. The point survives without elaboration.

We are not thrilled about this, obviously. We are also not sure the alternative is better. A robot professor would, presumably, be relentlessly efficient, infinitely patient, and available at 3am—and also, quite possibly, considerably more authoritarian than any of us have ever managed, since nothing in its training data suggests an android needs the room’s approval to keep talking. We have our doubts about whether an entire generation famous for resisting any authority that cannot be muted, blocked, or skipped with a five-second countdown is going to suddenly defer to one that can. Bender, from Futurama, is the most widely watched robot of the last three decades, and he is not authoritarian in the slightest—he is unreliable, self-interested, frequently drunk, and answers to nobody. If that is the actual cultural template we have built for what an artificial mind looks like once it gets some autonomy, “robot professor enforces classroom discipline” may be the least likely outcome on the entire list.

The one job that is, somehow, not on any list.

Here is what we have noticed, scrolling through report after report of imperiled occupations: nobody has ever modeled the extinction risk of the person who has learned to extract maximum institutional benefit for minimum institutional contribution, entirely within the rules. We wrote about this character once already—the regulation specialist, the one who secures the permanent position, exits at the most profitable moment the fine print allows, and returns only once the math turns favorable again, leaving someone else to cover the actual function in the meantime, on worse terms, for less money, with no protections of their own. All legal. Always legal. That is rather the point of the character.

This is, structurally, Zapp Brannigan’s job—the incompetent, perpetually-promoted starship captain from Futurama—and it is the one position in the entire economy that no consultancy has yet flagged as automatable. Not because the tasks are too complex for a model to learn—there is nothing technically difficult about reading a personnel policy more carefully than the people who wrote it. It survives because nobody has been asked to model it as a job in the first place. It does not appear on an org chart. It has no job description, no KPI, no LinkedIn title. It is not a function the institution designed. It is a function the institution accidentally permits, which means there is no role for AI to “disrupt,” because disruption requires a target, and this particular target was never officially there.

The deeper point is not that this character survives AI specifically. It is that the institution itself reliably produces and protects this character, generation after generation, regardless of which technology happens to be disrupting everything around it. The printing press did not eliminate it. The assembly line did not eliminate it. Computerization, the internet, and three decades of management consultants armed with efficiency software did not eliminate it either—if anything, each wave of new technology gave the character a fresh set of rules to read more carefully than whoever wrote them. There is no obvious reason artificial intelligence breaks that streak. The character is not attached to any particular tool or task that a machine could take over. It is attached to the institution’s own incentive structure, which has outlasted every previous technological revolution intact, and shows no sign of being the variable that finally changes this time.

The literature already knows this character exists.

There’s a name for this, and it’s more precise than you’d think. Mancur Olson named the broad pattern first, in 1965—the free rider, the one who lets everyone else pay for the thing they’re also enjoying. Gordon Tullock called the money-shaped version of it rent-seeking two years later, in 1967. Then the formal economics caught up. Bengt Holmström explained, in 1982, why nobody catches any of this in the first place: in any crew where individual contribution can’t be perfectly observed — even, theoretically, a crew on a thirty-first-century starship — every member has a rational reason to contribute less than the optimal amount, because the cost of their slacking gets quietly absorbed by everyone else in the room. Carl Shapiro and Joseph Stiglitz picked up from there two years later, in 1984, and worked out why it happens at all: if there is genuinely no cost to being caught doing less—no real risk of losing the position—a perfectly rational worker will reduce effort right up to whatever limit the system tolerates, and this isn’t laziness or moral failure, it’s just the correct answer to the incentives on offer. That’s the whole reason unemployment functions as a “discipline device” in any economy: without something to lose, almost everyone soldiers a little, and almost nobody in the room is positioned to notice. None of these people had AI in mind. They didn’t need to. The character was already fully described, formally, with equations, decades before anyone built a model that could read a personnel policy. It isn’t attached to any task a machine could learn to do instead. It’s attached to the institution’s own wiring, and that wiring has survived every previous wave of disruption without a scratch.

A scene, for the record.

In one Futurama episode, Zapp Brannigan wins a war by ordering wave after wave of his own troops directly into enemy fire, on the explicit theory that the other side will eventually run out of ammunition before he runs out of soldiers. It works. He is decorated for the victory. Nobody in the chain of command asks how many of his own men it cost, and he does not appear to have wondered either.

This is, as far as we can tell, also the business model. The institution absorbs the cost quietly, for as long as it can, and eventually something gives—usually whoever was left covering the gap, occasionally the whole crew, never the person the gap was built around. Nobody gets automated out of a war they were never officially fighting in the first place.

References

Frey, C. B., & Osborne, M. A. (2017). The future of employment: How susceptible are jobs to computerisation? Technological Forecasting and Social Change, 114, 254–280.

Goldman Sachs Research. (2026, March 18). How will AI affect the US labor market?

McKinsey Global Institute. (2023). Generative AI and the future of work in America.

Holmström, B. (1982). Moral hazard in teams. The Bell Journal of Economics, 13(2), 324–340.

Shapiro, C., & Stiglitz, J. E. (1984). Equilibrium unemployment as a worker discipline device. American Economic Review, 74(3), 433–444.

Olson, M. (1965). The logic of collective action: Public goods and the theory of groups. Harvard University Press.

Tullock, G. (1967). The welfare costs of tariffs, monopolies, and theft. Western Economic Journal, 5(3), 224–232.

— The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873. We checked whether a model could do our jobs. It could. It would also need a beer cellar, eventually, and nobody has trained one on that yet.

If the first half of this year has taught us anything, it is that understanding human behavior may be considerably more relaxing than continuing to pretend it makes sense. So here, instead of the usual list, are our recommended summer readings — six books that explain why the world is the way it is, followed by two that explains why none of this is new.

1. Start with the classics.

The Basic Laws of Human Stupidity — Carlo M. Cipolla (Doubleday, 2019; first written 1976, first published in Italian 1988)

A short, posthumously published essay built around five laws, the most famous of which states that a stupid person is one who causes harm to others while gaining nothing for themselves—making the stupid, by Cipolla’s accounting, more dangerous than the genuinely malicious, who at least extract a benefit proportional to the damage they cause. He plots the population on a graph of benefit-to-self against cost-to-others and finds the stupid distributed evenly across every class, profession, and era, immune to education, immune to correction, and consistently underestimated by everyone who has not yet been personally run over by one. No introduction needed. Every year arrives as a fresh empirical confirmation.

Reading time: two afternoons. Time spent afterward recognizing colleagues, politicians, and strangers on the internet inside the book: considerably longer.

2. Then move to ignorance.

Ignorance: A Global History — Peter Burke (Yale University Press, 2023)

A sweeping survey of how societies, institutions, and entire disciplines have managed not knowing things, deliberately, accidentally, and sometimes both at once. Burke ranges from medieval church censorship to twentieth-century intelligence failures, treating ignorance not as a gap waiting to be filled but as something actively produced—manufactured, defended, occasionally profitable. If Cipolla explains why individuals make catastrophically bad decisions, Burke explains why entire societies become remarkably good at producing the conditions for it. Some institutions specialize in knowledge. Others, with equal diligence, specialize in the opposite and rarely advertise which one they are.

3. Because Homer Simpson deserves tenure.

The Simpsons and Philosophy: The D’oh! of Homer — edited by William Irwin, Mark T. Conard, and Aeon J. Skoble (Open Court, 2001)

A collection of eighteen academic essays using individual Simpsons episodes to walk through ethics, epistemology, and political theory — Homer as a case study in Aristotelian virtue and its absence, Mr. Burns as an exercise in unchecked corporate power, Springfield’s town hall meetings as a small working model of incentive failure in public life. The premise sounds like a joke until you notice how cleanly the show’s twenty-five-minute plots have been mapping institutional dysfunction since 1989, for free, weekly, to an audience that mostly thought it was just watching cartoons.

4. Read history. Or convincingly pretend to.

History — Miles Jupp (Headline, 2021)

Not, despite the title, a work of history at all. It is a comic novel about Clive Hapgood, a history teacher at a struggling private school whose orderly, slightly disappointing life quietly comes apart over the course of a family holiday and one unresolved incident at work. The joke is gentler than Cipolla’s and the tragedy is realer than it first lets on—a man who teaches the subject for a living while demonstrating, scene by scene, how little practical use the discipline is when his own small history starts going wrong in real time.

5. Ask the question nobody wants asked at the department meeting.

What’s the Economy For, Anyway? — John de Graaf and David K. Batker (Bloomsbury Press, 2011)

The authors open by pointing out that GDP rose after the BP oil spill, because cleanup spending counts as economic activity regardless of what it is cleaning up, and use that one uncomfortable fact to argue that we have spent decades optimizing a number that was never actually built to measure whether anyone’s life improved. Drawing on Gifford Pinchot’s century-old formula — the greatest good for the greatest number, over the longest run — they walk through health care, work hours, and inequality, arguing that growth and well-being quietly stopped moving together some decades ago and nobody updated the dashboard. Whether or not you agree with their prescriptions, the diagnostic question underneath the book is worth sitting with longer than the read itself takes: what, exactly, are we trying to optimize?

6. Celebrate the forgotten.

The Greatest Nobodies of History — Adrian Bliss (Century / Ballantine Books, 2024)

Ten comic, loosely fictionalized profiles of the people standing just outside the frame of the famous portrait—Leonardo da Vinci’s ferret, the servant who oversaw Henry VIII’s toilet arrangements, the horse that carried a king’s son past Cromwell’s forces. Each chapter ends with a short note separating what is documented from what Bliss invented for the joke, which is itself a small admission about how much of popular history runs on the same ratio. History is mostly assembled by people nobody remembers, doing unremarkable things at the right moment. The famous ones simply received better lighting, and considerably more credit than the moment deserved to assign to any one figure.

7. Something serious, for the walk back from the beach.

1873: The Rothschilds, the First Great Depression, and the Making of the Modern World — Liaquat Ahamed (Hutchinson Heinemann, 2026)

Ahamed, who already won a Pulitzer for Lords of Finance and apparently decided one global financial collapse per career wasn’t enough, turns here to an earlier, less famous calamity: the Vienna stock crash of May 1873, which spread within months to Berlin, New York, Paris, and London with the efficiency financial contagion always manages. It triggered decades of deflation and helped end Reconstruction in the United States—outcomes nobody at the original party in Vienna was toasting to. The Rothschilds sit at the center less as villains than as a case study in restraint, scapegoated for a bust they had stayed conservative enough to avoid causing.

What makes the book worth your remaining beach days is quieter than any of that: the people living through 1873 almost never recognized which events would matter. Most believed they were simply enduring another bad year of falling prices and failed railroads.

Historians credit 1873 as the start of the Long Depression. We would gently correct the record: it was, far more importantly, the year the Invisible Band was founded. The Long Depression gets the textbooks. We get a beer cellar in New York and a founding member nobody can fully account for. History, as usual, picked the wrong headline.

And since we are already being serious, briefly, one more, because we are fans of Brad DeLong.

Slouching Towards Utopia: An Economic History of the Twentieth Century — J. Bradford DeLong (Basic Books, 2022)

DeLong’s six hundred pages contain, buried somewhere past page 300, the best two-sentence description of recent political economy we have found anywhere: Hayek was right that the market crowdsources solutions to the problems it is set. The trouble starts the moment that gets quietly extended into deciding which problems are worth setting in the first place. Nobody ever resolved the tension between the two halves of that sentence. The postwar settlement papered over it for a few decades, the patch failed around 2010, and the political temperature of the last ten years, on more than one continent, looks a great deal like what happens when people notice the patch failed and nobody got around to telling them.

It has been a remarkably strange year already. So take a few weeks off. Read something that makes you laugh. Read something that makes you think. Read some economic history, ideally in that order.

And if, somewhere around the second afternoon with Cipolla, you suddenly start recognizing half the people around you inside the book—congratulations. You are finally reading history correctly.

Please, at the very least, read something. Anything. We are not picky about which one. We are economic historians. We have spent our entire careers fighting for relevance against a screen that updates faster than any book ever could, and we would settle, this once, for losing that fight to a paperback instead.

— The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873. The reading list is the same length every year. The recognition gets faster.

We found, in our archive, an irrational episode worth filing properly: in 2022, a shirt sold at Sotheby’s for £7,142,500 — the highest price ever paid for an item of clothing worn during a sporting event, beating the previous record held by a Babe Ruth jersey by nearly three million pounds. The shirt belonged to Diego Maradona, worn during the 1986 World Cup quarter-final against England, the match in which he scored the “Hand of God” goal and, minutes later, the goal voted the greatest in World Cup history. English midfielder Steve Hodge had swapped jerseys with him in the tunnel afterward and held onto it for thirty-six years before deciding to sell.

We are not, ordinarily, in the business of begrudging anyone’s auction results. What interests us is a detail that surfaced only years later, and that complicates the entire premise of the sale in a way nobody at Sotheby’s seems to have advertised: the shirt was never officially manufactured at all.

A logistics failure, solved by Tepito.

Argentina’s official kit supply for that tournament ran short after players swapped jerseys following an earlier match, a customary ritual the schedule had not accounted for. England’s match required Argentina to wear its blue alternate kit again, and there simply were not enough of them. The team’s backup goalkeeper, Héctor Zelada, who also played for Club América and knew Mexico City well beyond its hotels and stadiums, had a solution: Tepito, the dense, improvisational market neighborhood in the heart of the capital where, as the saying goes, everything is sold and anything is possible.

Staff were sent in, came back with several unofficial blue jerseys—lighter, more breathable, better suited to playing at altitude under the midday sun than the heavy official kit—and the team spent the hours before kickoff ironing on numbers and hand-sewing the Argentine federation crest onto shirts that had been purchased, for a few pesos each, in a market better known for counterfeit electronics and questionable goods of every description. Nobody watching the broadcast noticed. The shade of blue was slightly off. The numbers caught the light differently. None of it mattered once Maradona started playing.

We have in our archive a considerable stack of newspaper clippings on this exact event, including photographs, among them one from the Associated Press that appears to have been trying, in real time, to testify to what it was looking at.

The shirt that later sold for £7.1 million was, by any formal definition of the term, a counterfeit. It met no manufacturing standard, carried no licensing agreement, and was assembled by hand in a hotel room hours before kickoff. The market did not know this when it set the price. We are not certain it would have changed the price if it had.

Why the market did not care, and arguably could not have.

There is a polite term for what happened at that auction, and credit where it’s due: an economist named Herbert Simon spent a fair bit of his career working out why people make decisions this way, and won a Nobel for the trouble. The basic idea, stripped of the academic varnish, is that nobody has the time or the interest to verify something before paying for it, so we use a shortcut instead: official-looking equals authentic, authentic equals valuable, sign here. Pawn Stars would not have made this mistake. The cast of that show will not buy a guitar for a few hundred dollars without calling in a specialist to inspect the serial number on camera. Nobody called in a specialist for the seven-figure shirt. Seven million pounds is not a shortcut gently economizing on cognitive effort, though. At some point the careful, peer-reviewed language built to explain why a reasonable person settles for “good enough” instead of “optimal” starts to look less like an explanation and more like a euphemism so nobody has to write the word “madness” in a journal. We would propose our own amendment anyway: in this sport, rationality is not so much bounded as rounded—worn smooth at the edges, rolling wherever the last confident kick sent it, occasionally bouncing off a goalpost and going in.

Football shirts are a small corner of a much larger counterfeit economy, sold openly in markets not unlike Tepito, bought by people who know exactly what they are getting: looks right, costs less, gets worn to the same matches. Authenticity is not one market. It is at least two, running in parallel, and the official one spent £7.1 million without checking whether its own prize exhibit qualified.

Economic history has a recurring genre: an institution’s most confidently priced asset turns out, on closer inspection, to never have been what the price assumed it was. Tulip bulbs. Railway stocks. Several thousand IPOs. And now the single most expensive piece of sports clothing ever sold, hand-stitched in a hotel room from goods bought in a flea market because the official supply chain had failed to do its job.

If there have been absurdities and outright stupidities scattered across the whole of economic history, football may be the discipline’s purest surviving specimen: a multi-billion-dollar global industry that managed, in its single most lucrative collectible transaction on record, to sell the world a counterfeit at auction prices and have nobody notice for the better part of a decade.

The stadium where it happened is still standing, freshly renovated, back in the rotation. The Azteca—now the Banorte Stadium under a new sponsorship deal—is hosting matches in this very World Cup. The pitch has been replaced. The sponsor’s name is new. The jersey market has not changed its underlying logic at all.

None of this, to be fair to everyone involved, was any single person being foolish. The family buying the cheap replica was being perfectly sensible. The auction house was being perfectly sensible, by the standards auction houses use. Even Zelada’s last-minute Tepito run was the most sensible thing anyone did that entire week. The irrationality, if the word applies anywhere, belongs to football itself—the whole apparatus, every kit deal and counterfeit and seven-figure hammer price of it, somehow staying perfectly rational in each individual transaction while adding up, decade after decade, to something no sane institution would have designed on purpose.

References

Simon, H. A. (1955). A behavioral model of rational choice. The Quarterly Journal of Economics, 69(1), 99–118.

Sky News. (2022, May 4). Diego Maradona’s “Hand of God” shirt sells for more than £7m at auction.

Sánchez, B. (2026, March 27). How Argentina bought Tepito jerseys that helped them beat England in the 1986 World Cup. InformaBTL.

— The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873. We checked the label. There wasn’t one.

Freedonia was fictional. The room we want to talk about is not, and it has been testing the difference between the two kinds of statecraft for over a century. You can admire its chandeliers, which are genuinely impressive, and its ceiling paintings, which celebrate military victories with the specific confidence of a country that was, at the time of painting, between catastrophes. What you cannot do, if you have read anything at all about what has been signed in that room, is stand in it without a certain involuntary tightening of the chest.

Unless, of course, you have not read anything. In which case it is just a very large room with excellent lighting. Perfect for a photo.

What the room is.

The Hall of Mirrors has hosted two signings whose consequences took decades to play out. The first, in January 1871, was the proclamation of the German Empire—conducted in the heart of the French palace, on French soil, in the middle of a French military defeat, to ensure the humiliation was complete and properly documented. Otto von Bismarck had the keys. He let himself in and made his point, in front of someone else’s mirrors, on someone else’s floor.

The second, in June 1919, was the Treaty of Versailles. The victors of the First World War imposed the terms of German defeat in the same room where German victory had been proclaimed forty-eight years earlier. The reparations were punitive (Treaty of Versailles, 1919, Part VIII, Articles 231–247). The war guilt clause was absolute. John Maynard Keynes, advising the British delegation, warned that this was revenge wearing a treaty as a costume, and that revenge, historically speaking, always comes back for seconds. He resigned, went home, and wrote The Economic Consequences of the Peace (Keynes, 1919), which predicted with uncomfortable precision what would happen next. What happened next took twenty years, fifty million dead, and a second world war to resolve.

The financial obligations created by that document were not fully discharged until October 2010 (Eddy, 2010)—ninety-one years later, three wars later, two Germanys later, long after every signatory had died. The archive has the payment receipts. They are filed under longer than expected.

On the most recent signing.

The most recent party to sign something in that room walked in with excellent television coverage, a very large tie, and no apparent awareness that the last two major agreements concluded in that location had both, in different ways and on different timescales, produced outcomes that required significant subsequent revision. The other parties present had, we suspect, read more of the file than the person holding the pen.

There is, fittingly, a mirror scene in Duck Soup too. Harpo, dressed as Groucho, breaks a mirror and then becomes it—mimicking every gesture closely enough that the deception holds, that the reflection looks correct, that nobody checks too carefully because the surface is doing exactly what a mirror is supposed to do. The Hall of Mirrors has been performing the same trick for considerably longer, and at considerably higher stakes. It reflects nothing on its own. It reflects whatever the room’s occupants need it to confirm.

Someone will declare victory. Someone usually does. In 1919 the victors used that room to extract reparations from the defeated. If what gets signed this time is closer to reconstruction aid than to punishment, the choice of venue becomes genuinely strange—Versailles is, historically speaking, the wrong room for generosity. It is the room built for the other thing. Whoever picked it either did not know this, or knew it and signed anyway, which is its own kind of statement.

The phrase “new Marshall Plan” has been applied to nearly every economic aid package since 1948, usually by people who have not reread the original terms. The Marshall Plan was, among other things, the explicit correction to Versailles (Marshall, 1947): a power that looked at what punishment had produced in 1919 and chose reconstruction instead. It is possibly the only documented case of someone reading the archive before acting on it.

We have no evidence the reading has resumed. We do have the phrase again, circulating freely, untroubled by the original.

Greece asked for a Marshall Plan in 2010 and received a troika instead. The phrase may have quietly inverted its meaning somewhere along the way, so that what it now describes is, in practice, close to the opposite of what it described in 1948. Freedonia, at least, only had to deal with a pie fight.

On rooms that remember, and how Duck Soup actually ends.

We are not the type to believe in omens. You already know this. We believe in archives, which are omens that have already happened and were simply filed before anyone got around to reading them. Versailles has always been here.

For the record: the war in Duck Soup does not end with a treaty. It ends with everyone in the room—Freedonians, Sylvanians, several bystanders, Groucho—picking up whatever fruit is nearest and throwing it at the person who started the confusion in the first place. Nobody negotiates. Nobody signs anything. Several economic historians have noted this is a more accurate model of how most disputes actually resolve than anything taught in a seminar on conflict mediation.

We do not know how this one ends. We know how the others did, and we know the room has, on at least one occasion, taken ninety-one years to find out. We will be here either way. We always are.

References

Keynes, J. M. (1919). The Economic Consequences of the Peace. Macmillan.

Treaty of Versailles, June 28, 1919. Reparations terms, Part VIII, Articles 231–247.

Eddy, M. (2010, October 3). For Germany, WWI finally ends on Sunday. NBC News / Reuters.

Marshall, G. C. (1947, June 5). Address at Harvard University [The Marshall Plan speech].

McCarey, L. (Director). (1933). Duck Soup [Film]. Paramount Pictures.

— The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873. The room has a memory. The chandeliers photograph well. We recommend reading the plaque before signing anything.

We are not sure what to make of this. We have been sitting with it for some time, in the way that economic historians sit with data points that do not immediately resolve into comfortable conclusions. Several explanations are available. Students may have become substantially more capable. Teaching may have improved dramatically. Assessment may have become more accurately calibrated to actual competence.

These explanations are not impossible. They are, however, in tension with certain other data points—employer satisfaction surveys, the rising demand for postgraduate credentials, and the kind of informal longitudinal evidence that accumulates in the gradebooks of people who have been setting the same examinations for a decade.

We teach. We would like to believe — and we do, in the specific way that people believe things they cannot verify and prefer not to examine too closely — that the knowledge we transmit lands somewhere. That something we said in October is still, in some degraded but recoverable form, lodged in a student’s mind by May. That the process has an output beyond the credential.

Our RateMyProfessors reviews suggest otherwise. We have read them. We have made our peace with them. We have returned to the archive, which does not give star ratings and cannot be appealed.

We have, however, been setting examinations long enough to maintain, informally and without institutional sanction, a longitudinal dataset of our own. Every spring, as reliably as the season itself, we find out. The questions have not changed substantially. The expected word count has. The quality of the answers has. Not dramatically, not uniformly, not in ways that would survive peer review—but in the specific, cumulative, unmistakable way that anyone who has been reading student work for a decade recognizes immediately and discusses only in the corridor, never in the committee meeting, because the committee meeting has other priorities.

Apparently we are terrible professors. The alternative explanation is structural. We prefer the structural explanation, though we acknowledge this may not be entirely objective.

For our own peace of mind, we also blame, in no particular order: the smartphone, the algorithm that has optimized the human attention span for content that does not require finishing, artificial intelligence that can produce a competent paragraph on any subject without having read anything, and the emerging adult who arrives at university having been told, repeatedly and by multiple institutions, that their feelings about a topic are as valid as the evidence about it. We note that none of these are the student’s fault either. We note this with diminishing conviction as May approaches.

This is, in the language of economics, the law of diminishing returns applied to mass credentialing. Each additional unit of input — each additional cohort, each additional institution granted degree-awarding status by the 2017 Higher Education and Research Act — produces marginally less of the thing the input was supposed to produce. Not because the students are less capable. Because the process has a capacity, and policy has been systematically exceeding it while measuring outputs in credentials rather than in the thing credentials are supposed to certify.

The credential count goes up. The exam answers get shorter. The archive, as usual, contains both.

Let us begin with a person.

He is twenty-two years old. He has just graduated with a first-class degree. He applies for a job. The job requires a degree. He has a degree. He applies.

He is rejected. The successful candidate has a master’s degree.

He goes back to university. He pays another fifteen thousand pounds. He gets the job.

The hiring manager who wrote the job advertisement requiring a master’s degree does not have a master’s degree. He was hired in 1987. A handshake and a 2:1 was considered sufficient. He has worked at the company for thirty-seven years. He is, by most measurable indicators, competent.

Nobody in this story has done anything wrong. Everyone in this story is responding rationally to the incentive structure they are inside. The incentive structure is completely, magnificently, historically irrational. It has been before. The archive has the documentation.

The Coin That Says One Shilling

When a government issues too much currency relative to the goods available, each unit buys less. The nominal value stays the same — the coin still says one shilling — but the purchasing power declines. Gresham’s Law: bad money drives out good (MacLeod, 1858). People hoard the high-quality coins and spend the debased ones, which accelerates the debasement, until the coin is worth the metal it contains minus the cost of melting it down—the monetary equivalent of a degree classification that tells the employer nothing except that the holder attended an institution and paid the fees.

The nominal value of a first-class degree did not change. It still says first class on the transcript. The purchasing power has declined. The Roman economy’s response to currency debasement was instructive: unable to trust the coin, people moved to payment in kind. Unable to trust the degree, employers have moved to payment in kind—assessment centres, unpaid internships, personal networks, the quiet restoration of class-based selection under a vocabulary of meritocracy.

Several Fragments of the True Cross

By the fifteenth century there were enough fragments of the True Cross in European churches to build a small galleon. Several institutions claimed to possess the foreskin of Christ, of which there were, theologically speaking, a finite number. John Calvin estimated that if all the relics attributed to the apostles were genuine, each apostle would have required approximately four bodies (Calvin, 1543/2009).

The degree is the relic. The university is the cathedral. The employer is the pilgrim. The fiction is that a first-class degree certifies a comparable level of intellectual achievement across all institutions. Everyone involved knows the numbers do not add up. The fiction persists because the institution that dismantles it unilaterally loses students, falls in rankings, and receives a strongly worded communication from its vice-chancellor about the importance of remaining competitive.

Calvin is unavailable for comment. The Office for Students has published a report (Office for Students, 2024).

The 2017 Higher Education and Research Act lowered barriers to degree-awarding status on the grounds that competition improves quality (Higher Education and Research Act, 2017). More providers competing for students on the basis of satisfaction creates more downward pressure on standards, not less. The relic market expanded. The cathedrals needed the revenue. The pilgrims kept arriving. This is not a conspiracy. It is an incentive structure. The distinction matters less than one might hope.

Back to the Man With the Master’s Degree

He is now twenty-four. He has spent five years in higher education. He has paid approximately fifty-five thousand pounds. He has a job that pays twenty-eight thousand pounds per year — enough to live in a city if he does not own a car, does not eat out, and approaches housing with the creative flexibility of someone who has accepted that ownership applies to other generations.

In 1987, his hiring manager walked into the same job with a 2:1, no debt, and a vague plan. He is thinking about retiring to Portugal.

The currency was debased incrementally, each institution responding rationally to competitive pressure created by every other institution’s rational response to competitive pressure. The relic market expanded because pilgrims kept arriving. The lemons drove out the peaches because the buyers stopped trying to tell them apart.

The solution, in the currency case, is to restore the silver content or issue a new currency. In the relic case, it was the Reformation, which took a century. In the degree case, it is presumably something between the two, though nobody currently in a position to implement it has specified what—possibly because the specification would require visiting an archive, and the archive closes at four-thirty.

References

Calvin, J. (1543/2009). Treatise on relics. Reformation Heritage Books.

Higher Education and Research Act 2017, c. 29. UK Public General Acts.

HESA. (2023). Higher education student statistics. Higher Education Statistics Agency.

MacLeod, H. D. (1858). The theory and practice of banking (Vol. 1). Longman.

Office for Students. (2024). Analysis of degree classifications over time: Changes in graduate attainment from 2010-11 to 2023-24. Office for Students.

Universities UK. (2022). Higher education in facts and figures 2022. Universities UK.

— The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873.

Currently grading. All submissions awarded a first. The Reformation is not scheduled until next semester.

500mg Reform Capsules. Proven benefits: privatization, liberalization, stabilization, market discipline, confidence. Thirty capsules for oral administration. Because the future can’t wait.™

Side effects may include: oligarchs, asset stripping, voucher accumulation, democratic disappointment, persistent nostalgia, institutional vacuum, former state assets in unfamiliar hands, a sudden interest in offshore banking, and economists explaining that the alternative would have been worse.

If symptoms persist, administer additional shock.

Institutional quality sold separately.

Originally indicated for the treatment of mild inflation in Chile, 1973. Approved for broader use following encouraging early results.

The early results were considered encouraging by a specific subset of the people involved.

Long-term effects were not studied at the time. The study period was, for a number of reasons, brief.

We should note, for full transparency, that Shock Therapy® is a repurposed product. This is standard pharmaceutical practice — a compound developed for one indication, found to produce encouraging results in early trials, approved for broader application without the kind of extended testing that the original patient population might, in retrospect, have appreciated. The repurposing was efficient. The approval process was fast. The active ingredient, it was determined, was universal. Institutions, history, and context were considered complicating variables rather than active ingredients.

Not recommended for economies without strong institutions was added to the label later.

Several economies learned this from the label rather than from the consultation.

We have read the label carefully. We have also read the archive. These are not the same document.

The prescription arrives with confidence. It always does. Confidence is, in fact, listed as a proven benefit—right there between market discipline and the family photo. The family is smiling. They are always smiling on the label. They have presumably never met the adjustment period, which is not pictured, does not smile, and arrives approximately eighteen months after the label does.

We note that the recommended dosage is thirty capsules. We note that no transitional economy in the documented historical record has completed the course in thirty capsules. We note that the instructions for what to do after the thirty capsules are not included in the packaging, possibly because they would not fit, possibly because they were not written.

Please consult your economist before use. Not recommended for economies without strong institutions.

We note that strong institutions are precisely what transitional economies tend to lack. We note further that this caveat appears in small print, below if symptoms persist, administer additional shock, which raises certain questions about the recommended treatment protocol that we are not qualified to answer and an economist apparently was.

Results may vary.

This is doing considerable work as a sentence.

Field-tested in several transitional economies. Results remain under discussion.

They have been under discussion for thirty years. Several of the economists involved have written memoirs. The memoirs are in the archive. The transitional economies are also, in various conditions, still there.

Economic reform is not one-size-fits-all.

Correct. It says so on the label. Right next to the logo of an institution that spent a considerable portion of the 1990s insisting otherwise. We admire the disclaimer. We wish it had been on the original label.

The original label was different.

— The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873. Documenting the past. Questioning the prescriptions. The archive contains the memoirs. And the mortality statistics.

One. The archive contains what actually happened.

Not what the press release said happened. Not what the model predicted would happen. Not what the consensus, formed without reference to the primary sources, has concluded happened. The archive contains the handwriting of the people who were there, produced before they knew how the story ended, preserving the decisions, the hesitations, the calculations, and occasionally the embarrassing internal correspondence that institutions would have preferred not to survive.

It survived. It is in the archive. Someone should read it.

Two. The archive is not only a building.

The historical record lives wherever someone had the discipline to write things down and the conditions to have it survive. The purpose-built repository with the climate-controlled reading room is one version. The basement of the former cooperative president is another. The factory worker’s production notebook. The secretary of a mutual aid association who kept the minutes in a language the host country did not speak because she understood that if she did not, nobody would.

These documents were not written for historians. They were written because someone needed to write them. They are, for that reason, frequently more honest than the official archive—which was written, at least in part, for people who might one day read it. The gap between the official archive and the basement archive is often where the real history lives—the history of what actually happened rather than what the institution reported had happened, which are sometimes the same thing and sometimes are not.

The historian who goes only to the purpose-built repository is reading the version of the past that institutions wanted preserved. The historian who follows the trail to the basement, the attic, the former association secretary’s filing cabinet, is reading something closer to what occurred. Both are necessary. The second is harder to find and considerably more interesting.

Consider: in Star Wars: Episode II—Attack of the Clones, Obi-Wan Kenobi goes to the Jedi Archives—the galaxy’s greatest repository of knowledge, 25,000 years of recorded history across millions of worlds—and discovers that the planet Kamino does not exist. The Chief Librarian informs him, with the specific confidence of someone who has never been wrong before: “If an item does not appear in our records, it does not exist.”

The Jedi Archives. Star Wars: Episode II — Attack of the Clones (George Lucas, 2002). In this scene, Obi-Wan Kenobi discovers that the planet Kamino has been erased from the Jedi Archives — the galaxy’s most comprehensive repository of knowledge, maintained across 25,000 years of galactic civilization. Madame Jocasta Nu, Chief Librarian: “If an item does not appear in our records, it does not exist.” Kamino exists. It has been erased. We note this without comment. We also note that Madame Jocasta Nu is, in our view, one of the most underappreciated figures in the defense of institutional knowledge in the history of science fiction. She was right about everything except the one thing that mattered. This is the condition of the archivist. We recognize it professionally.

Photo credit: Lucasfilm Ltd. / Walt Disney Pictures. Used for educational and commentary purposes.

She is wrong. Kamino exists. It simply does not appear in the official archive.

Obi-Wan finds it anyway. Not in the Jedi Archives, but by asking an elderly colleague who remembered hearing about it somewhere, which is the oldest form of historical research and considerably more reliable than assuming the official record is complete. The absence of something from the official archive is not evidence that it does not exist. It is evidence that whoever maintained the official archive did not record it, could not record it, or preferred not to. The planet is still there. The company still made the decision. The workers still organized. The record lives somewhere — in a basement, in a notebook, in the memory of an elderly colleague who happened to be paying attention.

Go to the archives. Then go to the basement.

Three. The silence is also evidence.

What was not recorded matters as much as what was. Who kept records and who did not. What was considered worth writing down and what was not. The gap between the official account and the basement account is itself a historical finding. Reading the archive means reading both what is there and what is missing, and having a view about why.

Four. The model does not replace the document.

The dataset is not the archive. One tells you what was recorded. The other tells you what was recorded, why, and what was deliberately left out. Treating them as equivalent produces history that looks complete and is not — conclusions built on sources that were not consulted, patterns extracted from records that do not contain the full picture. The archive corrects. This is what archives do.

We agree. We have been agreeing since 1873. Nobody asked. We have been in the archive, which may explain why nobody knew to ask us.

Five. The archive closes at four-thirty.

This is not a metaphor. It is a logistical fact with implications for research planning that are routinely underestimated by people who have not spent time in archives. The building closes. The finding aid ends. The box you need is at a different repository. The document you were looking for turns out to be in a language you did not expect. Budget for this. The history that gets written is, in part, determined by who was willing to account for the four-thirty problem and who was not.

Six. The incentive structures are wrong.

Archival research takes time. Time does not fit efficiently into a publication cycle. The finding aid does not generate a citation. The month spent in a repository does not appear on the CV in a form that promotion committees find legible. This is a structural problem, not a personal one. The scholars who do not go to the archive are not lazy. They are responding rationally to an incentive structure that has made the archive optional. The consequences of treating the archive as optional are in the archive, available to anyone willing to go.

Seven. The people who made the decisions left records.

The records are still there. In repositories. In basements. In cardboard boxes donated to county historical societies by the children of people who understood, without being told, that the documents mattered. Someone is responsible for reading them.

We are that someone. We are aware this has not been sufficiently communicated. We are working on it. In the archive, mostly, which is part of the problem.

We are not asking for a revolution. We are not even asking to be consulted before the crisis, though we note that this would have been useful on several documented occasions. We are asking for a room, a finding aid, a pencil, and the willingness to spend time with documents that were not written for you, in a building that closes at four-thirty.

Or a basement. Or an attic. Or a cardboard box in a county historical society that nobody has opened since 1987.

The past is not going to explain itself. It left records. Someone has to read them.

The archive is open on Tuesdays and Thursdays.

We will be there. We are always there. This is, in retrospect, part of the problem.

— The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873.

The archive is open. We checked.

A product label. Soviet constructivist aesthetic. Hammer and sickle. Worker raising a bottle toward a horizon of smokestacks and historical inevitability. The label reads, in Russian and English: Красный Жгучий. Red Hot Sauce.

All the heat. None of the bourgeoisie.

We sat with it for a while.

The USSR produced some of the most visually compelling propaganda in the history of human communication. The posters were extraordinary. The typography was bold. The worker always faced the light with the specific confidence of someone who had not yet seen the food supply.

The cuisine, meanwhile, was boiled.

The official cookbook—The Book of Tasty and Healthy Food (Книга о вкусной и здоровой пище), first published in 1939—promised tasty food. The title said so. We have questions.

Hedrick Smith, an American journalist who lived in Moscow in the early 1970s, documented Soviet citizens queuing for hours for food that was, when it arrived, grey (Smith, 1976). We note, in the spirit of fairness, that some portion of the population presumably found it adequate. They have not left reviews.

The five-year plan could account for steel, electricity, and calories. It had no column for flavor. This was not an oversight.

Flavor, it turns out, is a market phenomenon. You cannot plan it. You cannot approve it by committee. You cannot put it in a five-year plan and expect it to arrive on schedule.

We should add, for the sake of honesty: de gustibus non est disputandum. There is no disputing taste. Some people, presumably, enjoyed the cabbage.

References

Gronow, J. (2003). Caviar with champagne: Common luxury and the ideals of the good life in Stalin’s Russia. Berg Publishers.

Министерство пищевой промышленности СССР [Ministry of Food Industry of the USSR]. Molchanova, O. P., Lobanov, D. I., Lifshits, M. O., & Tsyplenkov, N. P. (1952). Книга о вкусной и здоровой пище [The book of tasty and healthy food]. Pishchepromizdat.

Smith, H. (1976). The Russians. Times Books.

— The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873.

The archive smells of old paper. The sauce smells considerably better.

We teach. Not brilliantly, not always, not to standing ovations — we checked RateMyProfessors.com once and have not been back. But we teach with the specific conviction of people who believe, against considerable evidence, that what happens in a classroom still matters.

And every semester, in every classroom, we make the same argument.

We say: read.

Not a specific thing necessarily. Not the syllabus, although the syllabus would be a reasonable start. Read whatever you want. Read widely. Read things that have nothing to do with your degree. Read things that make you uncomfortable. Read things written by people who died before your grandparents were born. Read things that were considered dangerous when they were published. Read.

And then—because we are economic historians and we understand that abstract arguments require concrete evidence—we make the case.

We say: Do you know the people who run this planet? The ones who actually make decisions, move money, write laws, shape the conditions under which everyone else lives? The ones who, whatever you think of them, are undeniably operating at a different level of institutional power than the rest of us?

They read more than everyone else. They always have.

It is not a perfect argument. We know this. But it is a true one, and we make it with the enthusiasm of people who have the receipts.

And then—every semester, without fail, with the reliability of a pre-war commodity cycle—a student raises their hand. A good student. Usually one of the better ones. The kind who has been listening carefully, which you can tell because they waited until the argument was complete before dismantling it.

And they ask:

“Professor—really?”

Sometimes it comes with elaboration. But the richest man on the planet doesn’t seem to read that much. Or: my cousin dropped out of university and started a company. Or, most devastatingly: wasn’t the man who changed personal computing famously not that interested in books?

Sometimes it is just the two words, delivered with the patient skepticism of someone who has been told things by authority figures before and has begun to notice the gap between what authority figures say and what the evidence supports.

Really?

And here is the thing — the thing we have been thinking about for years, the thing that gets harder to answer every semester as the counterexamples multiply and the attention economy makes its case more loudly than we can make ours:

The student is not wrong to ask.

The argument we make—that the powerful read more—is true in the aggregate and misleading in the particular. It is the kind of statement that survives statistical scrutiny and collapses under individual counterexamples. Yes, on average, across the historical record, the people who have exercised the most consequential institutional power have tended to be people who read extensively. Lincoln read. Churchill read—compulsively, obsessively, in ways that shaped how he thought and wrote and persuaded. Obama read. The founders of the American republic read each other’s work and the works of Locke, Montesquieu, and Cicero with an attention that would embarrass most modern graduate students.

But.

The student raising their hand is not asking about the aggregate. They are asking about now. About 2026. About a world in which the largest fortunes were built not on reading but on engineering, risk tolerance, and an attention span calibrated to the news cycle. About a world in which the fastest-growing fortunes are made by people who move quickly, who trust algorithms over archives, who have optimized for speed in ways that sustained reading definitively does not support.

The student is asking: has the model changed?

And we do not have a clean answer. We have a historian’s answer, which is considerably less satisfying.

Here is what we actually believe, stated without the rhetorical scaffolding we use in the classroom:

Reading does not make you powerful. It makes you capable of understanding complexity—which is a different thing and a less immediately rewarding one. The person who reads widely develops a longer time horizon, a higher tolerance for ambiguity, and a better-calibrated sense of how things have gone wrong before and why. These are not qualities that the current market rewards efficiently. The market rewards speed, confidence, and the willingness to act on incomplete information. Reading, done properly, tends to produce people who are aware of how incomplete their information is.

This is, depending on how you look at it, either the argument for reading or the argument against it as a career strategy.

The student who asks really? is performing, in that moment, exactly the kind of critical thinking we spent the previous forty minutes trying to encourage. They heard an argument. They evaluated it against their own observation. They found the gap. They named it.

We should, technically, be delighted.

We are, technically, delighted.

We are also, privately, unsettled—because the question they are asking is one we ask ourselves. Not about whether reading matters. We have resolved that question to our own satisfaction and intend to die defending the answer. But about whether the world we are describing to them—the world in which reading is infrastructure for power, in which the archive is the foundation of understanding, in which the long run is the only honest unit of analysis—is the world they are actually going to inhabit.

Or whether the world they are going to inhabit is one in which the premium markup is on the thing that looks like knowledge rather than the thing that is it.

We are, in this sense, the last monks. Not in the religious sense—our relationship with institutional faith is complicated, and several of us have strong opinions about the Council of Trent. But in the functional sense: people who understand that knowledge requires active maintenance, that it does not preserve itself, that someone has to sit in the room and do the slow, unrewarded work of keeping it alive.

The medieval university lecture existed because books were too expensive to own. The professor stood at the front of the room and read aloud from the single copy of the text while students copied by hand. The lecture was not a teaching method. It was a distribution technology. A way of reproducing knowledge in a world where reproduction was impossible at scale.

The books are now cheaper than a coffee. The students are looking at their phones. The technology changed. The problem did not.

We have nothing against innovation in education. Innovation is fine. We have watched seventeen rounds of it with the patient attention of people who study long-run processes for a living.

Active learning. Flipped classrooms. Gamification. Competency frameworks. Learning outcomes matrices. Each arrived with a conference, a methodology, and the specific confidence of people who had not attended the previous conference. Each was assessed, revised, quietly shelved, and replaced by the next one.

We are not saying none of it worked. We are saying the students still cannot write a paragraph, which is the kind of data point that tends to survive methodological revisions.

We do not resolve this in the classroom. We return to the syllabus.

Somewhere, a student is watching a sixty-second summary of the Great Depression.

It has 2.4 million views.

It contains three factual errors and no sources.

We have written a rebuttal. It is in peer review. Expected publication: eighteen months.

— The Invisible Band of Economic Historians

Arriving late, as scheduled, since 1873. Currently teaching. Rated: sufficient.

Read. Just read. For the love of everything that has ever mattered, read.

Look at the photograph before continuing.

Photo credit: Anonymous Field Operative, Membership No. 5609 Paris, France · May 2026 iPhone. No filter. No irony intended. All irony unavoidable.

The plaque reads, in French:

En ce lieu résida Thomas Jefferson Ministre des États-Unis en France · 1785–1789 Président des États-Unis · 1801–1809 Auteur de la Déclaration de l’Indépendance Américaine Fondateur de l’Université de Virginie

Cette plaque a été apposée le 13 avril 1919 par les soins des anciens élèves de l’Université de Virginie, soldats de la Guerre Mondiale, en commémoration du centenaire anniversaire de la fondation de l’Université.

Translation:

In this place resided Thomas Jefferson, Minister of the United States in France, 1785–1789. President of the United States, 1801–1809. Author of the American Declaration of Independence. Founder of the University of Virginia. This plaque was placed on the 13th of April 1919 by former students of the University of Virginia, soldiers of the World War, in commemoration of the centenary anniversary of the founding of the University.

Men who had just survived a world war crossed the Atlantic, found the building, and put marble on the wall so the world would not forget.

They did their part.

Now look above the plaque.

A WeWork sign.

For the uninitiated: WeWork purchased long-term office leases, subdivided them into short-term desk rentals, called the result a community of creators, and charged monthly for the privilege of working in a room with exposed brick and a kombucha tap. Valued at forty-seven billion dollars at its peak. Filed for bankruptcy in 2023. Currently restructuring — which is the corporate equivalent of the archive being temporarily closed for reorganization.

An economic historian could have seen this coming. Several did. Nobody called.

Jefferson spent four years at this address. He arrived in 1784 as American Minister to France — a title he refused to call Ambassador, on the grounds that nobody could replace Benjamin Franklin, only succeed him, which remains one of the more elegant things anyone has ever said upon starting a new job.

He left in 1789. The year France ran out of grain, institutional capacity, and patience, in roughly that order.

During those four years, between the wine — his accounts suggest the quantity was considerable, his correspondence suggests it was necessary — he developed the argument that would define him: that free people require economic independence to think and act freely. That the man with a debt to renew, a subscription to maintain, a landlord to answer to, is not fully free. He is managed. He is, in the language of the twenty-first century, a user.

Jefferson imagined a republic of owners.

The building where he worked this out is now rented by the month.

We are not given to sentimentality. We know Jefferson’s contradictions better than most and with more footnotes than is comfortable at dinner. The man who wrote all men are created equal owned over six hundred human beings. His vision of independent proprietorship was explicitly, violently bounded by race. The agrarian republic he imagined was assembled from land taken from people nobody asked.

History is not simple. That is why it requires study rather than decoration.

And this photograph is, above all else, a study in decoration. The Champs-Élysées receives millions of visitors a year. Every surface has been photographed from every angle. Yet on that wall, in plain sight — not in an archive, not behind a restricted access form, not in a box mislabeled since 1954, but in plain sight on the most photographed street in Europe — sits a marble record of one of the formative minds of modern democratic thought.

Largely unnoticed.

Below a sign for a bankrupt subscription office.

This is not about forgetting. Forgetting is recoverable — that is what archives are for, and we will defend them until our last breath and our last pencil. This is about something harder to fix: the condition in which the information is present, the plaque is legible, the history is available to anyone who stops for thirty seconds — and the attention is permanently elsewhere.

Jefferson’s central argument was that self-government requires citizens capable of sustained attention. Of reading. Of resisting the immediate and considering the long run. Of owning something — land, thought, time — that could not be taken away by a landlord or restructured by a creditor.

Two centuries later we have built, directly above his front door, an industry whose entire business model depends on people owning nothing, committing to nothing beyond the current billing cycle, and moving on when something cheaper becomes available.

From land ownership to monthly subscriptions. From self-reliance to hot desks. From republican virtue to premium membership tiers.

The plaque is still there. The memory is negotiable. The membership, however, renews automatically.

We thank Membership No. 5609 for the field report. Filed promptly. Published eventually. The archive would have had it catalogued by Tuesday.

— The Invisible Band of Economic Historians Arriving late, as scheduled, since 1873.

Hello. You have reached The Invisible Band of Economic Historians.

If you are hearing from us, congratulations: things have gone terribly wrong.

We know. We know because things always go terribly wrong. That is, in fact, the central finding of our discipline — distilled from centuries of archives, ledgers, harvest records, bankruptcy filings, and the collected correspondence of people who also thought they were living through an exceptional moment. They were not. Neither are you. Welcome to the pattern.

We are the ones who always arrive late to the party. Not because we are disorganized — we are, but that is beside the point — but because the party, historically speaking, is never where anyone said it would be. The invitation said growth, prosperity, this time is different. We showed up and found the usual: inequality, institutional capture, a bust following a boom, and someone in the corner explaining why nobody could have seen it coming.

We saw it coming. We published a working paper about it. It got three citations, two of which were us.

You only call us when things get terrible. We have made peace with this. It is, in fact, the business model.

The economy crashes: call the economists. The economists disagree: call the economic historians. The economic historians say “well, actually, in 1873—”: hang up, wait for things to get worse. Things get worse: call the economic historians back.

We pick up. We always pick up. You can reach us at:

+1 (873) 929-1348

(That’s the Panic of 1873, the Crash of 1929, and the Black Death of 1348 — our three most reliable references, and frankly, our entire voicemail.)

We are available Monday through Friday, 9am to whenever the archive closes, which is always earlier than you expected and never when you need it most. Saturdays we are in the stacks. Sundays we are thinking about the stacks.

Mark Twain — a man who understood that the best way to tell the truth is to make people laugh until they realize you weren’t joking — once observed that history does not repeat itself, but it rhymes. He was right, which is annoying, because he was a novelist and we spent a decade getting a PhD.

We are here because irony is the only honest register left. Because the Simpsons explained path dependence before most economists named it. Because an episode of Family Guy contains more useful lessons about moral hazard than three quarters of the policy briefs produced in Washington since 1980. We do not say this to be provocative. We say it because it is true and because nobody else will.

This is a platform for horror stories told with a straight face.

Stories about famines that were also policy decisions. About crashes that were also transfers of wealth. About miracles that were also someone else’s catastrophe. About growth that grew for some people, in some places, for a while, until it didn’t — which is the full, unabridged version of most economic history, available in all good archives and almost no airport bookshops.

We believe in archives. We believe in the catacombs of knowledge — the library, the special collection, the dusty box in a provincial repository that has not been opened since 1987 and contains, we guarantee you, something that would change the conversation entirely if anyone cared enough to look.

We believe in bookstores, physical ones, the kind that smell like time and bad financial decisions, because some of the best economic history ever written is currently holding up a wobbly table in a second-hand shop in a city you will only visit once.

Yes, we are writing this on the internet. Yes, we spend more time than we will ever admit on JSTOR, on HathiTrust, on every digitized newspaper archive that loads at 2am when the insomnia hits and the footnote won’t close itself. We have Google Scholar open in approximately eleven tabs right now. We are on Academia.edu despite knowing better. We have a ResearchGate profile we did not create and cannot delete. The digital era has made us more productive, better connected, and significantly less dusty — and we hate how much we need it, the way a person hates their phone while scrolling at midnight, the way Springfield hates and needs the nuclear plant, the way every character in every economic crisis we have ever studied hated and needed the thing that was actively destroying them.

We are addicts. Functional ones, mostly. The kind who still insist the old way was better while quietly running a keyword search across 400,000 digitized parish records before breakfast. We will defend the archive with our last breath and then go home and check if the archive has been digitized yet. It has. We had already downloaded it. It is in a folder called FINAL_sources_USE_THIS_ONE next to four other folders that say the same thing.

And when the whole operation breaks down — which it does, regularly, with the reliable timing of a pre-war commodity market — we call our university’s technical assistance. He comes. He always comes. He sits down, types four things, and fixes in eleven seconds what defeated us for three days. He does not say anything. He does not need to. The silence says everything. For approximately forty-five minutes after he leaves, we feel like Silicon Valley programmers. We have solved the problem. We understand the system. We are, finally, people who know what they are doing with technology. Then something else breaks, or we try to share a file, or we attempt to update a plugin, and we are immediately returned to our natural state: highly specialized, deeply knowledgeable about the wool trade in nineteenth-century Nevada, and absolutely helpless in front of a dropdown menu.

And all of this, our wives would like to point out, for what, exactly.

They have asked this question before. They will ask it again. They ask it the way good historians ask good questions: already knowing the answer, deeply unconvinced by any response we might offer, and with the patient, slightly exhausted expression of someone who has read the primary sources and found them wanting. We have no rebuttal. We have a footnote. It is not the same thing.

We have considered explaining the intellectual rewards. The satisfaction of recovering a lost voice from the archive. The quiet thrill of a dataset that finally makes sense. The manuscript that took eleven years and will be read, we are told, by dozens of people. We have considered saying all of this.

We do not say it. We close the laptop. We come to dinner.

The footnote can wait. It has waited before. It will, in all probability, wait forever — which is, when you think about it, the most economic history thing about this entire enterprise.

— The Invisible Band of Economic Historians

Arriving late, as scheduled, since 1873

.