The Herd Has No Head
Four centuries separate the tulip taverns of Haarlem from the meme-stock subreddits, and the distance between them is close to zero. When prices stop measuring value and start measuring each other, something ancient takes over, and the most dangerous sentence in finance becomes: everyone else seems to know something.
In the winter of 1636, in the back rooms of Dutch taverns, grown men gathered nightly to trade flowers that did not exist. The tulip had arrived from the Ottoman world a few decades earlier and become the status object of the young Dutch Republic, and the most coveted varieties, the flamed and feathered ones like Semper Augustus, striped crimson on white, owed their beauty, botanists later learned, to a virus, which is the sort of detail history includes when it wants to be sure you get the joke. Because tulips spend most of the year as bulbs underground, the trade evolved into contracts on future flowers, paper promises passed from hand to hand in smoke-filled “colleges” where the wine flowed and the prices only rose. In the final weeks of the mania, single bulbs of the great varieties were changing hands, on paper, for the price of a fine canal house in Amsterdam; one famous tally for a single Viceroy bulb ran to wheat, rye, oxen, swine, sheep, wine, beer, butter, cheese, a bed, a suit of clothes, and a silver drinking cup. Then, on the third of February, 1637, at a routine bulb auction in Haarlem, the bids simply did not come. No news, no crop failure, no edict. The auctioneer lowered the price, and lowered it again, and the room, which for months had contained only buyers, discovered it now contained only sellers. Within days the paper was worthless from Amsterdam to Utrecht, and men who had been rich at Christmas were negotiating with magistrates by spring.
I should pause to be honest about this story, because honesty makes it stranger, not tamer. The version everyone knows descends from Charles Mackay’s 1841 classic Extraordinary Popular Delusions and the Madness of Crowds, and Mackay, a journalist with a moralist’s instincts, embellished freely; modern historians, above all Anne Goldgar, have combed the Dutch archives and found no wave of bankruptcies, no bodies in canals, an episode largely confined to a few hundred merchants and craftsmen rather than a whole nation possessed. Some economists have even argued the late-stage prices were briefly rational responses to a rule change that converted contracts into cheap options. And yet the correction, valuable as scholarship, changes nothing essential, because the mechanism the taverns revealed did not need to bankrupt Holland to be true, and it has since reproduced itself at every scale the world can offer: the South Sea Company in 1720, where Isaac Newton, according to the famous and possibly polished anecdote, sold early at a profit, watched his friends keep riding, bought back in near the top, lost the modern equivalent of millions, and reportedly said he could calculate the motions of the heavenly bodies but not the madness of people; the Nifty Fifty; Japan in 1989, when the grounds of the Imperial Palace were said to be worth more than California; the dot-com years, when adding “.com” to a company’s name measurably raised its price; and January 2021, when a subreddit’s affection for a dying mall retailer briefly made it worth more than half the companies in the S&P 500, and grown men once again traded flowers that did not exist, this time with rocket emojis.
The mechanism deserves to be stated plainly, because it is not stupidity, and calling it stupidity is how each generation ensures its own turn. A market price is a compression algorithm: it takes everything everyone knows and squeezes it into a single number, and most of the time this is the most useful invention in economic history. But the algorithm has a failure mode. When I buy because I have studied the asset, my purchase adds information to the price. When I buy because the price is rising, because you bought, because your buying suggests you know something, my purchase adds no information at all; it adds only confirmation, and the price begins to feed on itself. Economists call the individual logic an information cascade, and the cruel elegance of a cascade is that every participant is behaving rationally given what they can see; it is genuinely reasonable to infer that a thousand buyers know more than you do. What no one can see is that the thousand buyers each made the same inference about the nine hundred ninety-nine before them, so the crowd’s apparent mountain of knowledge traces back, like a rumor traced to its source, to almost nothing. Keynes gave the phenomenon its permanent image in 1936: the newspaper beauty contest in which the prize goes not to whoever picks the prettiest face, but whoever picks the face the other entrants will pick, so that everyone is devoting their intelligence to anticipating what average opinion expects average opinion to be. The herd, in other words, has no head. There is no one at the front doing the thinking. There is only the aggregate of everyone watching everyone, a creature with ten thousand eyes and no brain whatsoever, galloping.
What makes bubbles durable rather than momentary is that standing apart from the herd is not free, and the market makes you pay in the most painful currency there is, which is watching your neighbors get rich. The fund manager who correctly identifies the mania faces a horrible clock: clients do not compare him to his eventual vindication, they compare him to this quarter’s leaderboard, and careers rarely survive three years of being right too early, which is why so many professionals privately doubted the dot-coms while publicly riding them, a posture summarized forever by Chuck Prince of Citigroup in July 2007, months before the music actually stopped: as long as the music is playing, you have got to get up and dance. Prince was widely mocked for that sentence, and the mockery has always seemed to me slightly unfair, because he was not describing his own foolishness; he was describing, with the precision of a man who knew, the incentive structure of every intermediary in every bubble in history. The tavern colleges of 1636 had the same structure. So does a subreddit where selling is treason and diamond hands are a moral identity. The herd does not merely reward joining; it makes leaving feel like betrayal, and it makes the sober man at the edge of the room feel, for exactly as long as it matters, like the only fool present.
I have a machine in the lab that makes the mechanism visceral: two hundred forty simulated traders, each glancing at its neighbors, with a single dial that raises how much weight they give to each other versus their own information. At low settings the crowd argues with itself and the price wobbles honestly around reality. Turn the dial past a threshold and, with no news whatsoever, patches of agreement form, spread, and suddenly the whole grid moves as one animal while the price departs on a journey to nowhere, and what I find quietly horrifying every time I run it is that no individual trader changed character; the only thing that changed was how much they watched each other. That is the whole story of four centuries in one slider. Financial innovation changes the plumbing, futures contracts in Haarlem, margin accounts in 1929, options flow in 2021, and the plumbing determines how fast the water moves, but the water is us, and the water has not changed at all since the Dutch invented, almost simultaneously, and I do not think this is a coincidence, the joint-stock company, the stock exchange, and the speculative crash.
The uncomfortable ending, which the genre of bubble essays usually softens, is that there is no reliable vantage point outside the herd, and everyone who believes they have found one should study Newton on his second entry into the South Sea. You cannot simply resolve to be contrarian, because most of the time the crowd is right; that is what makes cascades work. The honest disciplines are smaller and less heroic. Ask, of any position you hold, what percentage of your conviction traces to the asset and what percentage traces to its price chart and its popularity, and be genuinely suspicious when the second number dominates. Write down, while calm, what evidence would make you sell, because the herd will make sure no moment ever feels like that moment. Size positions such that being wrong with the crowd, or early against it, is survivable, since either fate will eventually be yours. And when you feel the specific warmth of consensus, the relief of finally agreeing with everyone, treat it as data, the way a sailor treats a falling barometer. It is the best feeling in markets, and it is the one the wagon of fools was built to sell. The painting hung in Haarlem within months of the crash. They knew. They always know, afterward. The whole trick of a financial life is arranging, in advance, to be someone who can afford the tuition of the lesson, because the lesson has been on the syllabus every forty years or so for four hundred years, and attendance, sooner or later, is mandatory.
Notes & sources: the tavern “colleges,” the February 3, 1637 collapse at Haarlem, and the Viceroy goods tally are from standard histories, with Mike Dash’s Tulipomania (1999) and Anne Goldgar’s corrective Tulipmania (2007) on the limited real damage; Mackay’s 1841 account remains the myth’s source; Semper Augustus’s breaking pattern was caused by the tulip-breaking virus; Newton’s South Sea losses per contemporary accounts and John Carswell’s The South Sea Bubble (1960), the famous quote possibly apocryphal; information cascades per Bikhchandani, Hirshleifer & Welch (1992); the beauty contest from Keynes’s General Theory, ch. 12; Chuck Prince’s dancing remark from the Financial Times, July 9, 2007. Painting: Hendrik Gerritsz Pot, c. 1637, public domain.