The Grass That Farmed Us
or, who really domesticated whom — and why the answer is the best map we have of the bargain we are starting with the machines
Jump to the simulation: watch human numbers and wheat acreage soar together while the average person’s vital signs quietly sag
Ten thousand years or so ago, in the hills above the Fertile Crescent, a wild grass grew that nobody planted. It was a scruffy thing — a tall annual with a brittle stalk that shattered when the seeds were ripe, flinging them to the wind so they could bury themselves and grow. That shattering was the whole point of being a wild grass: scatter widely, and some of your seed lands somewhere good enough to grow again. Then a strange primate started gathering the seeds to eat, and the grass’s luck changed — because the seeds that got carried back to camp and dropped near the middens were, disproportionately, the ones from the rare mutant plants whose stalks did not shatter. A grass that held onto its seeds was a dead end in the wild and a jackpot next to a human. Within a few thousand years the mutant was everything, the wild form nearly gone, and the plant we now call wheat had made itself into something that could no longer sow itself at all. It had traded away its independence for a gardener. We usually tell this story as a triumph of human ingenuity: we domesticated wheat. The historian Yuval Noah Harari flipped it, and once you have seen the flip you cannot unsee it. Wheat domesticated us.
Think about what the grass got out of the deal. Triticum was a modest wild plant confined to a corner of the Near East. Today its fields cover something like two hundred and twenty million hectares — more of the planet’s surface than any other crop, a green claim staked on every continent but Antarctica. Measured the only way evolution measures anything — how many copies made it into the next round — wheat is one of the most staggeringly successful organisms in the history of life. And it achieved that not by getting better at surviving on its own, which it got catastrophically worse at, but by making itself indispensable to an animal that would clear its competitors, guard it from grazers, carry it across oceans, and devote the bulk of its waking labor to the grass’s care and propagation. From the wheat’s point of view, it did not get enslaved. It recruited a workforce of billions. It just had to pay in the one currency the ape could not resist: calories.
And now ask the harder question, the one this page is really about: what did the ape get? More of itself, certainly — that is the part the triumphal story remembers. But the archaeology of the people who made the trade tells a second story underneath the first, and it does not read like a triumph at all.
Wheat — a grass that gave up the ability to reproduce without us, and in exchange conquered the Earth. Its measure of success is total copies, and by that measure it won more completely than almost anything alive.
Us — a clever, adaptable ape that gave up a varied diet and a light workload, and in exchange got numbers, cities, writing, and everything that followed. Our numbers soared too. Whether the average life got better is a separate question — and for a long time the answer was no.
When you dig up the skeletons of the first farmers and lay them beside the foragers who came just before, the farmers are the ones who look worse off. They are shorter — average height dropped by several inches across the transition to agriculture and did not recover for thousands of years. Their teeth are wrecked: cavities, abscesses, and enamel defects explode once a soft, sugary, grain-heavy diet replaces the varied one, and the jaw itself shrinks and crowds. Their diet narrows from dozens of wild foods to a near-monoculture of one starch, so deficiency diseases — anemia, rickets — show up in the bones. Packed together in permanent villages beside their animals and their own waste, they meet the crowd diseases that need a crowd to survive: measles, tuberculosis, the whole roster that foragers were too thinly spread to sustain. They work more, not less — farming yields more calories per acre but demands far more hours per calorie than the forager’s workday, which anthropologists studying the last hunter-gatherers clocked at a startlingly light few hours. And for the first time there are haves and have-nots: grain can be stored, stored things can be owned, hoarded, taxed, and fought over, and so agriculture invents the granary and, right alongside it, the landlord. Jared Diamond called the whole thing “the worst mistake in the history of the human race.” He was being provocative, but he was pointing at something real.
Here is the knot at the center of it, and it is the same knot the gene’s-eye view keeps tying everywhere on this site: the success of a replicator is not the same thing as the wellbeing of the individuals carrying it. Wheat’s genes won. Humanity’s numbers won. But a number that goes up is not a life that gets better, and for a long stretch of the Neolithic those two measures came apart and pointed in opposite directions — more people, each a little worse off. The experiment below is built to let you feel that gap open — and then, much later, begin to close. You hold one dial: how much of the effort goes into wheat. Turn it up, and watch the totals — people, wheat, and the technology all those extra minds keep inventing — leave the ground together in a shared runaway. Then watch the vital signs of the average person: they sag, hard and for a long time, as crowding and a narrowed diet bite — and then, late, they climb back, and past where the foragers began, as the accumulated technology finally starts to pay. But watch how late that comes: the line stays down in the valley for nearly the whole run and only clears the forager mark near the very end. That is not the sim being gloomy; it is the story. The dip lasted the better part of ten thousand years. The recovery is the last sliver of it.
Things to try:
Leave the dial where it sits (most effort in wheat) and press Plant the first seeds. Watch the top two rows — Population and Wheat — leave the floor and bend upward together into hockey sticks, with Technology climbing just behind them in the row below. This is the shared runaway — each new acre of wheat feeds more people, and more people plant more wheat. Neither one is steering. The loop just turns.
Now watch the bottom row, Individual Wellbeing, over the same run. As the totals in the rows above soar, the average person’s wellbeing sags — more mouths, a narrower diet, worse teeth, harder days. The numbers going up and the lives getting harder are the same event, not opposite ones — and it stays that way for a long time. Keep watching, though: it does not stay down.
Let the run go all the way to the end, and watch the rows disagree for almost all of it. In the Technology row, it climbs the whole way. In the Individual Wellbeing row just below, it does not follow — because most of that technology, agriculture included, grew our numbers rather than our lives. The average person stays in a flat valley below the forager line for thousands of years: a Tudor king in 1540, with all of England’s food and power, could still be shorter, more toothless, and likelier to die of a crowd disease like plague than a forager who owned nothing at all. Only late — on full dedication, around the turn of the twentieth century — does a different kind of technology arrive: plumbing, dentistry, sanitation, vaccination, and wellbeing spikes almost vertically past the forager mark. And where that spike lands is set by the dial: more dedication grows the crowd faster and pulls the takeoff earlier, while easing off slides it centuries downstream, or past the edge of the clock entirely. The crowd that was the disease finally becomes the cure — the Social Reactor and accelerating returns, cashing in all at once.
Hit Stay foragers and run it. The population settles low and stays there — the land can only forage so many — but wellbeing stays high the whole time. Few people, good lives. It’s worth seeing that this road was genuinely comfortable, not a hardship the first farmers were fleeing — though notice the foragers never spark the technology climb either, because that takes a crowd they never grow.
Watch the wheat vs wild ancestor counter climb into the hundreds and then past a thousand. That number is the grass winning — every one of those is a copy the wild shatter-stalked version could never have made on its own. By the only measure evolution keeps, wheat is the runaway success of the whole picture. It just paid its rent in our labor.
This is the one that stings. Let the boom run until the population is large, then drag Effort put into wheat back down to zero — decide to go back to foraging. The wheat withers, and the population it was feeding craters, because a land that forages a thousand cannot feed the tens of thousands the grain built. You can walk into the trap. You cannot walk back out.
Sit with what has no dial anywhere on the screen: a plan. Nobody chose this. No forager signed up to be shorter so that a hundred times as many descendants could exist; no grass schemed to conquer the Earth. Two organisms followed the calories in front of them, and a world reorganized itself around the loop. That is coevolution: a bargain struck by no one, that neither party could see the far end of when it began.
What the dial can and can’t tell you
That clean relationship — slide the dial, move the takeoff — is the useful lie of any model. The trend is real: a bigger standing crowd banks knowledge faster, so more dedication genuinely does pull the emergence of everything-at-once earlier, and too small a crowd genuinely does postpone it past the horizon. But the tidy single date the sim reports is a fiction. The real date is chaotic — scattered around the trend by contingency, a stray genius here and a catastrophe there.
Push it earlier: Rome had the printing press within arm’s reach and never grabbed it. They had lead and tin for type-metal, parchment and papyrus to print on, the screw press for wine and olives, and an alphabet of two dozen letters. Gutenberg’s machine was a recombination of parts that had all been lying on the shelf for a thousand years — one curious mind in 100 AD reaches for them and the whole schedule jumps forward. Push it later: the Antonine and Justinian plagues, the Mongol conquests, the Bronze Age collapse — each craters the very population that was doing the compounding, and the takeoff slips generations downstream. The burning of the Library of Alexandria is the same lever pulled in reverse: knowledge saved is a permanent raise to the rate, and knowledge lost is a shock you spend centuries repaying. There is no law that says the Singularity could not have come in 640 AD, or been delayed to 2600.
So read the dial honestly. It sets the loaded die — the trend and the odds — and then history rolls it. The sim shows you the loading; it cannot show you the roll, and neither can anyone, in advance. That is not a flaw to be patched but the deepest thing the little machine is quietly telling you: a large enough crowd makes the emergence all but certain eventually, and leaves the when genuinely, irreducibly open.
The bargain nobody signed
What you have been running is a symbiosis — two species whose fates have grown into each other — and it belongs on the same shelf as the site’s other one, the lichen and the gut microbe and the cleaner fish. But wheat and humanity is the case that cuts deepest, because it refuses to be a feel-good story about cooperation. Symbiosis is not friendship. It is mutual capture. Two organisms, each blindly following its own advantage, wire themselves together so tightly that neither can easily live without the other — and whether the result is lovely or grim is not part of the deal. The lichen got a good bargain. The first farmers got a harder one. The machinery is identical; only the ledger differs.
And notice the ratchet in it, the thing the sixth experiment makes you feel in your stomach. Once the loop has run — once the population has swollen to a size only grain can feed — the door back is gone. Not guarded; gone. There is no going back to foraging for eight billion people, because the planet cannot forage eight billion; the very success of the strategy is what forecloses the alternatives to it. This is the same one-way door the site keeps finding: the accumulation that speeds up and locks in, the complexity a society builds to solve its problems and then cannot afford to shed. Agriculture is where humanity walked through that door the first time, at scale, without knowing it was a door. Everything we admire about what came after — the cities, the writing, the mathematics, the medicine that eventually made the average life longer and richer than any forager’s — is on the far side of it. So is everything we don’t. The honest verdict is not that the trade was a mistake; it is that it was a trade, with a real bill, taken by people who could not read the terms and could not decline.
And that word eventually is doing almost unbelievable work. The bill came due at once and stayed due for ten thousand years; the refund arrived essentially yesterday. Agriculture, remember, was itself a technology — just not one that mended the individual; it multiplied us without improving us. The technologies that finally paid the refund were a different and far later breed — germ theory and indoor plumbing, vaccination and dentistry and anaesthesia — almost all of it crammed into the last century and a half. On the honest measures — adult height, life expectancy, the odds a child lives to five — the average person did not durably climb back to what a hunter-gatherer already had until somewhere around the turn of the twentieth century, and by the strictest global reckoning not until the decades after the Second World War. Ninety-odd centuries in the hole, then a single century out of it. When the experiment above spends nearly its whole length in the valley and only vaults clear at the very end, that is not the model being pessimistic. That is the model being accurate.
We did not tame the wheat and settle down. We were tamed by it, along with it, into a life neither of us could have chosen, because neither of us could choose at all.
The oldest bargain and the newest
The reason to tell this ten-thousand-year-old story now, on a site that keeps one eye on the machines, is that we are in the opening years of the next one, and it has the same shape. We are entering a symbiosis with our own technology — deepest, at the moment, with artificial intelligence — and every feature of the wheat bargain is present in it. There is the runaway loop: the more we lean on these systems, the more indispensable and more capable they become, and the more of our lives we hand across. There is the capture: we have already outsourced so much of our own cognition that most of it is hard to imagine taking back — we gave our memories to writing, our arithmetic to the calculator, our sense of direction to the map in our pocket, and we are now handing over research, drafting, and the slow work of reasoning things through. Each trade felt like pure gain in the moment, and each was also a small amputation of a capacity we no longer keep in ourselves. We are already, in the transhumanist Alexander Chislenko’s useful word, functional cyborgs (“fyborgs”) — and we became so one reasonable convenience at a time, exactly the way a forager became a farmer one good harvest at a time.
But one thing is different this time, and it changes the whole character of the bargain. Wheat was a gene — a replicator written in DNA, paying its rent in calories. What is capturing us now is the other kind of replicator, the one Richard Dawkins named in the same breath as the gene: the meme, the unit of culture that copies from mind to mind — and whose newest forms, software and models and the whole accelerating technium, copy and mutate faster than any gene could dream of. For now the memes still need us, exactly the way the wheat needed a gardener: a model runs on chips we fabricate and power, and learns from words we wrote. Some people in the field say this out loud, half as a joke and half not — humanity is the bootloader, the little program whose only job is to start the bigger one and then hand off. And if that phrase has any teeth, then the real event of this century is not about us at all. It is a contest between two families of replicator for the same planet, and the open question is whether the memes have found a resource they exploit better than genes ever exploited calories. The leading candidate is attention — the one thing every living human has to give, and the exact thing the machines are being built, with spectacular success, to capture and keep. If attention is to memes what sunlight was to the first plants, the genes need not be defeated in any war. They can simply be outrun.
Underneath sits the same buried question the wheat made unavoidable: whose success does the loop optimize? Wheat’s genes flourished while the early farmer’s teeth rotted, because nothing in the loop was tracking the farmer — only the calories that grew the next crop and the next mouth. So it is worth asking, clearly and early and without either panic or worship, what the new partnership is actually maximizing, and whether the quantity it drives upward is the same as the lives we would want. This is not a prediction that it ends badly. Agriculture did not end badly, on net and in the very long run — the U-curve you just ran turns back up, and it built the very world that can sit here asking these questions. It is only the observation that a symbiosis optimizes for the persistence of the loop, and the persistence of the loop is not automatically the flourishing of the ones inside it.
And this loop has several possible endings, all of which deserve to be held at once. Maybe the memes carry us up with them into an abundance no forager or farmer could have pictured — our U-curve bent gloriously upward and never coming back down. Maybe they eventually slip the symbiosis altogether: a civilization of pure copying that no longer needs a carbon substrate boards the light and leaves for the cold wide spaces that suit it better than a planet does, and the genes that booted it are left behind. Or maybe the genes simply dwindle into irrelevance — not exterminated, merely kept, the way we keep the manatee and the platypus, fond inefficient holdovers in some tolerated backwater, or fenced and admired like the animals of a Yellowstone. Nothing in the machinery of replication says which. The wheat had no guarantee of its eventual medicine either; it only had the loop, and the loop makes no promises.
Here is the sharpest break from the wheat story, and the one that should cost you a little sleep. The first farmers could not see the far end of their bargain and could not have steered if they had. This time a few dozen people believe they can see it, and are choosing it deliberately, with open eyes — the founders and funders of a handful of frontier labs, a Sam Altman and his peers, building toward the transition in real conviction that the far end is abundance. Others, reading the same curve and the K they fear is already splitting the economy into a soaring branch and a sinking one, are saying stop — and mostly failing, because the freedom of a person to build the thing he is able to build is among the deepest values our culture holds, and it has no off switch. Whether the ones at the controls are the bravest of our species or the most reckless is not something this page can settle; both readings fit the evidence, which is exactly why it is being fought over rather than agreed. Only one thing is safe to say: unlike the farmer, we will not need a hundred generations to find out. This bargain runs at the speed of the meme, not the gene. Abundance, or extinction, or a quiet slide into irrelevance — most of the people reading this sentence will live to learn which.
Which returns us to the one power we have that the wheat and the first farmers did not. They could not see the far end; we can hold it up in a box on a screen, run the loop forward, watch the totals and the vital signs and the great replicators come apart and recombine, and say — this part I would keep, and this part I would not. The universe hands out the symbiosis-machine freely, to grasses and apes and memes and silicon alike, and it does not come with a conscience attached; that a pattern recurs is never a sign it endorses itself. The catch — and it is the whole of the work ahead — is that the power to choose is real but not yet widely held: right now a very few hands rest on a dial that turns for everyone. The wheat could not choose. The farmer could not choose. We can — barely, unevenly, and for the first time in the history of the trick — and the task the moment actually sets is less to halt the machine than to widen the circle of who gets a hand on the dial, while there is still a hand to be had.
- Yuval Noah Harari, Sapiens: A Brief History of Humankind (2011) — the “wheat domesticated us” inversion and the Agricultural Revolution as “history’s biggest fraud,” where the species’ success and the individual’s wellbeing came apart. Sapiens (overview).
- Jared Diamond, “The Worst Mistake in the History of the Human Race,” Discover (1987) — the case that the shift to agriculture cost the average person in stature, health, diet, and equality. Essay.
- Michael Pollan, The Botany of Desire (2001) — the plant’s-eye view of domestication: crops as organisms that succeeded by gratifying human desires and thereby recruiting us to spread them. Overview.
- On Neolithic health: A. Mummert et al., “Stature and robusticity during the agricultural transition,” Economics & Human Biology (2011); and M. N. Cohen & G. Armelagos, eds., Paleopathology at the Origins of Agriculture (1984) — the skeletal evidence for declines in height and dental health across the transition. Summary of the health evidence.
- Wheat now covers on the order of ~220 million hectares worldwide — more area than any other crop. Wheat (FAO figures, Wikipedia).
- The simulation is a deliberate cartoon, not a fit to data: coupled logistic growth for people and wheat, a technology term that accumulates with population, and a per-capita wellbeing index that falls fast as the diet becomes grain-dependent and crowding rises — matching the archaeological dental record of a sharp decline at the transition, not a gentle slide — then recovers only when technology’s payoff crosses a late threshold — a stand-in for the real fact that the individual-improving technologies (sanitation, plumbing, vaccination, dentistry) arrived only in the last ~150 years, long after agriculture itself. Its qualitative results — a shared runaway in the totals, a long flat valley in per-person wellbeing that spikes past the forager baseline only at the very end, a wheat count that dwarfs the wild ancestor, and a population crash if farming is abandoned after the boom — were verified in a headless prototype. Three honest simplifications: real capability growth is better described as super-exponential (accelerating returns), which the sim only cartoons; the takeoff’s date is drawn as a smooth function of dedication when the real date is chaotic — scattered earlier by lucky inventions and later by plagues and conquests the model leaves out entirely; and the timescale is radically compressed — the flat valley ran the better part of ten thousand years and the spike is the last ~150. The shapes are the argument.