Humanity has already lived through the disappearance of its biggest job. It went fine. It also went badly. Both parts matter, because we’re about to do it again.
The numbers first
In 1800, farming employed about 75 percent of American workers. By 1900 it was 40 percent. Today it is under 2 percent, and output is higher than ever. The tractor, the combine, fertilizer chemistry, and hybrid seeds removed the need for most human hands in the single activity that had defined human labor for ten thousand years.
What the sanitized story leaves out
The transition was not automatic and it was not gentle. It produced the Populist revolts, child labor in the mills, urban slums, and the Dust Bowl migrations. The reason it ultimately worked, the reason displaced farm families’ grandchildren ended up better off rather than permanently stranded, is that America invented an entire institutional layer that had never existed anywhere.
Universal high school. In 1910, fewer than 10 percent of American 18-year-olds graduated from high school. By 1940 it was 51 percent, and nearly three quarters of teenagers were enrolled. Economists Claudia Goldin and Lawrence Katz call this the high school movement, and it was arguably the single best economic decision the country ever made: a massive, expensive, contested bet that the children of farmers needed an education nobody’s parents had needed.
The rest of the layer came in the same generation. Social Security in 1935, because old age could no longer be absorbed by the family farm. The 40-hour week in 1938. The five-day week, spreading after Ford adopted it in 1926. Labor law, unemployment insurance, the whole architecture we now mistake for permanent features of civilization. All of it invented, on purpose, under pressure, roughly a generation after the displacement began. That lag, the gap between when the old structure failed and when the new one existed, is where most of the suffering lived.
The job is where farm labor was in 1900
The job today sits directly under a compounding cost curve: the price of AI performance at a fixed capability level fell around 280-fold in two years, and even if capability froze, decades of ordinary deployment would absorb an enormous share of coordination and knowledge work, the way tractors absorbed field work. Not overnight. Farming’s decline took a century. But the direction is not seriously in doubt, and the task-by-task pattern is already visible: not jobs vanishing whole, but jobs quietly hollowing as the tasks inside them get absorbed.
What makes this bigger than a labor market story is what the job carries. Income, obviously. But also the structure of the day, health insurance, retirement savings, social standing, geography, and the answer to “so, what do you do?” Farm labor carried most of that in 1800, which is exactly why its disappearance forced the invention of school, pensions, and the weekend. Nothing else in modern life is as load-bearing as the job. That’s why this is the shift that cascades furthest.
What gets invented next
The useful question is not whether the transition happens. The curves answer that. The useful question is the 1910 question: what’s this era’s high school movement? What are the institutions that don’t exist yet, that will look as obvious in retrospect as universal secondary education looks now?
I’d bet on three requirements, whatever forms they take. Income security that isn’t welded to a single employer, the way Social Security decoupled old age from the family farm. Verification of what someone can actually do, decoupled from degrees, the way the diploma once decoupled opportunity from family land. And most important, formation: the deliberate development of the capacities machines don’t absorb, defining problems, exercising judgment, acting with agency when nobody hands you instructions.
That last one is the piece I have direct evidence on. I’ve spent over a decade watching teenagers build real companies, and the consistent surprise is not what they know. It’s what they become when handed real stakes and real responsibility: they learn to move without permission, to make decisions with incomplete information, to recover from failures that actually cost something. That is formation, and no cost curve absorbs it. The education system we have was built to produce reliable employees for the economy that’s dissolving. The one we need produces people who can operate without the scaffolding the job used to provide.
The high school movement started roughly a generation late, and the gap was ugly. This time we can see the curve from here. The interesting work of the next 20 years is not predicting the shift. It’s building the institutions early enough that the gap doesn’t have to be.
Sources: farm labor share via EH.net and USDA; high school movement via Goldin and Katz, “Why the United States Led in Education” (NBER); AI cost decline via Stanford AI Index 2025.
Laurie Stach 