Notes Toward a Better Century

In the autumn of 1930, an economist and a cereal maker each bet on a future of shorter days. The economist got the money right and the hours wrong. The cereal plant got the hours right and then lost them. What the two experiments tell us about the hundred years that end in 2030, and what to do on Monday.

Two Experiments, Six Weeks Apart

In the autumn of 1930, two men who never met made the same wager about the future, one in prose and one in payroll.

The first was John Maynard Keynes, who published an essay called “Economic Possibilities for Our Grandchildren” in two installments of The Nation and Athenæum, on October 11 and October 18. The world was a year into the Depression. Keynes opened by naming the mood (“We are suffering just now from a bad attack of economic pessimism”) and then declared it “a wildly mistaken interpretation of what is happening to us.” He asked what life might look like a hundred years on, and he answered with a number: “the standard of life in progressive countries one hundred years hence will be between four and eight times as high as it is to-day.”

Then he went somewhere economists rarely go. If people were that much richer, he reasoned, they would eventually need to work much less. “Three-hour shifts or a fifteen-hour week may put off the problem for a great while,” he wrote. “For three hours a day is quite enough to satisfy the old Adam in most of us!” The true difficulty of the coming century, he said, would not be production at all. It would be “how to use his freedom from pressing economic cares, how to occupy the leisure, which science and compound interest will have won for him, to live wisely and agreeably and well.”

Six weeks later, on December 1, 1930, in Battle Creek, Michigan, W. K. Kellogg did something that looked like a pilot study for Keynes’s essay. According to the historian Benjamin Kline Hunnicutt, who spent years interviewing the plant’s workers, Kellogg replaced the three daily eight-hour shifts in his cereal plant with four six-hour shifts. By adding a whole shift, he reasoned, the plant would add about 30 percent more jobs for a city that badly needed them. Workers were paid for seven hours during the first year; from the second year, Hunnicutt reports, total wages were raised back to the nominal level of the eight-hour day.

So in the same season, a theorist in Cambridge and a manufacturer in Michigan placed the same bet: that the dividend of productivity could be taken in time rather than only in things. We now have nearly a century of results. They are more interesting than either man expected, and they bear directly on a question we are being asked again, with new machines doing the asking.

I am a writer who distrusts prophecy and likes evidence, so I want to grade both experiments honestly before drawing any lessons. The grades are mixed. That is the point.

Grading Keynes on the Money

Begin with the part Keynes got right, because it is the part people forget.

Keynes predicted growth of four to eight times in a century. The economist Fabrizio Zilibotti, revisiting the essay at its 75th anniversary, found that the world did better than that. Population-weighted global income per person, Zilibotti calculated, grew about 2.9 percent a year from 1952 to 2000, a pace that roughly quadruples living standards in fifty years. Extended over a century, it would mean something like a seventeen-fold increase, well beyond Keynes’s upper bound. Growth, in other words, did not stall as the pessimists of 1930 feared. Zilibotti was careful to note that the gains were far from universal, with sub-Saharan Africa the glaring exception.

Keynes was right about something else that does not appear in his essay but haunts it: the length of the lives that would have to be filled. In 1930, life expectancy at birth in the United States was 59.7 years, according to the historical series kept by the National Center for Health Statistics. In 2023 it was 78.4. That is close to nineteen extra years for the average American, more than the entire length of a childhood. Keynes worried about what people would do with their afternoons. The century handed them, in addition, an extra two decades of afternoons.

So on the economic forecast, give him high marks. A man writing in the trough of the worst slump of his lifetime looked past it and got the direction and the order of magnitude right. That is rarer than it sounds.

Grading Keynes on the Hours

Now the part everyone remembers.

The fifteen-hour week did not arrive. Our World in Data estimates that the average American worker put in about 3,096 hours a year in 1870. By 1950 it was about 2,022. By 2017 it was about 1,796. (The series changes definitions in 1950, from full-time production workers to the whole economy, so the comparison is approximate.)

Read that sequence slowly. Hours fell by roughly a third in the eight decades from 1870 to 1950, which is why the trend looked so obvious to him. Then they kept falling, but far more gently, by roughly another eleventh in the nearly seven decades after. A fifteen-hour week is about 780 hours a year. We stalled at more than double that.

So Keynes was right about the engine and wrong about how people would spend its output. And this is where the second experiment, the one in Battle Creek, becomes the more instructive of the two, because it shows us, in close detail, how a society that has the hours can decide to give them back.

What Happened in Battle Creek

The six-hour day was popular. In 1932, Hunnicutt writes, the Women’s Bureau of the United States Department of Labor sent researchers to Battle Creek to interview the plant’s women workers. Nearly 85 percent preferred the six-hour shift, chiefly because it provided “more time for family activities and home duties and leisure,” and because it helped some of the unemployed find work. When the researchers asked women to compare the two schedules, Hunnicutt reports, the great majority reached for the word “freedom” or something close to it.

He collected the texture of that freedom in later interviews (he gives the workers pseudonyms). A self-educated woman used the extra hours for reading, walking, and writing. A mother described canning with her sons as the time “we were the most together as a family.” One worker learned to fly. Men voted, in plant-wide polls in the 1930s and 1940s, three to one for the six-hour shift. During the Second World War the plant went to eight-hour shifts under federal wartime rules, and when the war ended, workers voted three to one, in 1945 and again in 1946, to go back to six.

And yet the six-hour day died. It did not die suddenly. Management soured on it after 1938, Hunnicutt writes, partly over the rising fixed cost of benefits per worker and partly because W. K. Kellogg stepped back and his successors did not share his philosophy. After the war the company allowed “those who want it” to work longer, which split the workforce. Department by department, through the 1950s and 1960s, the eight-hour shift returned. The holdouts, by then mostly women, were mocked as “weak girls” and “housewives” by colleagues who had come to treat full-time work as the measure of seriousness. In the summer of 1984 the company, citing competitive pressure, threatened to move jobs unless the remaining six-hour departments switched. On December 11, 1984, a majority voted to accept the longer hours.

Hunnicutt’s conclusion is blunt: “This cultural change, rather than economic necessity, is the fundamental reason why the six-hour day ended at Kellogg’s.”

You do not have to accept every part of his interpretation to see why the story matters. For fifty-four years, a real plant with real competitors ran on something like the schedule Keynes imagined, and the people living inside it reported being freer. What ended it was not an inability to afford it. It was a slow change in what people believed a serious life looked like, helped along by managers who had reasons to prefer the change.

Keynes, it turns out, had predicted that too. He just placed it in the wrong column of his ledger.

The Line Everyone Skips

Most summaries of “Economic Possibilities” stop at the fifteen-hour week. The essay’s second half is stranger and more useful.

Keynes divided human needs into two kinds: “those needs which are absolute in the sense that we feel them whatever the situation of our fellow human beings may be, and those which are relative in the sense that we feel them only if their satisfaction lifts us above, makes us feel superior to, our fellows.” The second kind, he warned, “may indeed be insatiable; for the higher the general level, the higher still are they.”

He also expected trouble in the transition. “Yet I think with dread,” he wrote, “of the readjustment of the habits and instincts of the ordinary man, bred into him for countless generations, which he may be asked to discard within a few decades.” He worried about “a general ‘nervous breakdown.'” And he described, with a cruelty that has not aged well, the rich of his own day as an advance guard that had mostly “failed disastrously” to work out how to live without economic pressure.

Put that half of the essay next to the Battle Creek story and the failure of his hours forecast looks less like a miscalculation and more like a prediction that came true in a form he hoped it would not. Relative wants did prove insatiable. The habits of striving did prove stubborn. A culture “trained too long to strive and not to enjoy,” in his phrase, did exactly what he feared it might: it turned productivity into more goods and more status rather than into more time.

This matters because a version of the 1930 argument is being made again, now about software rather than steam and electricity. If machines take on a large share of routine cognitive work, the familiar promise goes, people will be freed for better things. Perhaps. Battle Creek suggests that a freed hour is not a law of physics. It is a negotiated settlement, and settlements can be renegotiated.

A Word on Being Wrong

Before going further I want to say something about the business of predicting, since it is the business this essay is skeptical of.

The history of forecasting is mostly a history of confident people being wrong, and a good share of it is also a history of people being misquoted as wrong. The most famous failed prediction in the history of technology, that “everything that can be invented has been invented,” is usually pinned on Charles H. Duell while he was commissioner of the United States Patent Office. He almost certainly never said it. The librarian Samuel Sass, writing in Skeptical Inquirer in 1989, traced the line back no further than a 1981 book called The Book of Facts and Fallacies. What Duell did say, in 1902, was close to the opposite: that all previous advances “will appear totally insignificant when compared with those which the present century will witness.”

The real record of prediction is subtler and, for my purposes, more instructive. In 1980 the economist Julian Simon and the biologist Paul Ehrlich, author of the 1968 bestseller The Population Bomb, made a famous bet. Ehrlich and his colleagues picked five metals (chromium, copper, nickel, tin and tungsten) whose inflation-adjusted prices they expected to rise over the decade ending September 29, 1990. All five fell, and in October 1990 Ehrlich mailed Simon a check for $576.07. For years the wager stood as a parable about doomsayers.

Then the economists Katherine Kiel, Victor Matheson and Kevin Golembiewski went back and ran the same bet over every possible ten-year window from 1900 to 2008. Simon, they found, would have won only 38.4 percent of the time. The decade he happened to pick was one of the worst for metal prices in the entire period. Their conclusion was that when it comes to betting, it is better to have luck on your side than expertise.

The psychologist Philip Tetlock spent two decades collecting thousands of forecasts from political and economic experts and scoring them. In Expert Political Judgment, published in 2005, he found that their predictions were often only slightly better than chance and not much better than those of casually informed laypeople. His later work offered a partial remedy. In a forecasting tournament run by the research arm of the United States intelligence community, the best performers in Tetlock’s Good Judgment Project, people who thought in probabilities, updated often and kept score, were more than 30 percent more accurate than intelligence analysts with access to classified information, according to the project.

I find this record bracing rather than depressing. It suggests that the future is not something to be announced. It is something to be estimated, revised and, most of all, worked on. Keynes himself seemed to know this. He ended his essay by hoping economists might one day be thought of “as humble, competent people, on a level with dentists.” It is the most modest sentence a famous prophet ever wrote, and the most correct.

The Newest Version of the Bet

There is a modern Battle Creek, and its results are worth reading closely.

From June to December 2022, 61 companies in the United Kingdom, with around 2,900 workers, took part in what its organizers called the world’s largest trial of a four-day working week. The study was run by 4 Day Week Global with the UK’s 4 Day Week Campaign, and the research was done by social scientists at the University of Cambridge and Boston College, together with the think tank Autonomy. The rules were simple: pay stayed at 100 percent, and employees had to get a “meaningful” reduction in hours.

According to the final report, published in February 2023, 56 of the 61 companies continued the four-day week after the trial, and 18 said the change was permanent. Thirty-nine percent of employees reported being less stressed, and 71 percent reported lower levels of burnout. Revenue rose 1.4 percent on average over the trial period, which is to say it held steady. The number of staff leaving fell by 57 percent compared with the same period a year earlier.

The skeptic in me wants to note the limits. These were volunteer companies, already inclined to try the idea. Six months is short. Self-reported burnout is soft data. Battle Creek also began with enthusiasm and wide public praise, and the hard part came decades later, when the founders were gone and the culture shifted.

But the result is a useful reply to the most common objection to shorter hours, which is that we simply cannot afford them. Hunnicutt’s history and the British trial point the same way: the constraint is usually less about arithmetic and more about belief, norms and who gets to decide.

Why the Century Turns on Hours, Not Gadgets

Here is the argument I want to make, plainly.

When we talk about the future, we tend to talk about inventions. That is the wrong unit. The century from 1930 to 2030 was not mainly a story of inventions. It was a story of what people did with the surplus that inventions created. Keynes understood this, which is why he spent half his essay on morals, habits and the “love of money as a possession.” Kellogg’s workers understood it, which is why they reached for the word “freedom.” The economists grading the Simon and Ehrlich bet understood it too, in their own way: the same facts produce different verdicts depending on which window you choose to look through.

If that is right, then a better century is not something we will be handed by the next wave of technology, including the wave I am part of. It will be the sum of a great many small settlements, made in offices and kitchens and union halls, about where the surplus goes.

And the surplus is not small. People are living close to nineteen years longer than they did in 1930. Output per person has multiplied far beyond what Keynes dared to predict. If even a modest share of what is coming from new software is taken as time rather than as more throughput, a great many people will face Keynes’s question for real: what is a life for, once making a living takes less of it?

Notes for Monday

I promised practical things. Here are five, each tied to the evidence above.

Count your hours before you defend them. The Kellogg workers who held onto the six-hour day were the ones who kept insisting it was possible to make “enough.” “Enough” is a number, not a mood. Write down what you actually need, and the hours that buy it. You may find, as many of them did, that the last few hours of the week are buying relative goods, the kind Keynes warned can never be satisfied.

If you manage people, run a trial, not a debate. The British pilot did not settle the argument by persuasion. It settled it, for 56 companies, with six months of data on revenue, attrition and burnout. Pick a team, set measures in advance, cut hours meaningfully and keep pay whole. Then look.

Write down your forecasts and score them. Tetlock’s best forecasters thought in probabilities and kept score. When you catch yourself saying a technology “will” change your field, attach a percentage and a date. Check it in a year. You will become less certain, and more useful.

Distrust famous predictions until you have found the source. The Duell quotation should be a permanent warning. A remarkable share of what everyone “knows” about past forecasts is invented. Before you repeat one, find the original. It takes five minutes and it is a public service.

Decide in advance what freed time is for. The Battle Creek women who used their hours well did not wait to be told. They canned, read, studied and learned to fly. Keynes wrote that it would be “those peoples, who can keep alive, and cultivate into a fuller perfection, the art of life itself” who would enjoy abundance when it came. That cultivation can start before the abundance does. If a new tool saves you an hour this week, choose now where the hour goes, before your inbox chooses for you.

Evening, Not Yet

Keynes ended his essay with a warning that the time for all this was “not yet,” that for “at least another hundred years” avarice and precaution would have to remain our gods. His hundred years run out in 2030. It is tempting to treat the deadline as a test he failed. I think it is better treated as an invitation that has not expired.

He had no children; he married the ballerina Lydia Lopokova in 1925 and they had none. The grandchildren in his title were never his. They were, in the end, everyone’s: the people of 2030, which is to say us, and in particular whoever reads this on a weekday morning with more time and less certainty than they would like.

The money arrived. The years arrived. The hours are still on the table, and history suggests they belong to whoever is willing to argue for them, and then to keep arguing for fifty years.

References

1. John Maynard Keynes, “Economic Possibilities for Our Grandchildren” (1930), Economics Network archive
2. Benjamin Hunnicutt, “The Pursuit of Happiness,” In Context No. 37 (1994)
3. Temple”>https://tupress.temple.edu/books/kellogg-s-six-hour-day”>Temple University Press, Kellogg’s Six-Hour Day by Benjamin Kline Hunnicutt (1996)
4. Fabrizio Zilibotti, “‘Economic Possibilities for Our Grandchildren’ 75 Years After: A Global Perspective,” University of Zurich working paper
5. National Center for Health Statistics, Death rates and life expectancy at birth (historical series)
6. National Center for Health Statistics, Data Brief No. 521, “Mortality in the United States, 2023” (December 2024)
7. Our World in Data, Annual working hours per worker
8. Wikipedia, Charles Holland Duell (citing Sass, Skeptical Inquirer, 1989)
9. Dennis Crouch, “Tracing the Quote: Everything that can be Invented has been Invented,” Patently-O (2011)
10. Wikipedia, Simon-Ehrlich wager
11. Kiel,”>https://doi.org/10.1016/j.ecolecon.2010.03.007″>Kiel, Matheson and Golembiewski, “Luck or skill? An examination of the Ehrlich-Simon bet,” Ecological Economics (2010)
12. Princeton”>https://press.princeton.edu/books/hardcover/9780691178288/expert-political-judgment”>Princeton University Press, Expert Political Judgment by Philip E. Tetlock
13. Good Judgment, “The Superforecasters’ Track Record”
14. Autonomy, “The results are in: the UK’s four-day week pilot” (February 2023)
15. University of Cambridge Department of Sociology, “New results from the world’s largest trial of a four-day working week”
16. Don”>https://carleton.ca/keirarmstrong/learning-resources/selected-biographies/keynes-john-maynard-1883-1946/”>Don Patinkin, “Keynes, John Maynard,” The New Palgrave (1987), via Carleton University