01 / THE SAVED HOUR

The hour after the machine arrives.

Imagine finishing a day’s work by lunchtime. The work is done. The afternoon is still yours—or so you think.

A new tool has made that possible. Yesterday, the same task took eight hours. Today, it takes four. For a moment, the change seems wonderfully simple: the machine has given you something back. You could go home, read a book, spend time with someone, or follow the question you keep postponing. A piece of the day that used to belong to necessity has opened up.

Then someone looks at the schedule. If the task now takes half as long, perhaps you can do twice as many. Or perhaps the same work can be done with fewer people. Or the price can fall, customers can buy more, and the business can grow. All of these possibilities begin with the same improvement. They lead to very different afternoons.

This is a hypothetical scene, but it is the question I want to carry through this essay. When we say a machine saves labor, whose labor do we mean? When we celebrate the time it saves, who gets to decide what happens to that time? The tool changes the practical limits of a task. The arrangement around the tool determines how that change enters a life.

Economists have a useful word for the first part: productivity, the output produced for a given input. The second part is about distribution: how gains and costs travel between workers, owners, customers, and everyone else. We can measure an improvement in the first without establishing an improvement in the second. A faster process is an achievement. Turning it into a better life is another piece of work.

I find that distinction more useful than choosing between excitement and fear. It lets us take the machine seriously while still looking at the people around it. Over the next few chapters, I’ll follow one selective thread through industrial Britain and later economic thought, then return to AI, scientific curiosity, and the kinds of work we choose to reward. This is a reading path through an argument, rather than a complete global history of technology.

Our first stop is a small factory making an almost absurdly ordinary object: a pin.

02 / 1776 · THE PIN FACTORY

A small object. An enormous change.

Historical portrait of Adam Smith
Adam Smith1776 · The organization of work

In The Wealth of Nations, Adam Smith follows pin-making through its operations: drawing wire, straightening it, cutting it, forming the point, and attaching the head. Dividing these operations lets workers acquire skill and avoid switching tasks. Tools and coordination are part of the arrangement. [1]

His example makes the gain tangible. Ten workers together produce around 48,000 pins in a day—about 4,800 per person. An untrained worker operating independently, he writes, could not make twenty. This is a historical illustration of differently organized work, not an experiment isolating the effect of a machine.

Smith’s pin factory, per worker

PINS PER DAY · HISTORICAL ILLUSTRATION, 1776
Working independently, without trainingFewer than 20
With division of laborAbout 4,800
≈ 240×using 20 as the comparison ceiling
Smith describes ten workers producing around 48,000 pins a day together. The 4,800 figure divides that output by ten. He contrasts this with fewer than 20 per untrained worker operating independently. The small bar is drawn at the upper bound of 20 on the same zero-based scale. This compares different work arrangements; it does not isolate the effect of machinery. Source: Smith, Book I, Chapter I ↗
Inspect the numbers
Pins per worker per day in Smith’s illustration
Work arrangement Value Basis
Independent, untrained <20 Smith’s comparison
Division of labor ≈4,800 ≈48,000 ÷ 10 workers

It is easy to look at that long orange bar and feel that the argument is over. Here is the extra productive capacity, visible almost at a glance. But the graph has a deliberately narrow answer. It can show the difference in pins. It cannot show whether the workers are paid more, whether they have greater security, or whether the extra output allows them to leave earlier.

Imagine, for a moment, being asked to describe the factory from two positions. From the table where finished pins accumulate, the striking fact is abundance. From the worker’s place, the questions might include repetition, pay, and control over the day. Neither position makes the other disappear. They ask for different kinds of evidence.

That is the first turn in our story. Better tools matter, but so do the ways tasks are divided and people are brought together. A workplace is an arrangement of skills, instructions, expectations, and claims on the finished product. Changing one part of that arrangement creates possibilities elsewhere.

Once we notice this, the next question becomes harder to avoid. If the factory can produce so much more, what happens to the people whose old skills have become less necessary? The extra pins do not answer that for us.

03 / 1811–1867 · THE TERMS OF PROGRESS

The view from the other side of the factory.

By 1811, the question had acquired a sharper form in Britain’s textile districts. The workers remembered as Luddites protested machinery and changes to their livelihoods, including through machine-breaking. Letters, petitions, and notices preserved by the National Archives place disputes over pay and working conditions alongside the machines themselves. [2]

The convenient version of the story makes them people who failed to understand progress. The documents ask us to linger a little longer. A worker can understand that a new process produces more and still object to the terms on which it is introduced. An increase in productive capacity does not make every wage dispute a misunderstanding.

This is where the scale of the story matters. An economy can change over decades. A household needs to get through the week. Saying that new work will eventually appear leaves open its location, its pay, the skills it requires, and who can afford to wait. The transition is lived by particular people, even when its eventual benefits are described in totals.

Historical portrait of David Ricardo
David Ricardo1821 · A change of mind

In 1821, David Ricardo made a consequential revision to his own thinking. Adding a chapter on machinery to the third edition of On the Principles of Political Economy and Taxation, he acknowledged that he had previously treated labor-saving machinery as a general good with mainly transitional inconvenience. He now accepted that owners could benefit while the demand for workers fell. [3]

Ricardo did not conclude that machinery should be stopped. His revision broke a more specific assumption: the interests of different groups need not move together. An improvement in the revenues enjoyed by owners could coexist with a deterioration in the circumstances of laborers. The account of progress needed to ask who was being counted.

I find the change of mind as interesting as the conclusion. It models a way of discussing technology that we could use more often: keep the useful mechanism, examine the assumption that did too much work, and revise the claim. A machine can be productive without every consequence of its introduction being beneficial. We do not have to deny the first fact to investigate the second.

Historical portrait of Karl Marx
Karl Marx1867 · Ownership and the working day

Marx later pushed the question toward ownership and control. In his account in Capital, machinery under capitalist production serves the extraction of surplus value. A capacity that could reduce toil can instead be used to lengthen or intensify the working day. The distinction lies between what a tool makes possible and the purposes for which it is employed. [4]

That is Marx’s theoretical lens, rather than an uncontested law. Its value here is the question it makes visible. Who owns the productive equipment? Who sets the target? Who can refuse a new demand? A discussion centered only on the engineering achievement can leave those relationships out of the frame.

Smith, Ricardo, and Marx are not interchangeable voices in a single chorus. They direct our attention to different things: the organization of production, the possibility of divergent interests, and power inside the workplace. Read together, they keep us from letting one impressive number stand in for the whole experience of economic change.

By this point, the saved hour has become a contested object. It could appear as profit, a lower price, fewer paid hours, or a more demanding target. But there is another possibility worth taking seriously. What if the gain became time away from work?

04 / 1930 · ANOTHER KIND OF ABUNDANCE

What if progress bought us time?

Historical portrait of J. M. Keynes
J. M. Keynes1930 · Economic possibilities

Writing in 1930, amid economic crisis, Keynes looked roughly a century ahead. In Economic Possibilities for our Grandchildren, he imagined rising productivity and living standards making much less paid work necessary. He also warned of technological unemployment during the adjustment. His essay held the difficult transition and the possibility of abundance in the same view. [5]

The memorable image was a fifteen-hour working week. It was a conditional vision, tied to assumptions about growth and material needs. It was not a measurement of working life, and the calendar alone could not guarantee its arrival.

“Three-hour shifts or a fifteen-hour week may put off the problem for a great while.”JOHN MAYNARD KEYNES · 1930 · Economic Possibilities for our Grandchildren

Put that possibility beside Smith’s pin factory and the meaning of progress shifts. One image asks how much more a group can make. The other asks how much less of a life might need to be spent making it. Both concern productive capacity. They choose different ways to spend the gain.

Picture Keynes’s fifteen-hour week.

ONE POSSIBLE ARRANGEMENT · EACH BLOCK = 1 HOUR
15 hours of work25 other hours within this comparison
A visual interpretation of Keynes’s possibility, using a hypothetical 40-hour comparison window. This is neither historical work-hours data nor a claim about today’s average week. Keynes was looking roughly a century ahead under favorable assumptions. Read the essay ↗

The unfilled blocks in this picture are an invitation to imagine, not evidence of hours anybody actually recovered. Even so, they make a useful absence visible. Output has units that are easy to count. An afternoon spent caring for someone, wandering through a book, or simply resting may be harder to fit into the same account. That does not make it valueless.

I would not expect everyone to spend a saved hour in the same way. Some people need more income; others need rest or time for care. The interesting question is whether they have a choice. A shorter paid day with an unaffordable loss of income is a different arrangement from a shorter day that leaves someone secure.

Keynes’s possibility therefore brings us back to distribution. Productive capacity can make a different life feasible. Reaching it also requires decisions about income, working time, and access to the gains. The machine can open a door. It does not establish who is able to walk through it.

05 / PAUSE · FOLLOW THE ARITHMETIC

One tool. Three possible afternoons.

Before moving from factories to software, it helps to strip the problem down. Imagine a workshop making 100 units in eight hours. A new tool doubles its productivity. We will hold quality and staffing fixed and temporarily leave prices, wages, demand, and the cost of the tool outside the picture.

The arithmetic permits several arrangements. Keep the eight-hour day and make 200 units. Keep output at 100 units and finish in four hours. Or work six hours and make 150 units, taking half the maximum time saving and using the remaining gain for output. The same improvement supports all three.

1× · SAME AS BEFORE3× · THREE TIMES AS PRODUCTIVE
OUTPUT PER DAY200 units Scale: 0–300 units
HOURS WORKED8 hours Scale: 0–8 hours

Keep the eight-hour day. At twice the productivity, output doubles to 200 units.

Illustrative arithmetic, not a forecast. Output = 100 × productivity multiplier × hours ÷ 8. We hold quality, staffing, and tool availability fixed, and leave out costs, prices, demand, and wages. “A bit of both” takes half of the maximum possible time saving and uses the rest for output.

Move the control and choose a different use for the gain. The bars change because we have changed an assumption about what the workshop wants to do. There is no hidden prediction in the experiment. It is a small model that lets us see which part is arithmetic and which part is a choice.

The omissions become revealing as soon as we put people back into the workshop. Does the worker who finishes in four hours keep the same income? Are there customers for the extra units? Who paid for the tool? Who has the authority to choose between the buttons? Those questions could change which scenario is attractive or even viable.

That is why a productivity demonstration can be convincing and incomplete at the same time. It establishes a capacity under particular conditions. Turning that capacity into an outcome requires the rest of the arrangement. When we return to AI, I want to keep both parts in view.

06 / THE PRESENT · TASKS AND TRANSITIONS

What changes when the tool is software?

Consider a hypothetical researcher preparing an analysis. The job includes finding relevant work, checking data, writing code, interpreting results, and explaining why any of it matters. These activities fit under one job title, but they do not all present the same problem to a machine. Speeding up one can change the demands placed on the others.

David Autor’s 2015 essay offers a useful way to think about this. Automation can substitute for labor in some tasks while complementing human work in others. Lower costs and changed demand can also affect employment. That helps explain why the automation of tasks has not historically meant a matching disappearance of jobs. [6]

A historical comparison gives a sense of the scale of reorganization. Autor reports that agriculture employed 41% of the US workforce in 1900 and 2% in 2000. These are employment shares, not counts of people or measures of agricultural output.

Agriculture’s share of the US workforce

EMPLOYMENT SHARE · TWO HISTORICAL SNAPSHOTS
190041%
20002%
Values reported in Autor (2015), p. 5. Common zero-based scale: 0–100% of the US workforce. Two observations do not show the path between them or isolate automation from other changes. These are historical figures, not current estimates or a forecast for AI. Source: Autor ↗
Inspect the numbers
US workforce employed in agriculture
YearShare
190041%
20002%

The comparison should make us careful about both reassurance and alarm. It records a major change in where people worked. It cannot tell us that any particular transition was easy, or that another century of adjustment would look the same. The existence of work elsewhere is not a record of what happened to each person leaving an occupation.

Acemoglu and Restrepo make the task distinction more explicit. Their framework separates automation’s displacement of labor from the creation of new tasks in which people have a comparative advantage. Productivity can increase while labor’s share falls; new tasks can push in the other direction. The balance matters. [7]

For our hypothetical researcher, a tool might shorten the time spent producing a first draft of code. That leaves questions about whether the code is right, whether the data justify the analysis, and whether the problem was worth asking. It might make those judgments more valuable. It might also invite a higher throughput target. This example describes possible arrangements, not a measured effect of an AI product.

I therefore want a more specific conversation than whether AI will take “the jobs.” Which task is changing? What becomes easier to supply? Which skills become more useful? Does new work appear for the people who lose old work, and on what timescale? What happens to their bargaining power in the meantime?

History cannot settle the speed or distribution of AI’s future effects. It can improve the questions we ask of a demonstration. A system finishing a task is the beginning of an economic story. We still need to follow the people, the demand, and the rules around it.

And if some of the gain really does arrive as an open hour, another question begins: what might become possible in that space?

07 / THE OPEN HOUR · CURIOSITY

Who gets the room to think?

I keep returning to a use for the saved hour that is difficult to promise in advance: following a question whose value I cannot yet explain. Reading something outside my field. Looking again at a result that does not fit. Trying an approach that might fail. These are activities whose significance can become clear only after they have been allowed to happen.

That makes them awkward companions to a system that wants every hour assigned to a deliverable. We can make the assigned work faster and still leave the unassigned question untouched. The calendar fills again. The tool was effective; the conditions for curiosity did not change.

An illustrated reader at an open notebook, with a gear and hourglass on the desk, looking into a night sky that flows toward the page.
A little room for an unfinished question. AI-generated conceptual illustration.

The familiar story of Newton during the plague years often enters this conversation. During the disruptions of 1665–1666, he spent extended periods at Woolsthorpe, away from Cambridge, and developed foundational ideas of calculus. It is tempting to turn the interruption into a neat explanation of discovery. [8]

The history is more interesting with its preparation left in. Newton had already been studying mathematics intensely. His work continued over years, and Leibniz later developed calculus independently. The plague did not place a complete theory into an empty afternoon. This was a trained mind continuing an inquiry under unusual circumstances.

What I take from the story is a question about conditions. What happens when someone with knowledge, an unresolved problem, and time to pursue it can keep working beyond an immediate demand? The historical example does not answer the causal question. It helps us name the pieces we would need to examine.

One piece is cognitive offloading. A written note, a reminder, or a calculator can move some work outside our heads and reduce a task’s mental demands. Risko and Gilbert’s review examines this relationship between external aids and cognition. It does not show that delegating a calculation automatically produces a new theory. [9]

Another piece is an interval away from a problem. In a 2012 experiment, Baird and colleagues found that an undemanding task during a break improved performance on previously encountered unusual-uses problems more than a demanding task, rest, or no break. That condition also involved more mind wandering. It is evidence about a particular creative task, rather than proof that downtime produces scientific breakthroughs. [10]

The third piece concerns permission and support. Azoulay, Graff Zivin, and Manso compared life scientists supported by HHMI, with latitude to experiment and tolerance for early failure, with similarly accomplished NIH-funded scientists. HHMI investigators produced more high-impact work and explored more novel directions. This was an observational comparison of funding arrangements, not a randomized test of removing annual evaluations. [11]

The tool, the interval, and the institution therefore deserve separate attention. A machine can reduce routine effort. A pause may help with a problem already in mind. A funding or evaluation system can make an uncertain direction more or less affordable. None is a substitute for the other two.

SUPPOSE A TOOL SAVES AN HOUR

Where could that hour go?

REFILL THE SCHEDULE

More of the known work.

More tasks completed. The same questions, with a higher target.

PROTECT SOME OPEN TIME

Room for a different question.

Read across fields. Follow an anomaly. Try something that might fail.

Two possible uses, not a measured causal model or an exhaustive choice. Rest, care, and other work belong here, too. Open time makes exploration possible; it does not guarantee a discovery.

My working hypothesis is that reducing routine mental work may help discovery when people also have knowledge, time, and freedom to choose what comes next. I would want to recognize a well-explored dead end or an unexpected connection as part of serious work. Otherwise, we risk becoming very efficient at answering only the questions that already fit the form.

Here the economic argument returns. Open time is easier to use when a person is secure enough to tolerate an uncertain payoff. Asking who gets the saved hour also means asking who can afford to explore, whose failures are supported, and whose curiosity has to justify itself immediately.

08 / PURPOSE · BEFORE SPEED

Busy doing what?

There is a final complication. So far, we have asked how the gains from doing work more efficiently might be shared. But before we automate a task, we can ask what purpose it serves. If nobody can explain why a report is needed, producing it in seconds does not resolve the missing explanation.

James Suzman brings this cultural question into view. In the interview below, he asks why paid employment retains such a central place in our lives as machines become more productive. He argues that institutional expectations help sustain some work, and imagines more time for chosen activities: making, learning, and creating. These are his arguments in the interview, rather than established facts about every workplace. [12]

WATCH / ABOUT 6 MINUTES

Why so many people work “bullshit jobs”

James Suzman · Big Think · 2021

Watch on YouTube ↗ Read Big Think’s transcript ↗

I connect this to the question about evaluation. Could the effort spent proving that we are busy crowd out the effort needed to understand something? A tool could help remove that burden. It could also make the evidence of busyness easier to manufacture, until there is more of it to read and less time to ask what it means.

David Graeber’s “bullshit jobs” argument gave this concern a deliberately provocative name. In his 2013 essay and subsequent book, he began with jobs whose existence the people doing them could not justify. The judgment concerned the purpose of the job, rather than the worth of the person. [13]

The distinction matters, and so does the evidence against the wider claim. Soffia, Wood, and Burchell tested central propositions using European survey data. They found a low and declining share of employees describing their work as useless, challenging Graeber’s account of its prevalence and growth. Their alternative explanation emphasized working relationships and management. [14]

Those results are reports of experienced usefulness, not an objective inventory of the social value of every occupation. Still, they are a reason to keep the argument open. A compelling description of a feeling should not become a statistic merely because many readers recognize it. We can take the experience seriously and examine the explanation.

For this essay, I want to keep the question narrower: what about the tasks inside an otherwise useful job? Imagine a researcher documenting an experiment so another person can reproduce the method, understand a failure, or find an error. The record is valuable even when the experiment yields no breakthrough. Repetition and an uncertain outcome do not make a task pointless.

Now imagine the same information copied into another form that nobody uses for decisions, accountability, or learning. If that description is accurate, the extra step deserves examination. These are hypothetical examples, and usefulness can be indirect or emerge later. The point is to ask what the work enables before deciding that it should disappear—or be accelerated.

This is the question I want to bring back to AI. Can we use a tool to improve the work that matters, examine the work whose purpose is unclear, and protect some of the time recovered? That requires judgment about the institution as well as the task. Speed alone cannot tell us whether we are moving toward something worthwhile.

09 / RETURN · THE AFTERNOON IS STILL THERE

The story continues after the demonstration.

Return to the afternoon we began with. The work that took eight hours now takes four. We have followed that possibility through a pin factory, a dispute over livelihoods, an economist’s revision, a vision of leisure, and a question about discovery. The machine has remained in the picture. So have the decisions around it.

Productivity tells us that something has become possible with less effort. It does not tell us how the gain will be divided, whether a displaced person can find a new place, or whether recovered time is theirs to use. Those answers depend on ownership, demand, bargaining, and the institutions through which change travels.

I am interested in the tools because they can change what we are able to do. I am interested in the rest of the story because it determines what that ability feels like to live with. A faster working day might contain more output, more pressure, more security, or more freedom. We need to look beyond the demonstration to know which one has arrived.

And I keep coming back to the unfinished question: the one that does not yet promise a paper, a product, or a place on an evaluation form. If a machine frees an hour, perhaps some part of it could belong to that question. Making room for it would be a choice about what we value—and about who gets the chance to find out.

Your notes stay in this browser. Nothing is sent.

Have a different reading or a source I should see? Send a thought or correction ↗

10 / OPEN THE SOURCES

Don’t take the notebook’s word for it.

Start with the original arguments. Publication years below refer to the works, not the dates their digital copies were uploaded. Originally published October 4, 2026. Narrative revised and chart sources rechecked October 12, 2026.

  1. Adam Smith · 1776
    The Wealth of Nations, Book I, Chapter I ↗The division of labor and the pin-factory example. This is an account in a historical text, not a modern controlled study.
  2. Workers’ perspectives · 1811–1816
    Why did the Luddites protest? · The National Archives ↗Letters, petitions, and notices documenting conflict over machinery, wages, and working conditions.
  3. David Ricardo · 1821
    On the Principles of Political Economy and Taxation, Chapter 31: On Machinery ↗The chapter added in the third edition, revising his earlier position on machinery and workers.
  4. Karl Marx · 1867
    Capital, Volume I, Chapter 15: Machinery and Modern Industry ↗Machinery, surplus value, the working day, and the distinction between technology and its capitalist use.
  5. John Maynard Keynes · 1930
    Economic Possibilities for our Grandchildren ↗His essay on technological unemployment, material abundance, and a possible fifteen-hour week.
  6. David Autor · 2015
    Why Are There Still So Many Jobs? ↗Journal of Economic Perspectives 29(3). Why automation can substitute for tasks and complement human labor.
  7. Daron Acemoglu & Pascual Restrepo · 2019
    Automation and New Tasks: How Technology Displaces and Reinstates Labor ↗Journal of Economic Perspectives 33(2). A framework for displacement, productivity, and new tasks.
  8. Isaac Newton · historical record
    The Newton Project: Life and Work at a Glance ↗ Chronology of his studies, plague-year work, later publications, and Leibniz’s independent development of calculus. Read alongside Newton’s draft letter to John Wallis ↗, which revisits dates and acknowledges earlier mathematical reading. A historical case, not a causal experiment.
  9. Evan F. Risko & Sam J. Gilbert · 2016
    Cognitive Offloading ↗ Trends in Cognitive Sciences 20(9), 676–688. A review of how actions and external aids change a task’s cognitive demands; not a demonstration that offloading generates new theories. Author-hosted PDF.
  10. Benjamin Baird and colleagues · 2012
    Inspired by Distraction: Mind Wandering Facilitates Creative Incubation ↗ Psychological Science 23(10), 1117–1122. A laboratory experiment on breaks and creative problem solving; improvements concerned previously encountered unusual-uses problems. University research record and DOI.
  11. Pierre Azoulay, Joshua S. Graff Zivin & Gustavo Manso · 2011
    Incentives and Creativity: Evidence from the Academic Life Sciences ↗ RAND Journal of Economics 42(3), 527–554. An observational comparison using propensity-score weighting and difference-in-differences. Relevant to research incentives; it does not isolate annual appraisals or AI use. Author’s page, with paper and DOI.
  12. James Suzman · Big Think · 2021
    Why so many people work “bullshit jobs” ↗ The video that prompted this section. Suzman’s perspective on work culture, technology, and chosen activity. Publisher’s transcript ↗. Read as an argument to examine, not a settled empirical account of all work or scarcity.
  13. David Graeber · 2013 / 2018
    On the Phenomenon of Bullshit Jobs: A Work Rant ↗ The original 2013 essay, followed by the 2018 book Bullshit Jobs: A Theory. His 2018 interview ↗ explains the definition in his own words. Both are hosted in the David Graeber archive.
  14. Magdalena Soffia, Alex J. Wood & Brendan Burchell · 2021 / 2022
    Alienation Is Not ‘Bullshit’: An Empirical Critique of Graeber’s Theory of BS Jobs ↗ Work, Employment and Society 36(5), 816–840. First online in 2021; journal issue in 2022. Tests five propositions using European survey data and challenges the theory’s account of prevalence and growth.
Image credits & how the visuals were made

The cover and the illustration beside the curiosity chapter are AI-generated conceptual images, not an archival photographic record. Charts and diagrams are original HTML/CSS visuals; sources, units, and assumptions are stated next to them. Historical portraits come from Wikimedia Commons, locally hosted. They are shown in grayscale via CSS.

END OF FIELD NOTE 001

A little less fog, perhaps.

Take the question with you. Share it with someone curious.

Enjoyed this edition? You can buy me a coffee ↗. It keeps the notebook going.