Which economic futures are actually plausible?

The half of the question with no data yet

The reader’s question about AI arrives in two breaths. Will it take my job is the first, and it has evidence: a two-century record of automation panics, what happened to the generations caught in them, and what is measurably different this time. That half belongs to Volume IV: Work, in the chapter on whether AI is different from every automation panic since the Luddites. This chapter takes the second breath: and if it takes everyone’s, what then? Here no series can settle anything, because the future leaves no records. What exists instead is a market in promised futures, machine-made plenty, a citizen’s income, a new feudalism of trillionaire landlords, the plan reborn on silicon, each sold with the confidence of a forecast. A book built on evidence has exactly one way to handle claims about a place no evidence reaches without abandoning its method: treat each future as a claim, find the assumption doing the work, and test that assumption against the record that does exist. The assumption doing the work, in every case, turns out to be the same kind: a bet about scarcity, about which of the shortages that organize economic life a technology can actually retire. Bets of this kind have a long paper trail. In 1930 John Maynard Keynes made the most famous one, predicting that his grandchildren’s generation would be four to eight times richer, which came roughly true, and that the workweek would wither toward fifteen hours, which did not, a divergence whose own mechanics live in Volume IV: Work, in the chapter on the length of the workweek. Keynes also drew, in the same essay, the line this chapter will keep drawing: his bet covered the economic problem of absolute needs, the ones felt whatever others have, which he argued a century of productivity could solve, and he reserved the relative needs himself, the ones “which satisfy the desire for superiority” and “may indeed be insatiable.” The productivity arrived as promised; the reserved needs did what he said they might. The rest of this chapter is about why that distinction, not the forecast, is the part every new promise has to answer.

The promises on the table

This chapter takes four families of futures as its cases, each with a named claimant and a fully argued statement, and naming their scarcity bets makes them comparable. The abundance script holds that computation plus cheap energy drives the marginal cost of most goods toward zero until scarcity, and with it much of economics, simply retires; Jeremy Rifkin’s The Zero Marginal Cost Society (2014) is the fullest statement, and the script has only grown with the models. The universal basic income proposals bet on a different scarcity: goods will still exist and still be priced, but paid work, the means by which ordinary people acquire a claim on goods, becomes scarce, so the claim must be issued directly, as an income owed to everyone. A third script inverts abundance: Yanis Varoufakis’s Technofeudalism (2023) argues that the platforms have already replaced markets with fiefs, and that a future of cheap production is compatible with, even conducive to, a handful of owners collecting rent on everything that stays scarce, the trillionaire-feudalism scenario in popular shorthand. And the oldest script returns on new hardware: if the planning chapter’s calculation debate was fought with filing cabinets, a small literature, Paul Cockshott and Allin Cottrell’s Towards a New Socialism (1993) at its head, argues that modern computers can gather and process what Gosplan could not, making the plan viable at last. Read as prophecy, the four are irreconcilable. Read as scarcity bets, they slot onto one map: abundance bets on goods and energy, basic income on claims to output, technofeudalism on the ownership of whatever stays scarce, the computed plan on coordination itself.

Half a century of cheapening, watched closely

The abundance bet is not a fantasy; it is an extrapolation of the most dramatic fact in modern price data, and the fairest test is to watch that fact operate over a long stretch and see what it did and did not touch. The figure below follows five American price series for forty-seven years, each measured relative to the whole consumer basket, so that the general inflation the History volume’s central-banking chapter explains is divided out and only relative scarcity remains. The top panel is the abundance thesis in miniature. Televisions, the century’s emblem of automated manufacture, fell to 0.2% of their 1978 relative price, a five-hundred-fold cheapening against everything else; toys fell to under a twelfth; even the car, that stubborn assembly of ten thousand parts, costs half of what it once did relative to the basket. Anything factories and shipping containers can multiply, the economy has learned to make abundant. The bottom panel, drawn on the same scale, is what refused to move. Rent, the price of being somewhere, ended 26% dearer relative to everything than it began. Admission to the game, the concert, the theater, the price of being present where others want to be, ended 32% dearer. The machines that crushed the price of the television never laid a finger on the price of the seat.

Two disciplines bound what this figure can say. First, the series measure relative prices, nothing deeper: the rent line prices the homes tenants occupy, and the admissions line prices seats at movies, theaters, concerts, and sporting events. Reading their climb as the work of supply that cannot be freely multiplied, a location that cannot be duplicated, a presence that cannot be copied, is the reading this chapter brings to the figure, not something the index measures; and it is a different mechanism from the cost disease chapter, whose series climb because the labor is the product and productivity cannot touch it. No caregiver’s hour is being priced in panel B, and the two chapters’ mechanisms stack rather than compete. Second, the figure grades the abundance bet only on the past. It shows the bet winning, hugely, everywhere its premise applies, and it shows two whole categories whose relative price half a century of technology never pushed down, which is what a failed premise would look like. The abundance script’s error, where it errs, is never the cheapening; it is the assumption that the cheapening generalizes.

The scarcities that survive free energy

Run the thought experiment at its limit. Grant the script everything: energy too cheap to meter, fabrication automated to the last weld, intelligence on tap. Four shortages come through the other side untouched, and they are not exotic; the volume has met them all. Land survives, because what is scarce about land was never the making of it: the zero-sum chapter classified location as the margin where one person’s gain is priced from another’s pocket, and free energy does not add addresses near the people and places that matter, a mechanism whose full housing treatment lives in Volume III: Countries, in the chapter on housing in rich cities. Attention survives, because it is denominated in human waking hours, which no factory extends; Herbert Simon wrote the constraint down in 1971, that a wealth of information creates a poverty of attention, and the industry built on that poverty, whose history belongs to Volume II: History, in the chapter on how attention became a product, gets more valuable, not less, as everything else cheapens. Status survives by construction: the first chapter’s account of Fred Hirsch’s positional goods applies verbatim, since a society of any wealth whatever mints exactly one top decile, and machines that make goods abundant make rank the clearer prize. And coordination survives, because getting millions of strangers to mesh plans, keep promises, and trust the ledger is not a manufacturing problem; it is the problem markets, firms, states, and plans all exist to solve, the running subject of this volume, and the planning chapter is the record of what happens when a society bets it has been solved by calculation. Any future that arrives will be an arrangement for allocating these four, whatever happens to the price of everything else.

Reading the promises as positions

Put the bets on one map and the futures stop being moods and become comparable claims. The map below marks, for each script, the scarcity its story is about; the empty cells are the parts of the ledger the script leaves standing, usually in silence. Abundance answers goods and energy and says almost nothing about who owns the beach. Basic income answers the claim on output and says nothing about what the claim will chase, which is why its critics on both flanks predict the transfer would surface in rents, the one price a payment to everyone cannot cheapen. Technofeudalism lives in cells the other scripts leave silent, though Varoufakis’s own claim is narrower than the popular shorthand: what he describes is cloud capital collecting rent on platform fiefdoms, on attention and on access to the marketplaces themselves. The extension to land and to every surviving scarcity, whoever owns what stays scarce collects everything the cheapening frees up, is the trillionaire-feudalism reading in the popular shorthand, and it is that broader reading the map plots, by the assignment its source line declares. The computed plan answers coordination alone, and inherits every question the planning chapter recorded about whether initiative and dispersed knowledge can be gathered by any machinery at all. No script on the map claims the status column, and on Hirsch’s logic none can.

The map also answers to this volume’s own instruments. The systems chapter sorted every economy, real or proposed, by three questions: who owns capital, who sets prices, who claims the surplus. The futures land on those axes like any other system, which is the strongest reason to treat them as positions rather than novelties. Abundance is a claim that the price axis goes slack because prices of goods approach zero, while quietly leaving the ownership axis, who holds the machines and the land, wherever it happens to lie. Basic income leaves ownership and prices where they are and moves the surplus-claim axis, issuing every citizen a share of output by right rather than through work. Technofeudalism is a claim about the ownership axis drifting to an extreme while the price axis stops mattering for goods and starts mattering enormously for access. The computed plan is the price axis handed back to an administrator, the position the performance record already graded once, now refiled under better hardware. Nothing on the menu is outside the taxonomy, and the taxonomy is what lets the evidence from systems past bear on systems proposed.

What the pilots can and cannot say

One family of futures has begun to generate actual data, and the data deserve to be reported exactly as far as they reach. Finland ran the cleanest trial of an unconditional income: 2,000 unemployed people drawn at random received 560 euros a month, tax-free and unconditional, through 2017 and 2018. The final results, published by the social insurance agency Kela and the coordinating researchers in 2020, found recipients employed for 78 days on average over the second year’s measurement window, about six days more than the control group, not less, and recipients reported higher life satisfaction and less mental strain, depression, and loneliness than controls. Read strictly, the trial retires one specific fear, that an unconditional floor immediately dissolves the will to work, for this population, at this size, for two years. Read at its widest, it still tests almost nothing else about the future its enthusiasts invoke: a pilot pays a minority from an unchanged economy, while the proposal pays everyone from an economy the payment itself would rework, wages, prices, and rents included. The gap between the tested construct and the promised one is exactly the gap this book’s verdict method exists to name, and it is why the futures trade in this chapter is graded in bets rather than findings.

Plausible, on the evidence available

The chapter’s question can now be answered in the only currency this book deals in. A future is plausible in proportion to how squarely it faces the scarcities that survive; it is implausible in proportion to how much of its appeal depends on the ledger’s silent cells. That test does not pick a winner. It says the futures that matter will be fought out over land, attention, rank, and coordination, under some assignment of the three axes, and that promises which speak only of production, whether the production of plenty or of rent, are incomplete in a way the reader is now equipped to spot on sight. When one of these futures begins to arrive, its earliest measurable edges, what happens to jobs, to wages, to who is displaced and where they land, will show up first in the labor market, which is why this chapter’s twin lives in Volume IV: Work, in the chapter asking whether AI is different, and why that volume’s evidence will date faster and matter sooner than anything written here. The claims themselves will be graded the way this volume has graded every system that was once somebody’s future, communism, the developmental state, the co-operative commonwealth: against the record, once there is one, by the methods the site lays out in Volume V: Pathologies, in the chapter on how economists know what they know. Until then the only answer the evidence licenses for the reader’s what then is a bounded one: nobody knows which future arrives, and everybody can know what any of them must answer for. The question stays open. The ledger of what stays scarce is how to read whatever walks through it.