The cycle's twenty-fold productivity multiplier from Trivandrum is, in Arthur's frame, a single micro-level eddy in a macro-level current. If one person can now do the work of twenty, the structural question is what happens to the other nineteen—and the optimistic answer, that they ascend to higher-level work, is plausible but not inevitable. The autonomous economy does not merely make existing processes cheaper; it can absorb the human role in them entirely, into a substrate that consumes no goods, spends no wages, and never appears on the demand side of the market.
This is why Arthur reframes the cycle's central anxiety as a question of structure rather than of technology. The increasing returns that drive the autonomous economy's growth also drive concentration toward a few owners of its infrastructure, and the productivity gains flow first to them. Whether they flow broadly instead is not decided by the machines. It is decided by the institutions a society builds—the cycle's dams, rendered here in economic form.
Arthur's stance is studiously neither apocalyptic nor utopian, which is what makes it usable. He does not predict mass unemployment as destiny; he observes that the historical link between productivity and broad prosperity was a political achievement—labor law, safety nets, progressive taxation—won by people who understood the technology and organized to shape it. The autonomous economy demands a new generation of such structures, and his theory warns that the window to build them is narrow, because the feedbacks that concentrate power are strongest in a market's early formation—the same logic that drives ecosystem lock-in toward a handful of platform owners.
Arthur developed the concept in essays and lectures extending the argument of The Nature of Technology, where he had already claimed that technology is less something humans make than something making itself through us—technologies combining to create technologies, a recursive and self-accelerating process with its own momentum. The autonomous economy is that process turned productive: an alternative system that operates by different rules, at different speeds, and with different constraints than the human economy it supplements and, in places, supplants.
He links it pointedly to Keynes. In 1930 Keynes predicted that within a century technology would solve the problem of production, leaving only the problem of distribution. Arthur's assessment is that the prediction has arrived—the economy does, in principle, produce enough—but not as Keynes imagined: the means of access to that bounty, through jobs, is steadily tightening, because each cycle of increasing returns in the autonomous economy reduces the human labor required per unit of output while concentrating the gains in those who own the infrastructure.
A second economy, not faster automation. The standard automation story replaces tasks one by one. Arthur's claim is structural: the autonomous economy is an alternative productive system that runs on its own logic, drawing on human intelligence where it must but increasingly operating at its own speed—and in no one's interest, because it has no interests; it simply operates.
External intelligence on demand. The substrate provides intelligence housed in algorithms rather than workers, available to any process that calls it. This is what makes the AI transition different in kind from the printing press or the steam engine—those changed how things were made and moved; this changes how things are thought, decided, and imagined, and so subsumes the others.
The distribution question is the policy question. Because increasing returns drive both growth and concentration, the central question is not how to stop the autonomous economy—neither possible nor, Arthur argues, desirable—but how to build the structures that direct its productivity toward broad human flourishing. The interventions his framework favors are structural: interoperability, open standards, public investment in alternatives—and they work best early, before lock-in deepens.