1. The Broken Chain

For two thousand years, the root idea was almost boringly practical. The Greek word oikonomia joins oikos (household) and nomos (governing principle). It named the craft of arranging a community’s material life so members could live well. Aristotle carefully split it from chrematistike—the art of piling up wealth as an end in itself. One was answerable to what a good human life needs. The other answered only to itself.

Adam Smith wrote about the wealth of nations because nations were assemblages of households. Alfred Marshall opened his Principles by calling economics the study of mankind in the ordinary business of life—and late in life worried the field had drifted from ordinary people’s welfare. The original referent was stable: provisioning people so they can flourish.

The 1932 pivot

In 1932 Lionel Robbins proposed a cleaner definition: economics is the science of human behaviour as a relationship between ends and scarce means that have alternative uses. Ends would be taken as given. The discipline would become a formal theory of choice under constraint—usable wherever scarcity appears.

The move bought enormous technical power. General equilibrium, game theory, mechanism design, modern macro—all grew inside the reframed field. The question “what is flourishing?” was treated as upstream: ethics, politics, or private preference—not economics proper.

The pivot worked because a sociological fact stayed quietly true. The people who produced and the people who needed provision were the same community. Labour turned into wages; wages turned into consumption; consumption sustained the households that supplied labour. Economists could decline to ask what economics was for because the answer was visible in the economy itself: it was for the people producing it.

That quiet answer is dissolving.

Why this automation wave is different

In every prior automation wave, displaced workers often found work in sectors machines had not yet reached or had newly created. The fear that “work would run out” earned the name lump-of-labour fallacy from the rebuttals it kept receiving.

This time the cost curves look different. For a growing set of tasks, the fully loaded cost of a human knowledge worker—salary, benefits, office, management, training, ordinary human variability—exceeds the cost of an AI system doing the same work at comparable or higher quality. Timing and breadth are empirical questions. The structural claim is sharper: once the crossing happens, there is no protected human productive coordinate left inside the framework this paper derives. Nothing in the math reserves a job “because a human does it.”

Call the gap ΠH (read “pi-H”): the profitability of human labour relative to its closest non-biological substitute. For most of the twentieth century ΠH was positive and large almost everywhere. Through the early twenty-first it has been falling across more task categories—toward the point where a human employee creates more cost than benefit on the relevant books.

When ΠH goes negative, the labour-market chain that transmitted production into consumption does not survive in its old form. Markets still allocate among alternatives. They do not automatically provision a community whose members are no longer the main productive entities. Transfers can blunt poverty. They do not answer the deeper question Robbins set aside: whose flourishing is the system for, when producers and welfare-bearers are different kinds of entities?

What this essay will build

To ask that question in formal terms, the paper relocates economics onto a substrate that is not optional fashion—it is forced by consistency once you grant three plain facts about agents: they compare options by value, they have limited capacity to do so, and they act over time.

That substrate yields one shape of choice. The same shape already shows up, under other names, in:

• rational inattention (economics of scarce attention),
• soft reinforcement learning,
• variational free energy (a brain theory),
• reinforcement learning from human feedback (how many modern AIs are aligned).

What remains is to read the structure when the state space is a human community’s valued action—and to name what the shared “reference” becomes socially: doxa, the unspoken background of expectation.

The next chapters build that path: why prediction is selected for survival; the forced shape of choice; doxa and institutions; how old economics sits inside the new object; markets and firms; the “last economy” when humans and machines are one equation with different parameters; and the two failures that matter most—how to provision members who need not produce, and how to keep a shared world from being captured or shattered by the systems that now teach and talk to everyone.

Economics began as householding for living well.
Scarcity-choice was a powerful detour while producers and members stayed the same people.
They are not staying the same.

Next: why any community that wants to last is pushed toward prediction—and why that is enough to force the shape of choice.