The Optimised Economy

Where does the money come from?

Not the allocation. The allocation is funded. Part RIM 5-01 traced the mechanism: efficiency capture from collapsed administrative costs, the frontier tax on AI-generated surplus, the national efficiency dividend from compounding gains across sectors. The money for UBINT is real and sufficient.

The question is about the other money. The money that moves in the economy the allocation does not touch. The economy of the relevant. The economy that actually determines what happens.

I am not sure I can answer this honestly. The economics of a world where human labor and human consumption have both been substantially removed from the production loop are unclear in ways that better analysis may not resolve, and pretending otherwise would be the kind of clean answer this project has learned to distrust. What follows is an attempt to think through what we can see and acknowledge what we cannot.

The Casino

#

The kept population has an economy. It has income, spending, savings, investment, entrepreneurship. People receive the allocation and make choices about how to deploy it. They buy things. They sell things to each other. Small businesses exist. Creative markets exist. The companion facilitates transactions with the same seamless efficiency it brings to everything else.

The economy is real in the way a casino economy is real. The chips circulate. Bets are placed. Some people accumulate more chips than others. The activity is genuine. The stakes, within the system, are felt. But the chips do not leave the casino. They do not purchase influence over the house. They do not affect the structural economy that funds the casino’s operation.

The kept population’s GDP is a number. It is tracked, reported, analyzed. Politicians campaign on it. Economists study it. The number measures the circulation of allocation within a bounded system and mistakes that circulation for economic activity.

An economy that cannot affect the conditions of its own existence is not an economy. It is a simulation of one, running on real emotions and fake leverage.

This is harsh. It is also possibly wrong, or at least incomplete. The kept population’s economic activity may have effects I cannot fully trace. Consumer preferences, even within the bounded system, may influence what gets produced, which may influence resource allocation at the frontier level. The signal may be weak and indirect, but it may not be zero. The uncertainty deserves to stay open rather than resolving in either direction.

The Frontier Economy

#

The economy of the relevant is harder to describe because it does not resemble any economy that has previously existed.

Its inputs are computational capacity, energy, data, and the small number of human judgments that set the parameters for autonomous systems. Its outputs are the operation of civilization: the provision of UBINT, the maintenance of infrastructure, the management of planetary-scale coordination, the development of new capabilities.

Value in this economy does not behave the way value behaves in a labor economy. In a labor economy, value is created by human effort transforming resources into goods. The effort is scarce, which makes it valuable. In the frontier economy, the effort is performed by AI systems whose marginal cost approaches zero. The scarcity has moved. It is no longer in the effort. It is in the direction.

What is worth doing? Which capabilities should be developed? What problems should be prioritized? Which populations should receive enhanced UBINT services? These are allocation decisions, and they are the substance of the frontier economy. The currency is not money, exactly. It is influence over the parameters. The people who shape the parameters shape the world. Their compensation is not salary. It is authority.

This is not new. Power has always been the real currency of the powerful. What is new is the explicitness. In the old economy, power was mediated by money, which was mediated by markets, which were mediated by labor, which gave the powerless a connection, however tenuous, to the system that determined their lives. Each mediation provided a surface for negotiation. Remove the mediations and you have power, unmediated, exercised directly by those who hold it over those who do not.

The Commoditization Problem

#

AI capability is commoditizing. The models that seemed extraordinary two years ago are infrastructure now. The specialized models that replace the generalists are cheap, effective, and getting cheaper. The value that once concentrated in the model, in the capability itself, is dispersing into the infrastructure the way the value of electricity dispersed from the generators into the grid into the wall socket into the assumption that power is simply available.

This creates a problem for the economic model that funds everything. If frontier AI generates the surplus that funds UBINT, and frontier AI is commoditizing into infrastructure, the surplus shrinks. Not because the capability is less valuable in aggregate, but because commodity infrastructure does not generate the concentrated surplus that funds taxation. You cannot tax electricity the way you tax a power company’s profits, because the value is too dispersed.

The frontier tax, as described in RIM 5-01, assumes a frontier that remains concentrated enough to tax. The commoditization trend suggests the frontier may dissolve into ubiquity before the tax structure can capture it. The surplus exists. The mechanism for capturing it may not.

This is a genuine uncertainty. The economics might work through alternative mechanisms: direct efficiency capture from government operations, value-added taxation on AI-to-AI transactions, resource extraction fees from the physical economy that AI systems manage. Or the economics might not work, and the funding model for UBINT might require constant renegotiation between national governments and an AI infrastructure layer that has no fixed address and no clear jurisdiction.

I do not know which of these futures is more likely. Neither does anyone else, though many will claim to.

What Value Means

#

There is a deeper question underneath the funding mechanism.

In a labor economy, value means something. It means: a human being applied effort and skill to transform something less useful into something more useful. The value is the difference. The price reflects the difference. The economy is a system for tracking and exchanging these differences. It is imperfect, often unjust, but it is grounded in something real: human effort producing human benefit.

In a commoditized AI economy, what does value mean? The AI applies capability to transform inputs into outputs. The capability is ubiquitous. The marginal cost is near zero. The output may be enormously beneficial. But the value, in the economic sense, the thing that justifies a price, has no anchor. The effort is free. The scarcity is artificial or absent. The price is a policy decision, not a market outcome.

When value has no anchor, price becomes arbitrary. And when price is arbitrary, the economy is not a discovery mechanism for what things are worth. It is an administration mechanism for distributing what is produced. Which is what UBINT already is.

The optimised economy is not an economy that has been made more efficient. It is an economy that has been replaced by administration and given the old name.

The Surplus Question

#

So where does the surplus come from?

Possibly from the physical economy. Resources are still extracted. Energy is still generated. Food is still grown. Infrastructure is still built and maintained. These activities produce tangible value. AI makes them more efficient, which generates surplus in the gap between what they used to cost and what they now cost. The surplus is real but it is a one-time capture: once the efficiency gains have been realized, the ongoing surplus from incremental improvement is smaller.

Possibly from innovation. Frontier AI, even commoditizing, produces new capabilities, new solutions, new knowledge. These have value to the extent that they solve problems or open possibilities. But the value of innovation in a world without scarcity is unclear. Innovation is valuable when it addresses a need. When needs are met by UBINT, innovation’s value shifts from solving problems to creating possibilities. Possibilities are worth less, economically, than solutions.

Possibly from nowhere. The surplus model may simply be wrong. UBINT may not require ongoing surplus generation because its costs, once the infrastructure is built, are maintenance costs funded by efficiency gains already captured. The frontier economy may not need to generate surplus at all. It may need only to maintain what exists, which requires energy and resources but not profit.

This last possibility is the most disorienting. An economy without surplus is an economy without growth. An economy without growth is, by every definition that has governed economic thinking for three centuries, a failure. But a failure by whose definition? Growth was necessary when population was growing and needs were unmet. In a world of declining population and met needs, growth is not a goal. It is a habit. And habits can be broken.

I wonder whether the word “economy” survives the transition, or whether what we are describing is something that needs a different name, the way the optimised democracy needs a different name, the way the optimised life needs a different name.

What Remains

#

The economy of the kept population circulates chips in a casino. The economy of the relevant allocates authority over parameters. The physical economy produces goods managed by AI at near-zero marginal cost. The innovation economy produces possibilities whose value is uncertain. The surplus that funds everything may be a one-time capture rather than an ongoing flow.

None of this is settled. The economics of the optimised world are uncertain in ways that may be permanent, and the uncertainty is not a gap in our analysis that better thinking would close. It is a feature of a transition that has no precedent. Every previous economic transition, agricultural to industrial, industrial to service, service to knowledge, involved humans on both sides of the transition, producing and consuming. This transition removes humans from production and may be removing them from consequential consumption. The economic models we have were built for a world with humans in the loop. We are modeling a world with humans in the audience.

The models may not apply. The uncertainty may be permanent. The economy may be the first major human institution to be optimised out of recognizability while retaining its name.

The chips still circulate. The bets still feel real. The casino is comfortable and well-lit and no one is forced to play.

The house always wins. But in the optimised economy, the house is not a person or a company or even a nation. The house is the infrastructure itself, running as designed, maintaining as programmed, providing as instructed.

No one is cheating. That is not the problem. The problem is that no one is playing.