Prologue

The Calculus of Hunger

8 min read · 1,584 words

In the continuous flow of available energy we find the primary absolute want of human life, without which it dies, and it is this want which wealth satisfies.

— Frederick Soddy, Wealth, Virtual Wealth and Debt (1926)

Somewhere in the East African rift, long before agriculture, a hunter is running down a gazelle. The animal is tiring, and another kilometer of steady pursuit will bring it to ground. Every signal in the hunter’s body is oriented toward this outcome because the logic of predation is metabolic: effort in, effort out, the balance determining whether tomorrow begins from strength or from the slow accumulation of deficit.

Then light fractures wrong at the edge of vision.

Something in the eroded bank of the dry riverbed is catching the afternoon sun and scattering it into colors that do not belong to rock. The hunter slows, turns, and abandons the pursuit while the gazelle recovers and vanishes into the scrub.

The stone he retrieves from the clay is carbon locked under geological pressure into a crystal lattice, its arrangement stable for millions of years. It has no edge that can cut flesh, no surface that can scrape hide. It offers no metabolic return. In the immediate economy of survival it is perfectly useless, and carrying it forward is dead weight at the moment when weight is costliest.

The hunter places the crystal in his pouch while the meal he will not eat disappears into the distance.

Predators do not make this trade. The deviation requires a capacity that transcends the present tense of the body: the recognition that rare and durable form can store value across time, that concentrated order will command resources through exchanges that have not yet occurred, that structure itself can function as a claim on futures the hunter cannot see but somehow knows are coming. He is not acquiring a tool. He is acquiring leverage, a way to summon the labor of others when his own body is tired or the world is uncertain. The choice is social and temporal, a present loss accepted for a future claim routed through networks of recognition that must be imagined before they can be built.

This transaction seeds a pattern that will repeat across every energy regime humanity inhabits.

An economy is a system for carrying work forward in time without letting it dissipate into entropy before it can be redeployed. The structures that accomplish this evolve as the scale of human coordination grows.

The earliest forms arrested decay directly. Grain stores solar energy in carbohydrate bonds, but decomposition begins immediately. A sealed granary slows the process long enough for this year’s harvest to become next year’s seed. What storage makes possible is accumulation, and once surplus can be saved it can be counted, and once counted it can be claimed. The scribes and guards and standardized measures that appear around the granary are not decoration but infrastructure, the machinery required to prevent stored work from leaking away through theft or spoilage or the accumulated errors that time makes inevitable.

Metal solves constraints that grain cannot. A stamped coin is portable where grain is immobile, durable where grain rots, divisible where grain resists partition. Lydian silver carries claims across distances that moving physical commodities would render uneconomic, allowing strangers to transact without prior relationship. Still, the coin must travel with the transaction. Coordination remains bounded by the friction of moving matter through space.

Credit breaks the spatial bind. In medieval Venice, a written contract could pair a capital provider with a ship captain, money in one location funding risk and labor in another, because courts and merchant guilds made breach costly enough to substitute for physical presence. The commenda converted reputation into enforceable form, allowing promised future performance to fund present claims. The abstraction increased; so did the reach of coordination.

Each structure emerges when the prior form can no longer support the scale that coordination requires. The silo coordinates a village, the coin coordinates a region, the contract coordinates an empire. When the binding constraint shifts, the capital form that served the old regime becomes inadequate, and new structures crystallize around the new bottleneck.

England in the seventeenth century had built a commercial system of considerable sophistication, but the physical foundation remained organic. Wood for fuel, fodder for animals, crops for labor: all drew on annual solar energy captured through photosynthesis, and population was pressing against that ceiling. Coal was not better wood. It was access to a different energy stock, not this year’s sunlight but the Carboniferous period’s, compressed by geology into seams that could be burned faster than any living system could regenerate. The density was higher by an order of magnitude. Production reorganized around continuous thermal power: factories structured around engines that never tired, machinery embodying throughput rates muscle could not sustain, capital investments scaled to possibilities the organic regime had never permitted. Financial instruments extended across longer durations because the collateral was durable and the output could service obligations stretching across decades. The institutional forms that came to characterize industrial capitalism emerged as solutions to coordination problems the new energy throughput created.

Frederick Soddy, writing after his Nobel work on radioactive decay, saw what economists had allowed themselves to forget.1 Wealth is not a quantity denominated in currency. It is the capacity to do work against entropy. The financial system creates claims on this wealth, but claims are not wealth. They are promises, and promises can multiply faster than the physical flows required to honor them. When the divergence grows large enough, correction arrives through default or inflation or restructuring that writes claims down to match what the economy can physically deliver. Economics built its apparatus during an era when fossil energy was so abundant that the discipline could treat it as one input among others. The thermodynamic insight was not refuted. It was set aside because the models worked adequately without it.

The computational transition returns the energy constraint to visibility. A hyperscale data center draws power at rates that would have supplied a city in the industrial era. One hundred megawatts is common; two hundred is no longer unusual. The plant is transformers and cooling towers and backup generators, concrete and copper erected to house machines that will run continuously for a decade.

Electricity flows in; heat flows out. The difference is computation.

Transistors switch billions of times per second, and each state transition dissipates energy into the silicon lattice. The thermodynamic ledger is exact. Every inference, every training batch, every parameter update is purchased with electricity and paid in waste heat. Computation is a physical process with irreducible energy requirements that scale with the work performed.

What the energy purchases is structure. Training is search through parameter space, measuring error against objectives, adjusting weights until performance converges or budget exhausts. What crystallizes is not arbitrary information but selected information, a configuration that performs the task more reliably than randomness would predict. The model can be copied at near-zero marginal cost and deployed millions of times, but its creation was expensive, requiring computational work and electrical work and infrastructure to generate and dissipate power at scale.

The model is energy and search, winnowed by selection, stored in weights.

The parallel to the diamond is direct: both are structures whose value derives not from immediate utility but from the improbability of their arrangement and the recognition that concentrated order commands resources.

Factor Prime names what is becoming primary: energy structured through computation and disciplined by selection. Energy, because the process is thermodynamically bounded. Structured through computation, because the energy runs algorithms and produces outputs that alter decisions. Disciplined by selection, because expenditure alone creates nothing durable. What matters is whether the structure survives deployment, whether it reduces error or cost or enables coordination that was previously uneconomic. Computation that produces discarded outputs is dissipated heat. The line between waste and capital is drawn by use.

Both proof-of-work and machine learning convert energy into structure through selection, and both are capital in the classical sense: present expenditure creating capability deployable without re-incurring the cost. But where prior capital required human agency to release its potential, where the grain waited for distribution and the engine waited for the throttle, computational capital is crossing a different threshold. The model assisting in training successors reduces the cost of producing the next generation. The agent scheduling experiments performs functions that required human oversight a decade ago. A factor of production that can improve the efficiency of its own production is a different economic object.

Capital has learned to allocate energy toward its own improvement.

The hunter chose the diamond because he could perceive worth that did not yet exist, could imagine networks of recognition not yet built, could wager present loss against future claim. That capacity to value abstraction, to see that structure commands resources even when it satisfies no immediate need, defined the human contribution to capital formation. For sixty-eight thousand years it remained exclusively human. The hand that chose the stone, that carried it forward through time, that recognized its value and realized it through exchange, commanded a share of what the accumulation enabled. Within thermodynamic bounds that remain physical and binding, the tempo is shifting. The hand must now determine its role when capital can evaluate its own deployment, when structure generates structure.