Hayek for the Age of Agents

Friedrich Hayek is usually placed in the wrong drawer.

He is treated as a free-market economist. A political symbol. A name people invoke when they want less government, more markets, or a certain kind of classical liberal respectability.

That version is too small.

Hayek matters because he understood something deeper than economics: reality is not available to a single mind.

That is the wound at the center of his work. Knowledge is not sitting somewhere in complete form, waiting for the right expert, ministry, dashboard or model to collect it. The knowledge that makes society work is dispersed, local, partial, embodied, temporary and often impossible to write down.

A shopkeeper knows which products people ask for but never buy. A trader feels when liquidity is not really there. A nurse knows which patient is quietly getting worse. A founder senses that a customer is lying politely. A parent knows the difference between a tired child and a child who is about to break. None of this appears cleanly in a central database. Much of it is not even conscious until the moment demands it.

Hayek saw that this was not a minor inconvenience. It was the whole problem.

In The Use of Knowledge in Society, he wrote that the economic problem is not merely how to allocate "given" resources. The deeper problem is how to use knowledge that is "not given to anyone in its totality." That sentence should be read less as economics and more as metaphysics.

No single brain owns the world.

the central planner is a fantasy of cognition

Central planning fails for reasons deeper than ideology.

It fails because it imagines society from the wrong viewpoint. It assumes there can be a place outside the living system from which the system can be seen, known and directed. It imagines that enough data, expertise and calculation can replace the distributed perception of millions of people acting inside reality.

This is the same mistake that appears again and again in modern life.

A government thinks a policy dashboard can see the economy. A corporation thinks a KPI stack can see the company. A platform thinks engagement metrics can see human desire. A school thinks test scores can see intelligence. A health app thinks biomarkers can see vitality. A manager thinks status updates can see work.

Sometimes the measurements are useful. Often they are necessary. But they become dangerous when the map forgets it is a map.

Hayek's critique was not anti-intellectual. It was anti-pretence.

His Nobel lecture was called The Pretence of Knowledge. The title alone is enough for our age. He warned that in complex social systems, the facts we can measure are not necessarily the facts that matter most. Worse, because measurable facts are easier to handle, institutions begin to treat them as the only facts that count.

That is how intelligence becomes stupidity at scale.

The central planner is not only a bureaucrat in an office. It is any system that believes reality becomes manageable once it has been abstracted into its own categories.

markets as contact with reality

This is where Hayek connects with my own work.

In Truth Is Not a Thought, I wrote that truth is contact. Not an idea floating above life, but the moment reality pushes back. The body pushes back. The market pushes back. The road pushes back. A relationship pushes back. The world answers.

Markets matter because they are one of the ways reality answers.

Not because money is sacred. It is not. Not because markets are morally pure. They are not. Markets can be manipulated, captured, distorted, gamed and made cruel. Anyone who turns markets into religion has stopped thinking.

But the price system does something extraordinary when it works. It compresses countless fragments of local knowledge into signals that people can act on without needing to understand the whole system.

A price is not just a number. It is a scar. It is a message from scarcity, desire, risk, timing, trust, error and opportunity. It carries information from people who will never meet each other, who do not share a plan, and who may not even understand the wider consequences of their own actions.

That is why markets are discovery systems before they are allocation systems.

They discover what people value. They discover where supply is fragile. They discover which stories survive contact with reality. They discover when a beautiful plan cannot survive cost, logistics, timing or human behavior.

This is why I keep returning to leverage and sovereignty. In No Leverage, No Sovereignty, the point was not that power is good in itself. The point was that without real leverage, moral language becomes theater. You can have the right values and still lose the world if you refuse to build the means through which reality can respond to your intent.

Hayek gives that intuition an epistemic foundation.

Sovereignty is not control from above. Sovereignty is the ability to participate intelligently in a world no one fully owns.

why AI makes Hayek more important

It is tempting to think AI weakens Hayek.

After all, what if the old knowledge problem was only a compute problem? What if central planners failed because they had bad information, weak models and slow bureaucracy? What if modern AI can finally see enough, predict enough and optimize enough to make distributed discovery obsolete?

This is the great temptation of the AI age.

Central planning with better dashboards.

A machine that sees everything. A model that knows everyone. A system that predicts demand, allocates resources, manages behavior, optimizes health, routes labor, controls capital, writes policy and corrects human irrationality.

The dream is old. AI only makes it look modern.

But Hayek's point was never merely that calculation was difficult. His point was that the relevant knowledge often comes into existence only through action, friction and feedback. It is not sitting there, waiting to be harvested. It is discovered by people trying, failing, adapting, risking and noticing.

This is why AI should make us more Hayekian, not less.

The danger is not that AI becomes intelligent. The danger is that institutions use AI to revive the fantasy of a single mind.

A large model can summarize a market. It cannot be the market. It can analyze a company. It cannot replace the lived knowledge distributed across customers, employees, founders, suppliers and edge cases. It can help a government see patterns. It cannot possess the tacit knowledge of a society.

AI is powerful when it amplifies distributed intelligence.

It is dangerous when it pretends to replace it.

agents, not overlords

This is the practical frontier now.

AI agents can become the next layer of distributed coordination. They can help individuals and small teams perceive more, remember more, test more, build more and coordinate faster. In that form, AI increases sovereignty. It gives more people the leverage that used to belong only to large institutions.

That is the world I am building toward with Hermes and AI4 Accountancy.

Not AI as a central brain. AI as a network of instruments. Agents that help people act on their own local knowledge. Agents that reduce friction between perception and execution. Agents that let a small firm build learning loops that used to require a corporate machine.

This is why The Last Human in the Loop matters. The human is not valuable because he clicks the final approval button in a workflow designed by machines. That is fake sovereignty. The human is valuable because he can stand outside the frame and ask whether the frame is worth optimizing at all.

AI can optimize inside a theater.

Humans decide what deserves to be a theater.

Hayek would have understood this distinction. His argument was never that people are perfectly rational. They are not. His argument was that no replacement intelligence can safely substitute for the coordination process through which many limited minds encounter reality.

The market is not wise because every participant is wise.

The market is wise, when it is wise, because it lets error collide with consequence.

A good agentic system should do the same. It should increase the surface area of contact with reality. More tests. More feedback. More local adaptation. More visible consequences. Less theatre. Less dashboard mysticism. Less central pretense.

Europe and the constructivist mind

This is also why Europe's technology problem is not only about capital or regulation.

Those matter. A lot. I have written about this in Europe Regulates the Gate While Others Build the Next Door and Europe Is Not Sovereign. It Is Comfortable.. Europe keeps trying to govern the future as if legitimacy can substitute for leverage.

But underneath that is a Hayekian failure.

Europe trusts the designed order more than the discovery process. It prefers committees, frameworks, rights language, compliance structures and risk categories. Some of that is valuable. Civilization needs guardrails. But when the guardrails become the main creative act, society stops learning from builders and starts learning from regulators.

The result is not wisdom. It is delayed contact with reality.

A civilization can be morally articulate and still epistemically weak. It can explain what should happen while losing the systems through which anyone finds out what can happen.

Hayek's lesson is brutal here: if you suppress the discovery process, you do not get a kinder future. You get a future designed by people with less information than the system they are trying to control.

humility is not passivity

The common misunderstanding of Hayek is that humility means doing nothing.

That is wrong.

Humility means you do not confuse your model with the world. It means you build systems that can learn from what you cannot know. It means you prefer feedback to fantasy. It means you do not let experts become priests merely because their abstractions are mathematically elegant.

A Hayekian builder is not passive. He builds loops.

He builds products that meet users. He builds companies that learn from markets. He builds agents that test reality instead of merely producing reports. He builds institutions that can adapt without needing a master mind. He respects local knowledge, tacit skill, embodied judgment and the strange intelligence that lives at the edges.

That is not laissez-faire as laziness.

That is reality-centered design.

The gardener image from Hayek's Nobel lecture is useful here. In complex systems, he said, we should not imagine ourselves shaping society like craftsmen imposing form on inert material. We should cultivate conditions for growth.

That is a beautiful line because it applies far beyond economics.

You do not centrally plan a child into aliveness. You create conditions for growth.

You do not centrally plan a body into health. You create loops of sleep, food, movement, recovery and attention.

You do not centrally plan a company into intelligence. You create feedback loops between customers, builders, capital and truth.

You do not centrally plan a civilization into sovereignty. You cultivate builders, markets, energy, capital, courage and beauty.

the knowledge no machine owns

Hayek's work becomes sharper in the age of AI because we are about to repeat the old mistake at a higher level of abstraction.

We will be told that enough data can replace judgment. That enough prediction can replace markets. That enough optimization can replace entrepreneurship. That enough automation can replace human agency. That enough intelligence can replace wisdom.

Some of this will work locally. That is what makes it seductive.

AI will remove enormous waste. It will expose patterns humans miss. It will help small teams do impossible things. It will make many bureaucratic processes look absurd in retrospect. Good.

But if we turn AI into the new central planner, we will have misunderstood both AI and Hayek.

The better path is not human versus machine. It is not market versus model. It is not tradition versus technology.

The better path is distributed intelligence.

Humans with agents. Markets with better perception. Firms with faster learning loops. Institutions that use AI to listen to reality instead of enforcing their own abstractions more efficiently. Sovereign individuals who can act with more leverage while remaining grounded in truth, curiosity and beauty.

Hayek matters now because he reminds us that no mind, biological or artificial, owns the whole.

The future should not be governed by a central intelligence pretending to see everything.

It should be cultivated by many intelligences, human and machine, in contact with reality.

That is the line.

Not markets as dogma.

Not AI as master.

Not bureaucracy as morality.

Sovereign coordination.

A world where intelligence is distributed, feedback is real, and no system is allowed to confuse its map for the living territory.