Untitled

The twentieth century produced an extraordinary set of ideas for understanding the world: modern economics, statistics, AI, and more They transformed civilization. They also grew into separate disciplines, each becoming more sophisticated while sharing assumptions so deep that they are rarely questioned.

We think those assumptions break down in an important class of systems. There are remarkably simple processes that can generate behavior of extraordinary complexity—without randomness, without human irrationality, and without anything going “wrong.” That suggests that phenomena we currently treat as separate problems in economics, statistics, biology, and artificial intelligence may actually be different manifestations of the same missing science.

If we're right, the implications are enormous. It could mean rethinking some of the foundations of how we forecast economies, design markets, evaluate medical evidence, make social policy, and align intelligent machines.

There has always been a cost to getting these questions wrong. But AI changes the stakes. We are beginning to embed our existing theories of prediction, optimization, welfare, and decision-making directly into machines—and then putting those machines inside financial markets, supply chains, corporations, healthcare systems, and governments.

A mistake in a paper is a mistake. A mistake embedded in infrastructure can become a crash. A mistaken theory of welfare embedded in a sufficiently powerful system can become coercion. And a mistaken model of how complex systems behave, deployed everywhere at machine speed, can become something we cannot easily undo.

So the opportunity is not merely to prevent disaster.

It is to build the missing science before the old one becomes automated infrastructure.

That requires a different kind of institution: one organized around the problem rather than the disciplines. Economists working with mathematicians. Computer scientists with biologists. Statisticians with people willing to question the foundations of statistics itself. Researchers with enough intellectual freedom to follow a problem even when doing so takes them outside the boundaries of an established field.

Technology should expand human freedom, agency, and flourishing—not administer them.