The Seldon Plan works until it doesn't, and the mechanism of that collapse is the only thing that matters.

Asimov shows us this precisely: the Plan functions as long as it's a secret held by a priesthood of psychohistorians, but the moment the Four Kingdoms learn the predictions exist, behavior shifts.

People know what's supposed to happen. They optimize against it, game it, misinterpret it under pressure — and the Plan's predictions were never wrong, they became irrelevant because the system's existence changed the game it was trying to predict.

The Plan's predictions were never wrong—they became irrelevant because the system's existence changed the game it was trying to predict.

The Foundation itself becomes a variable in its own equation. The Plan adapts through the Seldon Crises, through calculated reveals, through managed information — but it's always one step behind because reality has one more degree of freedom than the model. What Asimov grasped was this: foresight systems have a fatal expiration date baked in, they decay the moment they enter the world they're trying to control.

Every AI alignment researcher building constraint systems, game-theoretic models, or "value-locked" AGI architectures is constructing the next Seldon Crisis — designing prediction systems that assume the predicted entities won't know they're being predicted, or will cooperate if they do. The instant those systems become operational, they enter a reality that includes knowledge of their own existence, the model breaks, and there's no psychohistorian left to revise it.