The wall: search explodes

The programs worked beautifully on small problems. They failed on large ones.

Early AI often worked by trying possibilities, following the promising ones, and backing up when it got stuck. The Logic Theorist did this. The General Problem Solver did this. It worked when the search space was small enough to manage.

But the number of possibilities does not grow slowly. It explodes. In chess, the number of possible game sequences is so large that search quickly becomes impractical. Even if a machine explores millions of options per second, it can run out of time long before it runs out of possibilities.

That was the real barrier: not delay, but a wall.