The Promise and the Crash

The early results were real. Theorems proved, puzzles solved, problems falling in a matter of years that had stood for centuries. Reasonable people looked at the rate of progress, drew the line forward, and concluded that a thinking machine was perhaps a decade away.

They were wrong by more than half a century, and the interesting question is why the reasoning failed, because the reasoning looked sound.

What nobody had noticed was how much the early problems had in common. Each one came with its world already fenced off: every legal move written down, every object accounted for, nothing outside the description that could matter. Progress inside a fence tells you very little about the ground beyond it, and it turns out there are two entirely separate reasons why, neither of which is a matter of trying harder or waiting for better machines.

When the promised results did not arrive, the funding went, and the field spent years in the cold. It is easy to read that as a story about hype, and it is, but the more useful reading is that a genuine achievement and a badly unjustified forecast sat side by side and looked identical from outside. They usually do.