What the proof didn't say

So what did the proof actually cover? Look closely, because the gap matters.

Minsky and Papert proved that this design had a ceiling. One layer of dials, one pass, no more. That was the thing they showed could not be fixed.

What they did not prove was anything about a different design. A machine with more layers, built another way, might sail right past the limit. Nobody had tried yet. That door was wide open.

There's a name for assuming it was shut: hasty generalization. Jumping from "this version failed" to "none of them ever will." It happens in science, in business, in medicine. It happened here.

A small group of researchers read the proof carefully and noticed the gap. The book killed one design. It didn't kill the idea. They kept working with little funding through the 1970s and into the 1980s, on designs nobody had tried yet.

But before we get there: what exactly was the limit? What could the Perceptron actually do, and why couldn't it do more?