The idea wasn't new in 1986. It had simply been waiting for someone who refused to let it stay buried.
Two teenagers walked the same hallways at the Bronx High School of Science in the early 1940s, one grade apart. Neither had any reason to think much of the other — that's what teenagers in adjacent grades usually are to each other, background noise. But one of them would grow up to build a machine that could learn from its mistakes. The other would grow up to write the book that nearly buried it. And for the rest of their lives, they stayed friends anyway.
Their names were Frank Rosenblatt and Marvin Minsky, and this is the part of Episode 9 the video didn't have room for.
Two Bronx Science Kids, Twenty Years Later
By the late 1950s, Rosenblatt was a research psychologist at Cornell Aeronautical Laboratory, building the Perceptron — a machine that adjusted its own wiring based on whether its guesses were right. The press loved him. One 1958 write-up predicted his machine would eventually walk, talk, and reproduce itself. Minsky, meanwhile, had gone all in on the opposite bet: that intelligence was rules, written down and reasoned through, not patterns absorbed from data. For over a decade, the two men circled the same question from opposite corners of the same small field, and by most accounts of people who knew them both, they argued loudly at conferences and remained genuinely fond of each other the whole time.
Then came 1969. When Minsky and Seymour Papert published Perceptrons — the book the video's cold open treats as a verdict — they dedicated it to Rosenblatt. Not to a rival. To someone Minsky had known since he was a teenager. It reads less like an ambush and more like the closing argument in a very long-running conversation between friends who happened to disagree about everything that mattered to both of them.
Rosenblatt never got to make his reply. On July 11, 1971 — his 43rd birthday — he died in a boating accident on Chesapeake Bay. That was fifteen years before Rumelhart, Hinton, and Ronald Williams proved that the ceiling Minsky had identified only applied to the simplest version of the idea. Rosenblatt spent the last two years of his life defending an approach the field had already declared dead, and he never lived to watch it get dug back up.
What This Really Means
There's a second thread worth pulling, because it says something about how these arguments never really end — they just change families. Geoffrey Hinton, the man whose name the video holds until the moment neural networks come back to life, is the great-great-grandson of George Boole — the same Boole from Episode 2, whose 1854 laws of thought gave every rule-based system its logical backbone. A hundred and thirty years after Boole formalized reasoning as rules, his own descendant helped prove that intelligence might not need rules at all. That's not coincidence worth dwelling on for its own sake. It's a reminder that the "rules versus weights" argument the video stages so cleanly has been running, in one form or another, since before computers existed — and the people who resolve it are rarely strangers to the people they're arguing against.
Plain Machine keeps returning to this pattern because it's the real engine of the field's history: not machines competing with machines, but people who kept working on an idea everyone else had agreed was finished.
One Question the Post Can't Answer
None of this — the friendship, the funeral, the family tree — tells you what a neural network actually is, or why a single failing line on a four-dot grid was enough to freeze a decade of funding. That part needs the whiteboard, the diagrams, and fifteen minutes you don't have to rush. Episode 9 is where the video earns the reveal this post has been circling.
Some ideas don't die when the field says they're dead. They just wait for someone who was in the room the whole time.
