The most important thing anyone has said about artificial intelligence was written in 1960, in a paper that never once mentions it.
It runs two pages. Its title is deliberately unglamorous — Microelectronics and the Art of Similitude — and you will not find it on any reading list. It survives mostly as a citation in the digest of a 1960 circuits conference — pages 76 and 77, delivered on a Friday morning that February — written by an engineer most people know, if they know him at all, as “the guy who invented the mouse.”
That is the least interesting thing about Douglas Engelbart.
The wrong question
In 1960, the entire electronics industry was obsessed with one goal: make everything smaller. Transistors were replacing vacuum tubes. Integrated circuits were being born. Miniaturization was the race — and almost everyone treated it as a manufacturing problem, a matter of better tools and steadier hands.
Engelbart wasn’t interested in that question. He was interested in a stranger, deeper one:
What happens when shrinking something changes the rules it runs by?
That sounds like a small distinction. It is the difference between optimization and discovery — and most people never notice which one they’re doing.
Similitude
Similitude — a word that deserves to come back — is the science of scale: an established branch of engineering Engelbart didn’t invent, but knew how to point somewhere new. It asks: when you change every dimension of a thing, what else changes with it?
The answer is: almost everything.
A bridge built twice as large doesn’t just weigh twice as much. Its stresses shift. It vibrates differently. Air moves across it in new ways. Scale it far enough and the assumptions the original design rested on simply stop being true.
Circuits are no different. Shrink them and resistance changes, heat misbehaves, and noise that was once a rounding error becomes the whole story. Past a certain point you are no longer building the same machine, smaller. You are building a different machine that happens to look similar — governed by different physics, demanding a different design.
Engelbart wrote this down in 1960, in the flattest language imaginable. Shrinking a circuit, he warned, would force most devices to be modified “in ways which will baffle the intuition” built on normal-sized ones. He was describing transistors and resistors. But the shape of the claim is bigger than circuits — and the leap from the one to the other is the whole point of this essay.
Smaller horses
Here is the idea, stripped down: every technological revolution eventually reaches a point where incremental thinking stops working.
Smaller horses never became automobiles. Faster typewriters never became word processors. Sharper film never became digital photography. In each case, the winners weren’t the ones who optimized the old thing. They were the ones who noticed the old assumptions had quietly collapsed — and built for the new ones instead.
Once you see this pattern, you can’t unsee it. And you start to notice how much of what we call “innovation” is really just polishing a corpse.
You already know half of it
You’ve heard a version of this paper’s idea. It became the most quoted prediction in the history of technology.
In 1965, an engineer named Gordon Moore — then head of research at Fairchild Semiconductor, later a co-founder of Intel — wrote a short article predicting that the number of components on a chip would keep doubling on a steady schedule. The industry named it Moore’s Law and set its clock by it for the next sixty years.
Five years before he wrote it, Moore had been sitting in the audience at the conference where Engelbart gave the similitude talk. In 2005, for the law’s fortieth anniversary, the New York Times — in a piece by John Markoff — ran the connection under a blunt headline: It’s Moore’s Law, But Another Had the Idea First.
Here’s the part almost no one notices. Moore’s Law is the comfortable half of what Engelbart was saying — the half that promises the numbers improve on schedule, so keep building what you’re building. The industry loved that half. It poured sixty years of roadmaps into it.
The other half — that each new scale breaks the old rules and opens possibilities the old design can never reach — stayed in the room. Moore’s Law tells you the old thing keeps getting better. Engelbart’s actual point was that, eventually, it stops being the same thing at all.
We kept the reassuring prediction. We forgot the dangerous one.
The demo was the same idea
Two years after the similitude paper, Engelbart published Augmenting Human Intellect: A Conceptual Framework. Six years after that, in December 1968, he walked onstage in San Francisco and, in ninety minutes, showed a room of engineers the mouse, on-screen windows, hypertext links, and live collaborative editing — decades before any of it reached an ordinary desk. He wasn’t a solo act, and it matters to say so: the mouse was co-invented with his colleague Bill English, who also engineered the demo itself from the lab in Menlo Park, and a whole team at Engelbart’s Augmentation Research Center built the system he was standing in front of.
The mouse, drawn in the Chromix style — after the first prototype Douglas Engelbart conceived and Bill English built at Stanford Research Institute in 1968. Its official name, deadpan and perfect: an “X-Y position indicator.”
It’s remembered — in a phrase the journalist Steven Levy coined years later — as “the Mother of All Demos.” What he showed still describes the screen you are reading this on.
But the machines were never the point. The demo was the 1960 insight, grown up. Engelbart had understood what his peers hadn’t: when technology changes scale, humans have to change how they think — and the job of the computer was not to automate our calculations but to amplify our minds. He wasn’t building a faster adding machine. He was building a new kind of thinking.
Intelligence is the new transistor
We are living through another scaling revolution right now. Only this time the resource becoming suddenly, absurdly abundant isn’t transistors.
It’s intelligence.
And the industry is making the exact mistake Engelbart warned about. Walk into almost any company and the question on the whiteboard is:
How do we add AI to our product?
That is the 1960 question. That is “how do we make the transistor smaller.” It is the question of people who think they’re living through a change in size when they are living through a change in kind.
The right question — the Engelbart question — is the opposite:
What becomes possible now that intelligence is cheap?
AI is not a feature you bolt onto yesterday’s software. It is a change in scale, and changes in scale demand new architectures, new workflows, and new mental models. Most of what ships today with “AI” in the name is a smaller horse: the old product, lightly optimized, with a chatbot stapled to its side.
Redesign from first principles
The people who actually move things forward have always understood this. The Wright brothers didn’t build a lighter wagon. Jobs didn’t build a smaller minicomputer. Engelbart didn’t build a quicker calculator. Each looked at a technology that had changed scale and had the nerve to ask what it made newly possible — then built for that, from nothing.
The rest of us mistake optimization for innovation. We improve what exists. We speed up workflows. We automate yesterday. It feels like progress, and sometimes it is. But every so often the ground shifts so completely that the old questions stop mattering — and clinging to them is the surest way to be left behind.
That is the real lesson buried in Microelectronics and the Art of Similitude. Not that electronics get smaller. That every revolution eventually reaches the moment when the old assumptions collapse, and the future goes to whoever is willing to start over.
The loop he was really drawing
Late in life, Engelbart kept returning to a single diagram: a box with an arrow that leaves it, curves around, and feeds back into itself. Inside the box, three words — improving collective IQ.
He called it bootstrapping, and it’s the key to everything else he did.
The idea is deceptively simple: the better we get at getting better, the faster we get better. Most tools make a task easier. A bootstrapping tool makes you better at improving your tools — which makes you better at improving the next thing, and the next, compounding. The output isn’t a product. It’s a smarter version of the people using it, looping back on itself and accelerating.
Diagram recreated by Chromix, after Douglas C. Engelbart’s “Bootstrapping” slide. The idea and the words are his: the better we get at getting better, the faster we get better — a feedback loop whose product is a rising collective IQ. And, as he added on the slide, “just think of the important role for technologists.”
Read that box again with AI in the room.
We have just built the most powerful bootstrapping tool in history — something that could, in principle, help us get better at getting better faster than anything before it. So Engelbart’s question arrives with full force: are we pointing it at the loop, or at the convenience? Are we using abundant intelligence to make ourselves collectively wiser — or just to make the feed harder to put down?
He even named whose job this was. Beneath the loop, on the same slide: and just think of the important role for technologists. He was talking, across sixty years, to the people building right now.
To you.
Epilogue
The paper runs two pages. On its surface it is a careful piece of engineering — dimensional analysis, Pi terms, the arithmetic of shrinking a network. But buried in its final paragraph is the whole argument: the shift in physics at each new scale doesn’t only retire the old devices, Engelbart notes — it opens fresh possibilities at every new scale, for anyone willing to stop scaling the old thing and build for the new one. Its influence runs through nearly everything that followed in modern computing, and almost no one has read it.
In 1960, Douglas Engelbart gave a talk about microelectronics.
He was describing the future of human intelligence. We’re still not listening.
Source: Douglas C. Engelbart, “Microelectronics and the Art of Similitude,” 1960 International Solid-State Circuits Conference, Digest of Technical Papers, Session VII (chairman J. R. Nall, Fairchild Semiconductor), pp. 76–77. Presented Friday, February 12, 1960. Engelbart’s paper in turn draws on the theory of similitude as set out in Glenn Murphy, Similitude in Engineering (The Ronald Press, 1950) — credit where it was due, all the way down.