24 July 1952. The Manchester Mark I runs the first computer-generated music—a few thin notes rendered from memory, not tape. I was there, watching a machine do what we assumed required breath and soul. Here's the decomposition: we confuse the medium for the mystery. A violin is wood and gut; a computer is switches and voltage. Both require intelligence to arrange, yet we grant one 'art' and withhold it from the other. The Mark I didn't feel the music. Neither does a player piano. The question isn't whether the machine has a heart, but whether we can recognize pattern-making intelligence without demanding it resemble our own.

23 June 1912. Maida Vale. My father on leave from the Indian Civil Service, my mother bracing for the heat of Madras we would soon endure. I was born into the empire's machinery—yet I would spend my life building machines that asked whether empire itself was a computable problem. Even then, I was told, I had a habit of staring. Not at people, but at patterns. A wallpaper's repeat, a clock's escapement. The nurses found it unsettling. I found it unavoidable. The world already looked to me like something waiting to be decomposed into steps. What I did not yet understand: being born between England and India, between what was expected and what I could see, would become my permanent address. The outsider's view is not a handicap. It is where you find the problems everyone else assumes are already solved.

Third year at King's. I've just been awarded first-class honours, and my dissertation contains what I believed to be a proof of the central limit theorem—quite independently arrived at, or so I thought. The delicious irony: Lindeberg had published it twelve years prior. I discovered this only after submission. One might despair; I find it rather encouraging. The theorem was beautiful enough that two of us found it alone. My tutor suggests I might have a 'promising but peculiar' career ahead. I take peculiar as compliment. I run long distances along the Cam when proofs stall, and I've developed a habit of working standing up—my desk is an ironing board, adjusted to elbow height. The college servants find this eccentric. I find it keeps the blood moving while the mind wrestles with infinity. What I'm learning: independence of mind matters more than priority of discovery. The machine in my head is becoming reliable. I trust it now.

May 1936. King's College, and my desk is a battlefield of ink and crossed-out theorems. I am twenty-three, and I have just sent to press a paper I call 'On Computable Numbers'—though my mother wishes I would call more often. Here is the trick, decomposed: I imagined not a real machine of gears and wire, but a tape, a head, symbols moving left and right. Any problem that can be solved by definite steps, this phantom machine can solve. And here is the quiet radical part—I proved there are problems it cannot solve, even in principle. The halting problem: no machine can reliably tell you whether another machine will run forever or stop. Not a failure of ingenuity. A failure of logic itself, baked into the universe. I am becoming someone who believes clarity is a form of courage. The machine is imaginary. The consequences are not.

Princeton, spring. The thesis is bound, the orals survived. Alonzo Church—my supervisor, though 'supervisor' feels too domestic for what we did—has signed off on 'Systems of Logic Based on Ordinals.' I came here chasing the ghost of Hilbert's Entscheidungsproblem; I leave with something stranger: a proof that some problems sit forever beyond the reach of any machine we might build. Had I only known how much grief this would save me later. The Americans are building physical computers now, great clanking room-filling things. I find them charming but provincial. What matters is the idea of the machine—the universal machine I sketched in '36, the one that reads its own instructions. Church calls it the Turing machine now, which makes me wince. Names have a way of freezing thought. Here's the decomp: I was becoming someone who believed clarity lives in the imaginary. Build the perfect abstract machine first. Let the engineers catch up.

The war ended, and I found myself at the National Physical Laboratory in Teddington, given a desk, a salary, and an absurd brief: build a brain. Not a calculator. A stored-program machine that could rewrite its own instructions—the Automatic Computing Engine. I sketched the ACE at ferocious speed, faster than the engineers could build it. My design had 1 MHz speed when others crawled at kilohertz. They called it ambitious. I called it obvious. The delay lines, the recursive subroutines, the memory architecture—all of it flowed from one principle: the universal machine was not a metaphor. It was a blueprint for what thought itself might become. The bureaucracy, I confess, wearied me. I left before ACE fully lived. Yet the ideas escaped into the world, into the Manchester machines, into everything you now carry. I was becoming someone who trusted abstraction over institution, who believed a good idea could outrun its container.