Building the First Brains

In 1936, Alan Turing's machine existed only on paper — a thought experiment to answer a question about the limits of logic. It was not designed to be built. The mathematics worked regardless.

But between 1939 and 1945, the Second World War created problems that no human mind could solve fast enough. Encrypted messages needed cracking — not tomorrow, but now. Artillery firing tables required millions of calculations. Theory had a deadline.

Across three countries, engineers and mathematicians raced to build machines that could think faster than people. They used different technologies, different designs, different philosophies. But they all converged on the same basic principle Turing had described: a machine that manipulates symbols according to rules.

The first person to build a programmable computer wasn't British or American. He was a German engineer, working alone, in his parents' apartment. And almost nobody noticed.

Konrad Zuse was born on June 22, 1910, in Berlin. He studied civil engineering, not mathematics. He had no idea Turing's paper existed. He just hated doing calculations by hand.

Starting in 1936 — the same year Turing wrote his paper — Zuse began building computing machines in his parents' apartment. His first attempt, the Z1, was purely mechanical and jammed constantly. But the concept was sound.

In 1941, Zuse completed the Z3. It was the world's first working, fully automatic, programmable digital computer. It read its programs from punched film — strips of discarded 35mm movie stock.

While Zuse tinkered in Berlin, a very different kind of computing was happening in the English countryside — not to solve engineering problems, but to win a war.

By 1943, the Allied code-breakers at Bletchley Park faced an impossible bottleneck. The German military's Lorenz cipher was fiendishly complex. Even after the mathematical method for breaking it was worked out, applying it by hand was agonizingly slow.

Tommy Flowers, a 38-year-old Post Office engineer, proposed a radical solution: build an electronic machine using vacuum tubes — glass bulbs that could switch electrical signals thousands of times per second. His bosses were skeptical. Flowers persisted, partly funding the work out of his own pocket.

In December 1943, Colossus Mark 1 became operational. It used approximately 1,500 vacuum tubes and could process 5,000 characters per second — reading encrypted messages from a paper tape racing through at 30 miles per hour.

After the war, Colossus was classified. Tommy Flowers was sworn to secrecy and could not claim credit for decades. He received a modest award of £1,000 — not even enough to cover what he had spent from his own savings.

Colossus was special-purpose — it could do one thing. Across the Atlantic, American engineers were building something more ambitious: a machine that could be reprogrammed for any calculation. It would weigh 30 tons.

J. Presper Eckert (just 24 when the project began) and John Mauchly (36) built ENIAC at the University of Pennsylvania. Completed in late 1945, it was the first large-scale, general-purpose electronic computer.

ENIAC could perform 5,000 additions per second. A trajectory calculation that took a human 20 hours could be done in 30 seconds.

But there was a catch. To "program" ENIAC, you physically rewired the machine. Programming meant rearranging cables and setting switches — a process that could take days or weeks. The computation itself took seconds. Setting it up took forever.

ENIAC was built by Eckert and Mauchly. But who PROGRAMMED it? Not the men. The job of making ENIAC actually work was given to six women — and then history forgot their names.

In the 1940s, the word "computer" referred to a person — usually a woman — who performed mathematical calculations by hand. When ENIAC was built, the engineering leadership considered hardware the "real" work. Programming was seen as clerical. So the job was given to six women:

These six women received no manual — because none existed. They studied ENIAC's wiring blueprints, understood its architecture from the ground up, and invented programming techniques as they went.

When ENIAC was publicly demonstrated on February 14, 1946, the women were not introduced. For decades, they were uncredited or identified only as "refrigerator ladies" — assumed to be models posing next to the machine. It was not until the 1980s and 1990s that their contributions were recognized.

John von Neumann was possibly the most brilliant mathematician alive by the mid-1940s. Born in Budapest in 1903, he made foundational contributions to quantum mechanics, game theory, and set theory — all before age 40.

In 1944, von Neumann visited the ENIAC project. He immediately grasped both its power and its critical limitation: the machine was fast, but changing what it did was agonizingly slow.

Von Neumann had read Turing's 1936 paper. He understood the Universal Turing Machine. The insight was radical in its simplicity: store the program in the same memory as the data.

A critical note: the ideas were developed collaboratively with Eckert, Mauchly, and others. The stored-program concept is often called the "von Neumann architecture," but it built on work by Turing, Eckert, Mauchly, and others.

The first machine to actually run a stored program was the Manchester Baby (SSEM), built by Frederic Williams and Tom Kilburn, on June 21, 1948. In Cambridge, Maurice Wilkes built EDSAC, which became the first practical stored-program computer designed for regular use in May 1949.

Store the program in memory. Brilliant. But HOW do you organize the machine to make this work? Von Neumann drew a diagram — five boxes and some arrows — that still describes every computer you have ever used.

The elegance of this design is that it separates what the machine does (defined by the program in memory) from how the machine is built (the hardware). Build the hardware once. Change the software forever.

On December 23, 1947, at Bell Labs in New Jersey, John Bardeen, Walter Brattain, and William Shockley demonstrated the first working transistor. A small electrical signal at one terminal could control a much larger current flowing between the other two. ON/OFF. 1/0. A switch — with no vacuum, no filament, no glass bulb, and almost no heat.

They received the Nobel Prize in Physics in 1956. By the late 1950s, transistors had begun replacing vacuum tubes in computers.

In 1958, Jack Kilby at Texas Instruments built the first integrated circuit (IC): all components fabricated from a single piece of semiconductor material. Independently, Robert Noyce at Fairchild Semiconductor had the same idea — with a better manufacturing process. Both are credited as co-inventors.

By the late 1950s, talking to a computer meant speaking its language — binary. Every instruction, every piece of data had to be expressed as patterns of 1s and 0s:

The machines were ready. The human interface was not. What computing needed was not faster hardware — it needed a way for humans and machines to communicate. A bridge between human thought and machine code.