Stored-program computer

Shortly after 11 o'clock on the morning of June 21, 1948, the Small-Scale Experimental Machine—nicknamed 'Baby'—executed the first program ever run on a stored-program computer. Built at the University of Manchester by Frederic Williams, Tom Kilburn, and Geoff Tootill, the machine ran for 52 minutes to find the highest proper divisor of 2^18, a problem chosen not for its importance but because it required enough instructions to test the new architecture. The computation itself was trivial; the architecture was revolutionary. For the first time, a machine held both its instructions and its data in the same modifiable memory, treating programs as data that could be examined, modified, and replaced without rewiring.

The theoretical foundation for stored-program computing had been articulated in 1945 in John von Neumann's 'First Draft of a Report on the EDVAC,' which described a computer architecture where programs would be stored in the same memory as data. But theory is not implementation. ENIAC, completed in 1945, was electronic and programmable but required days of rewiring to change programs—its 'program' was the physical configuration of patch cables. The EDVAC design remained on paper until 1949. What Manchester achieved was making the stored-program concept actually work.

The critical enabling technology was the Williams tube, invented by Freddie Williams earlier in 1947. This cathode-ray tube memory could store binary digits as charge patterns on the tube's phosphor screen, with each bit readable and writable in microseconds. Previous electronic memory options—delay lines, flip-flop circuits—were either slow, expensive, or impractical for the random access that stored programs required. The Williams tube provided the first truly random-access electronic memory, enabling the computer to fetch any instruction or datum at will.

Why Manchester? The answer lies in wartime radar research. Williams had spent the war developing cathode-ray tube displays for radar, becoming expert in their electronic behavior. When he moved to Manchester in 1946, he brought this knowledge and began experimenting with using CRTs for memory storage. Tom Kilburn, his colleague, realized that a working random-access memory would make a stored-program computer feasible. The Baby was explicitly designed as a testbed for the Williams tube—the computation was secondary to proving the memory worked.

The convergent emergence was striking. In America, the ENIAC team was simultaneously converting their machine to stored-program operation, demonstrating it in April 1948—two months before Manchester's Baby. The Cambridge EDSAC, designed by Maurice Wilkes, ran its first program in May 1949. All three groups were racing toward the same inevitable destination once the 1945 EDVAC report had crystallized the concept. The adjacent possible had opened; multiple teams rushed through.

The cascade from this architectural innovation is effectively the entire digital world. Every computer built since—from mainframes to smartphones, from gaming consoles to cloud servers—follows the fundamental principle that Baby demonstrated: programs are data, stored in memory, fetched and executed instruction by instruction. The Ferranti Mark 1, commercialized from the Manchester Mark 1 (Baby's successor), became the world's first commercially available general-purpose computer in 1951. By treating software as mutable data rather than fixed wiring, the stored-program architecture created the possibility of operating systems, compilers, applications, and every layer of abstraction that defines modern computing.

The Manchester Baby was small and primitive—just 32 words of memory, capable of only subtraction and conditional branching. But it proved that the stored-program concept worked in silicon and phosphor, not just on paper. Every computation performed by every computer since traces its architectural lineage to that 52-minute program run in a Manchester laboratory.