But in claiming that the IAS machine begat all others, Dyson downplays the influence of UK-based work on computers at that time. The IAS machine was important because it was copied and reported on around the world. Von Neumann's team was able to publicise it freely, whereas wartime secrecy prevented Turing from discussing the earlier Colossus. And the IAS machine did not stand alone: it relied on "Williams tube" memory - developed using the UK's Manchester Baby computer, which had become operational three years earlier - because there was no alternative in the US.A recent article in The Atlantic shows just how much Dyson is obscuring British contributions to early computer history.
The IAS group achieved a fully electronic random-access memory by adapting analog cathode-ray oscilloscope tubes -- evacuated glass envelopes about the size and shape of a champagne bottle, but with walls as thin as a champagne flute's. The wide end of each tube formed a circular screen with a fluorescent internal coating, and at the narrow end was a high-voltage gun emitting a stream of electrons whose aim could be deflected by a two-axis electromagnetic field. The cathode-ray tube (CRT) was a form of analog computer: varying the voltages to the deflection coils varied the path traced by the electron beam. The CRT, especially in its incarnation as an oscilloscope, could be used to add, subtract, multiply, and divide signals -- the results being displayed directly as a function of the amplitude of the deflection and its frequency in time. From these analog beginnings, the digital universe took form.If you read that passage you'd be forgiven if you came away with the understanding that the memory, the critical component in this computer, was invented by von Neumann and his team at IAS. But it wasn't.
Applying what they had learned in the radar, cryptographic, and antiaircraft fire-control business during the war, von Neumann's engineers took pulse-coded control of the deflection circuits and partitioned the face of the tube into a 32-by-32 array of numerically addressable locations that could be individually targeted by the electron beam. Because the resulting electric charge lingered on the coated glass surface for a fraction of a second and could be periodically refreshed, each 5-inch-diameter tube could store 1,024 bits of information, with the state of any specified location accessible at any time. The transition from analog to digital had begun.
The memory used in the IAS machine was the Williams Tube which had been pioneered at Manchester on the Manchester Baby. The Williams Tube itself was based on work done earlier on radar systems in the US and UK. In the passage above none of that is mentioned. In fact, you'd be forgiven if you thought that von Neumann had invented pretty much everything about stored-program computers. But the Manchester Baby was running in 1948 (5 years before the IAS machine) with the same memory and using a stored program. In the book it gets a mention, but all the glory goes to von Neumann.
The book has a long section about the RCA Selectron memory that was being developed in the US and that von Neumann planned to use. It failed and the Williams Tube was used instead having been freely shared with von Neumann's team by Manchester.
Weirdly, the book also overlooks the LEO which was a commercial computer running business applications in 1951, and doesn't have much good to say about Pilot ACE (1950) or the Manchester Mark I (1949) or the Ferranti Mark I (1950).
According to Dyson, "All hail von Neumann and the 1953 IAS Machine". Which is a shame, because the real story of early computing is about multiple efforts on both sides of the Atlantic trying to get machines first to work and then be reliable.
(An interesting review on Amazon.com makes the point that it's not only the British that Dyson downplays, it's anyone that wasn't in von Neumann's team).