George Dyson:


"Turing's Cathedral" (2012)

(Copyright © 2015 Piero Scaruffi | Terms of use )
This book is a good example of how important marketing is. Many people probably bought it because it has Alan Turing's name in the title: no, it is not a book about Turing. I bought it because i thought it was a detailed history of how the electronic computer was invented and developed: no, it is not. If that's what you are looking for, stick with Paul Ceruzzi's "Reckoners" (1983) that has everything that you'll ever need to know about developments in Germany, Britain and eventually in the USA. Dyson's book will mislead you into thinking that the electronic computer was essentially a US invention, and nothing could be farther from the truth. Zuse's Z3 was operational at least 5 years before the first US computer, and of course the British had built the Colossus, the "Baby" and many other experimental computers before any of Von Neumann's computers in the USA. The first commercial computer was sold by Ferranti in 1951, two years before Dyson's story begins. This bias is visible from the very first pages, where Dyson claims that in March 1953 all the computers "on planet Earth" were in the USA. Nothing could be farther from the truth. This book even overlooks the truly pioneering computers made in the USA, both the commercial ones and the academic ones. Its reader can be completely misled about the state of the art in 1953.

The book wastes a lot of pages (a lot) in detailed biographies, and not only of the computer pioneers. Some of the biographies go back centuries. We are told details of how parts of the USA were colonized. The trivia might be interesting for some readers, but for others (like me) the trivia simply constitute "filler". Remove the bios, and very few pages are left.

The book is really about a narrow period of time after World War II (and after the introduction of the first computers), and centers on John VonNeumann's work at Princeton, and it probably gives VonNeumann more credit than he deserved (he deserved a lot in Quantum Mechanics and in Economics, but perhaps not as much in designing the computers that we used today).

The overwhelming multitude of biographical details ends up making the events more obscure. The chapter about the EDVAC is confusing at best: it starts with a stored-program version of the ENIAC that in Princeton and ends with the a code-named Hippo program running on an IBM SSEC in New York, but the transition from one to the other is not explained and the dates are not clear. The story of the MANIAC is maddeningly confusing: wasn't it built and used solely by the Los Alamos laboratory (in 1952) by Nicholas Metropolis? It doesn't come out of Dyson's narrative, who says that a copy was being built at Illinois in 1949 and it was used in 1950 by Schwarzschild (even the original IAS machine was never used before 1951). Dyson claims that MANIAC was the original name of the IAS machine, but every source says that MANIAC was the name chosen by Metropolis at Los Alamos, not the original name at IAS. Even if Dyson's story is true, it only creates confusion to call MANIAC the original IAS machine without saying that MANIAC would be the name of the Los Alamos machine.

Nowhere on the cover or the back does this book say that it is simply a semi-biography of John VonNeumann at the Institute for Advanced Study. If it did, i might rate it differently. But a book titled after Turing and subtitled about the invention of the computer that says absolutely nothing about the invention of the computer and doesn't even credit adequately Mauchly and Eckert for the "VonNeumann Architecture" cannot be rated as informative.

For me this book was mostly a waste of time. For the casual reader it can be misleading.

There's an old book that tells the same story, except that it was written by a living witness: Herman Goldstine's "The Computer from Pascal to von Neumann" (1972). A broader discussion can be found in William Aspray's "John von Neumann and the Origins of Modern Computing" (1990). Nicholas Metropolis and others edited the 600-page "The History of Computing in the Twentieth Century" (1980) that contains articles by the people who built those machines. "ENIAC in Action" (2016) by Thomas Haigh, Mark Priestley and Crispin Rope is much easier to read and better organized. For Ulam's and VonNeumann's theory of self-reproducing automata see Freeman Dyson's "Disturbing the Universe" (1979).

Thank you for introducing me to the work of Nils Barricelli.