When we become aware of the sweat, dreams, persistence, and enthusiasm of the people who began our digital age, all the effort of learning zeros and ones and the entire universe created from them begins to make sense; it becomes more human, closer to each of us. That’s why I decided to translate this text, written by science historian George Dyson for Make magazine. I thank George Dyson and Mark Fraunfelder, the magazine’s editor-in-chief, for kindly allowing me to translate and publish the text in Portuguese on this blog. They were very attentive and kind. You can find the original English text at this link. Now, here’s the article:
Looking for the beginning of the digital universe? Check the basement, by the washroom.
Sixty years ago, in the spring of 1946, several electronic engineers approached the Institute for Advanced Study in Princeton, NJ, and asked for a place in the basement where they could use their tools. They had no luck. “The only really usable space in our shed is that washroom next to the men’s room, where you are very welcome,” said the director of the Electronic Computer Project, John von Neumann. Despite this unpromising welcome, the engineers moved in, and in April the project’s accountants recorded the first US$4 (four dollars) for “electrical work.”
The Institute for Advanced Study was founded in 1930 as a refuge for scholars in mathematics, physics, history, and art. It had no laboratory facilities of any kind. “It would be wiser than giving resources to people who have time to think about what to do with them,” the founding director, Abraham Flexner, was told in 1939. Not a word about workbenches or tools.
John von Neumann was a pure mathematician who, along with many of his colleagues, turned to applied work during World War II. He liked it there and refused to leave. Having been exposed to the power of digital computing through his involvement with the nuclear weapons program at Los Alamos, he was eager to build the next generation of computers, and was well aware of the implications of Alan Turing’s 1936 results on Universal Machines. “We’re building one," he told his newly arrived engineers, and Turing’s paper in the IAS library was consulted so often that its cover came loose.
Other groups were working on similar projects, but none was as quick or bold enough to keep up with von Neumann.
“If he really wanted a computer,” explained Arthur Burks, “what we had to do was build it.”
Burks and Herman H. Goldstine, an Army colonel transferred from the ENIAC project at the Moore School in Philadelphia, settled on the second floor of Fuld Hall, in a small office that belonged to the logic theorist Kurt Gödel. There, under von Neumann’s guidance, they began defining the machine’s logical architecture.
“The heart of the system is a central clock, which needs a huge charge,” says the minutes of the first Electronic Computing Project meeting. The circuit should be modular, the engineers added, because “this kind of design is favorable for mass production.” “Words that encode instructions are accessed in memory exactly like numbers,” von Neumann explained, breaking the distinction between numbers that represent things and numbers that do things, thereby avoiding various problems through this simplification.
Instead of bringing a few mathematical logicians into a laboratory full of engineers, von Neumann brought a group of engineers into a place run by mathematicians. Although they had to start by building their own workbenches at the Institute and scrounging for surplus wartime components as best they could, nobody was around to say, “Well, Dr. von Neumann, that looks interesting, but we build computers like this…”
The chief engineer was Julian Bigelow, a former colleague of Norbert Wiener and co-author of the 1943 paper “Behavior, Purpose and Teleology,” from which the cybernetics movement emerged. His job was to take the abstract logical design produced by Burks, Goldstine, and von Neumann and bring it to life as a machine.
Every bit of the digital universe we know today can be traced directly to the first binary digits that came to life in the summer of 1951.
“Julian would come up with the ideas, Ralph Slutz would work out the details, and then James Pomerene and I would try to make the electrons work as planned,” explains Willis Ware, one of the original engineers. “Julian was the architect of that machine.” Bigelow’s group soon outgrew its room in the Fuld Hall shed and moved to a building hurriedly constructed on the edge of the Institute’s 600 acres.
At the Institute for Advanced Study, it was easier to find an expert in quantum mechanics than someone who had worked on their own car. Bigelow was the exception—the only permanent Institute member who routinely took engines apart with his own hands. He came to the new computer’s construction as a true engineer. The new machine’s innards (the first to have a substantial, fully random-access memory available at megacycle speeds) contained 40 “Williams tubes,” 20 on each side of the main frame, so the computer not only looked like a huge V-40 engine with many cylinders and valves, but worked like one too.
Every bit of the digital universe we know today can be traced directly to the first 40,960 binary digits that came to life in the summer of 1951, as a 32x32x40 array of dots charged on the surface of delicate cathode-ray tubes.
The first major calculation, a thermonuclear problem for Los Alamos coded by a team that included Hedy Selberg and Klari von Neumann, ran for six weeks without stopping and resulted in the successful hydrogen bomb. Not bad for an improvised operation.
Cosmologists may never know what existed before the Big Bang, but at the beginning of the digital universe, we know it was Julian Bigelow, James Pomerene, Herman Goldstine, Ralph Slutz, Jack Rosenberg, Willis Ware, and a small group of assistants who worked to realize the dreams of Leibniz, Babbage, Turing, Atanassoff, Zuse, von Neumann, and many others honored for their discoveries.
Team that built the IAS computer. Photo credit: Alan Richards
Photo: Making a revolution: Institute for Advanced Study staff on the Electronic Computer Project, ca. 1952. After John von Neumann decided to build a modern high-speed electronic digital computer with stored programs, several people had to design and build the machine and write the code to make it work. Some of the people whose identities are incomplete are shown here.
The entire computer was built in the primitive machine project’s workshop, and all the code was assembled by hand, using absolute, bit-by-bit addressing. The finished computer had a total of 3,474 vacuum tubes and an internal memory of 40,960 bits, stored as arrays of 32x32 dots charged on the surfaces of the 40 “Williams tubes” (cathode-ray tubes). It went into service in the summer of 1951, and its design was widely replicated around the world.