JOHN VON NEUMANN: a Calculating Man
A great mind teaches an electronic brain how to remember
Thinking was both business and pleasure for John von Neumann, and nowhere did he do it better than on the railroad trains that shuttled him between the many federal agencies and private companies he served as a blue-ribbon consultant. Staring at the rail-car ceiling, the great mathematician muttered intermittently as his mind raced along the path of logic like a steam locomotive. Before the train reached its destination, von Neumann usually reached his--or at the very least a way station in whatever intellectual journey he was on. He was an explorer of realms too abstruse to be penetrated by more than a small fraction of humanity--and yet of massive import. One stop on his itinerary was Los Alamos, N.M. (It was he who devised the implosion lens that triggered the plutonium-type atomic bomb.) Another stop was the University of Pennsylvania in Philadelphia. There and on arriving and departing trains, von Neumann figured out how to give the nascent electronic computer a storable memory--an invention as vital to the future of computing as the wheel was to chariot racing.
Born in Budapest in 1903 as Neumann Janos (the "von" came later when his father, a banker, was ennobled), he had been a classic prodigy child. He had scarcely outgrown baby food before he was gulping down science and world history. When he was not yet out of high school, mathematicians were already treating him like a colleague. In his 20s, the milestones flew by--the award of his doctorate, appointments to the Universities of Berlin and Hamburg, publication of papers noticed around the world, including one in which he mathematicized quantum mechanics. A summons from Princeton brought his migration across the Atlantic in 1930. And in 1933 the 30-year-old von Neumann became the youngest professor at the Institute for Advanced Study, a new think tank stocked with big catches, Albert Einstein among them.
In America, von Neumann developed a reputation for sharing the fruits of his genius with great generosity. Scholars (and later bureaucrats and businessmen) discovered that he could solve problems in mathematics and physics that baffled others. And he often seemed not to care who got the credit; his pleasure came from the brain work. At seminars, conferences, and sherry parties, admirers frequently tested the storied ability of this pear-shaped little man in bankers' pinstripes to make elaborate calculations in his head at incredible speed. They would pitch him a problem and watch while he went into deep concentration, hopping from foot to foot as he closed on his quarry. After a minute or two, he produced a solution that would have taken almost anyone else hours at best.
As much as he prided himself on these command performances, von Neumann took pains not to come across as arrogant or condescending. "Johnny" worked the room with an ingratiating smile and copious bonhomie. He might mumble "nebbish, nebbish" to himself when some long-winded confrere spouted bad science from the speaker's platform. Yet he treated everyone with pronounced politeness, whether Nobel laureate or graduate assistant, wise man or fool. When a disagreement became emotional, he broke the tension by telling a dirty joke (science was still a stag affair), the punch line all the funnier when delivered in an erudite Hungarian accent. He hosted lively parties at his rambling colonial on Westcott Road in Princeton. Bowing low like an Old World courtier, he somehow managed to introduce most of his guests without saying their names. His memory was prodigious but selective--he could recite page after page from works read in childhood yet had trouble remembering people's names.
The night shift. For one so often on the run (at one point he held 20 consulting contracts), von Neumann published prolifically. His major papers and journal articles totaled 150. His books included a 600-page hardcover classic co-authored with Oskar Morgenstern, Theory of Games and Economic Behavior, completed in 1943 when von Neumann was busy with war consulting. His output was enhanced not only by his mental speed but by his habit of sleeping four hours a night and thinking about his work almost every waking moment. Even in slumber his subconscious plowed on. He was notorious for waking at 3 a.m. with a burst of insight and phoning an associate immediately to report the illumination. He bought a new car each year, drove it fast, and didn't always let a spin behind the wheel interrupt his reading, so hungry was he for knowledge. His smash-ups were the talk of Princeton. He emerged from one with this bulletin: "I was proceeding down the road. The trees on the right were passing me in orderly fashion at 60 miles an hour. Suddenly one of them stepped out in my path. Boom!" On rail trips, he was untroubled by such hazards.
It was a chance meeting on a train platform in 1944 that led to von Neumann's involvement in the electronic computer. Herman Goldstine, a mathematician and the U.S. Army's liaison on the project, recognized von Neumann and introduced himself. Von Neumann agreed to go to the Moore engineering school at the University of Pennsylvania to see ENIAC (electronic numerical integrator and computer). Ten feet high and 100 feet long, stuffed with 17,000 vacuum tubes, ENIAC was bigger than a dinosaur. But it had no more memory capacity than a gnat. Unleashing it on a new task meant laborious manual reprogramming, with a flying wedge of operators throwing switches and replugging cables. ENIAC's designers, J. Presper Eckert and his partner, John Mauchly, welcomed the intrigued von Neumann as a consultant. They were completing ENIAC and designing its successor, eventually dubbed EDVAC (electronic discrete variable arithmetic computer).
Profits for prophets? The outlines of the stored-program computer took shape in the next several months as Eckert, Mauchly, and Goldstine met with von Neumann. At the core were five functional units, which von Neumann called central control, central arithmetic, input, output, and memory. The memory unit would hold both a program's instructions and the numerical data on which it operated. Eckert and Mauchly later claimed paternity of all these concepts when they sought unsuccessfully to patent EDVAC. They were, indeed, EDVAC's electronic wizards. But von Neumann was clearly its master logician; the elegant imprint of his mind was unmistakable. He described the stored-program computer in a 101-page "draft report." Mimeographed by the Moore school in June 1945 and finding its way to a small network of interested people in America and Britain, von Neumann's summary became the seminal work of a new industry.
While Eckert and Mauchly incorporated and sought to turn their computer work into riches, von Neumann fought to keep computer breakthroughs in the public domain as much as possible. He did not want science's new wonder tool held back by proprietary secrecy. Coaxing funds out of the armed services and private industry, he set up his own computer project in the basement of the Institute for Advanced Study. The computer he and his team built there--and reports they wrote and distributed widely--crystallized the "von Neumann architecture," as it came to be called. For the rest of the century, most computers would mirror it in some fashion. The IAS computer spawned 17 clones around the world, including MANIAC (Los Alamos National Laboratory, 1952), ILLIAC (University of Illinois, 1952), BESK (Stockholm, 1953), BESM (Academy of Sciences, Moscow, 1955), and SILLIAC (University of Sydney, 1956). The Rand Corp. clone, completed in 1954, was named in whimsical tribute. They called it JOHNNIAC.
With the acceleration of the cold war in the early 1950s, von Neumann was pressed into service as a military strategist. He applied game theory to the nuclear arms race, chaired a panel charged with unsnarling efforts to produce intercontinental ballistic missiles, and took a seat on the Atomic Energy Commission. In the summer of 1955, he felt a pain in his shoulder; the diagnosis was cancer. He died in 1957 at age 53. Born a Jew, von Neumann had converted to Catholicism in 1930 when he wed his first wife, a Catholic, but he had fallen away from the church seven years later. Just before his death, he went back to religious practice. Was the decision a leap of faith by a lifelong logician? "There probably is a God," he had said to his dying mother. "Many things are easier to explain if there is than if there isn't." Or was it a calculating man's swiftest calculation yet? To at least one visitor to his hospital room, von Neumann jauntily paraphrased Pascal: So long as there is any chance nonbelievers will be damned, logic dictates that you hedge your bets just before you give up the ghost.
Milestones
1951 UNIVAC I, the first commercial computer in the United States, is built for the U.S. Bureau of the Census.
1958 The North American air-defense system called SAGE introduces interactive computing, with punch cards giving way to continuous calculating. This leads to networked computers accessible by remote terminals.
1958 Integrated circuit are developed. Silicon chips take the place of many individual transistors wired together, dramatically reducing costs and increasing computer power.
1964 IBM System/360 is the first family of computers built around a common structure. The software industry is born when IBM is forced to let others write programs for the 360.
1975 MITS Altair, the first personal computer, is announced on the cover of Popular Electronics as a kit for hobbyists. 1991 The World Wide Web is developed by Tim Berners-Lee of CERN (European Council for Nuclear Research) so physicists can swap data on the Internet. With Berners-Lee's standards for addresses and data exchange, the Internet grows from a researchers' tool into a powerful new public medium.
This story appears in the August 17, 1998 print edition of U.S. News & World Report.