In 1914, when eleven-year-old John Atanasoff was attending a one-room schoolhouse in Florida, helping his father rewire the family house, learning to maintain, repair, and then drive the new Model T, as well as frustrating his teachers by surpassing them, Neumann Janusz (called “Jancsi”) was delighting his teachers, who were some of the best mathematical minds in Europe. Nobel Prize winner Eugene Wigner recalls, in Kati Marton’s The Great Escape, that “he was one grade below me, but in mathematics, two classes ahead. He already had an astonishing grasp of advanced mathematics … The way he described set theory and number theory was enchanting. The beauty of the subject, his intensity and facility of description made me feel we were close friends.” One well-regarded teacher tutored von Neumann without compensation, according to Wigner, for the sheer pleasure of “the brush with a special kind of mind.” There were other tutors, too. According to Macrae, “Before he finished high school [he] had been accepted by most of the university mathematicians as a colleague.” Jancsi was not a pest. He naturally and willingly fit in with his fellow students (Wigner recalled, “He joined in class pranks just enough to avoid unpopularity”) and pleased his teachers. He was so adept at mathematics that he could do difficult problems in several ways and gear his solution to the educational level of his associate if he had to. Perhaps we may say that whereas Atanasoff was a natural fixer and improver, von Neumann was a natural game player, always aware that the moves in any game could be made in more than one way and that each possible move would lead to a different outcome, which would in turn lead to other, different outcomes. And game playing, too, as demonstrated by Turing’s fascination with chess, was an aspect of computer innovation.
In 1920, when Neumann Janusz was seventeen, educational circumstances changed for Jews in Hungary. In a place where the vast majority of educated professionals (50 to 80 percent) were Jews, the post–World War I government instituted anti-Semitic quotas for university places—no more than 5 percent. By June 1921, when Atanasoff had saved enough money teaching and working so that he could attend the University of Florida, von Neumann was taking his exams (and worrying so much that as a result his papers were not perfect). In Gainesville, Atanasoff wanted to be a physicist, but the university offered electrical engineering, so he studied that. In Budapest, von Neumann wanted to be a mathematician, but conditions in Hungary made that impractical, so his father pushed him toward chemical engineering. Ironically, when, in September, Atanasoff left Brewster for Gainesville, von Neumann left Budapest for Berlin. But in this, too, he fell into the center of the world, or at least of the mathematical world. Marton writes, “From all over the globe, theoretical physicists gathered in Berlin, and in the medieval university town of Göttingen, three hours away. In those last years before the darkness fell on Germany, a revolution was taking place in the way we understand space and time.” This revolution was quantum mechanics, the very subject that Atanasoff was taking from John Hasbrouck Van Vleck at the University of Wisconsin at about the same time, and proving that he could comprehend in spite of a late start and missed classes.
By the time von Neumann encountered Atanasoff, he had exceptional connections, not only because he was a genius, and not only because he had been born and educated at the center of things, but also because he was worldly, charming, and personable—a connector as well as a maven, in Malcolm Gladwell’s terms. After completing his degrees at Berlin and Zurich (where a paper he wrote was sent to David Hilbert, the man who posed the problem that Turing addressed in “On Computable Numbers,” and so impressed him that he assiduously cultivated the young man), von Neumann went to the University of Göttingen in 1926, just about the same time that Atanasoff was first at Iowa State (and Flowers first went to work at Dollis Hill). In 1930, von Neumann was invited to Princeton, and two years later he was given a professorship at the Institute for Advanced Study, along with Albert Einstein and Kurt Gödel. It was there that he met Alan Turing, to whom he offered the job as research assistant in 1938. Clearly, von Neumann’s personality and biography meshed to produce a man who was perhaps preternaturally political in a way that was unusual in a mathematician or an inventor—he was not only completely at ease in all sorts of social situations, he was extraordinarily aware of the ramifications of larger sorts of politics. He was, after all, the man who was assigned to do the calculations at Los Alamos that were to estimate exactly how much damage an atomic bomb might be made to inflict upon the Japanese. His specific task was to calculate at what elevation the detonation should take place in order to achieve the greatest possible destruction. Other Manhattan Project physicists, notably Leo Szilard, von Neumann’s slightly older compatriot, preferred an intimidating demonstration of the weapon, but von Neumann was willing to make a list of good targets—according to Norman Macrae, he was instrumental in steering the air force away from the Imperial Palace, but, according to Kati Marton, he thought the Japanese holy city of Kyoto was a good target (of course, the final targets were Hiroshima, a shipping center and supply depot, and Nagasaki, a ship-building center).
Physicist Stanley Frankel, who performed many of the Manhattan Project calculations that predicted whether or not an atom bomb could be made to explode, and what would happen then, later said that von Neumann was aware of “On Computable Numbers” by 1942 or 1943 and made sure that Frankel studied it (Frankel went on to be a computer consultant after the war). With his experience on the Manhattan Project, von Neumann was one of the most influential scientists in the world.
But of course, although everyone knew that von Neumann was a genius, and an important man, in the summer of 1944 the Manhattan Project was highly classified, and in 1944, although one type of bomb had been developed (Little Boy), the method for detonating a more powerful bomb had not been worked out. Just about this time, von Neumann was approached by a young man on a train platform. The young man was Herman Goldstine. Goldstine went up to the famous mathematician (whose lectures he had once attended) and introduced himself, but von Neumann got friendly only when Goldstine began to chat about his (highly classified) work on a computer. A month later, in August, von Neumann visited ENIAC in Philadelphia for the first time. Von Neumann may have been a famous genius, but according to Norman Macrae, Pres Eckert, then twenty-five, viewed von Neumann’s visit as a test—for von Neumann. Eckert said to Goldstine that he would find out if von Neumann was really the genius he was supposed to be “by his first question. If this was about the logical structure of the machine, he would believe in von Neumann. Otherwise, not.” Forty-one-year-old von Neumann passed the test.
By the time of von Neumann’s visit, work on ENIAC had been moving at a fever pitch for fifteen months, but the speed of construction demanded by the army because of the difficulty of creating the firing tables meant that real innovation in every aspect of the machine (Mauchly’s and especially Eckert’s goal) had not been possible. They had to use parts that were already in existence (and because the machine was a low priority to the military, a percentage of these parts were defective, though not actual discards, like Zuse’s parts) and at least some ideas that derived from machines that were already familiar to the army, including Irven Travis’s machine at the Moore School that Mauchly was already familiar with by the time he met Atanasoff. Von Neumann grasped that the really new machine would be the next version, and Eckert grasped that, too—he had already begun making drawings for it.
After meeting Goldstine, Eckert, and Mauchly, and chatting with Atanasoff at the NOL (and, no doubt, with anyone else who seemed to know about computer theory), von Neumann went back and forth to Los Alamos, where he worked on the Manhattan Project—it wasn’t until December of that year that the detonation device for one of the bombs (Fat Man) was successfully tested. Work continued on the bomb, but in June 1945, von Neumann was not so busy at Los Alamos that he did not have time for other things—under his direction, Herman Goldstine wrote a description of an idea for the second version of ENIAC. The paper was 101 pages long and was entitled “First Dra
ft of a Report of the EDVAC, by John von Neumann.” EDVAC stood for “Electronic Discrete Variable Automatic Computer.” Mauchly and Eckert were told that the paper was “an internal summary of their work,” and Goldstine also told another concerned party that it was meant for internal use only; therefore it did not constitute classified material and could be reproduced. The fact that von Neumann was given sole authorship at first seemed to Mauchly and Eckert insignificant. The purpose of the paper, and its achievement, was that it expressed the logical and overarching theory of what the creators of ENIAC were trying to do, something that Eckert had hardly had time to attempt, and Mauchly had not been inclined to do, even though he had the time. Eckert had written a three-page memo in February 1944, describing a system for storing electrical impulses. A notable feature of Goldstine’s paper was that even though Eckert had described what he was building to von Neumann in August 1944 and subsequently, there was no mention of Eckert and only one mention of Mauchly (though Howard Aiken was mentioned several times). Partisans of von Neumann make the case that, as with everything else von Neumann did, he took the raw material of another man’s ideas and immediately transcended it, or, as Macrae says, “Johnny grabbed other people’s ideas, then by his clarity leapt five blocks ahead of them, and helped put them into practical effect.”
The most important contribution of the “First Draft” to computer design was that it laid out what came to be known as “von Neumann architecture”—that is, that the computer could contain a set of instructions in its memory like the set of instructions that Turing’s human “computer” would have been given and would have to follow day after day forever. The instructions would be stored in the memory, which the electronic computer could readily access (not like a paper tape or a deck of punch cards). This set of instructions in the memory would be called a stored program. Von Neumann described these ideas in terms of physical structures that had access to one another—the control unit was a self-contained space that could communicate back and forth with the memory. Separate from the control unit was the logic unit (conceived as a place where mathematical calculations were performed), which also communicated back and forth with the memory. The control unit and the logic unit communicated back and forth with each other. The problem to be solved, the input, was fed into the logic unit, and the solution, the output, emerged from the logic unit. But really these “places” were not physical structures—they were sets of instructions, an idea that von Neumann may have (or seems to have) gotten from “On Computable Numbers.” According to Macrae, “The primary memory would be fairly small, with rapid random access. Behind it would be a secondary memory. It should be able to transfer information into the primary memory automatically, as needed. The computer should be able to move back and forth through the secondary memory. Individuals should be able to enter information directly into the secondary memory.”
Although ENIAC was an army project and the war was still on when Goldstine wrote the paper, over the next few months Goldstine sent von Neumann’s report to twenty-four of von Neumann’s colleagues and friends in the United States and England. Their response was enthusiastic and included requests for more copies. Goldstine eventually sent out hundreds. It was this that finally alarmed Mauchly and Eckert, who wrote their own paper in September, describing their ideas for EDVAC and more carefully ascribing particular ideas to particular participants in the ENIAC project, but they hadn’t the gift—their report was neither as detailed nor as eloquent as Goldstine and von Neumann’s in conceptualizing the larger implications of the project. Nor did they have the connections or the reputation. Most important, they did not have the cooperation of the boss, Herman Goldstine. Goldstine, who was in charge of security classification for the project, marked Mauchly and Eckert’s report confidential, thereby ensuring that, unlike von Neumann’s report, it would not be widely read or, perhaps, read at all. There is no evidence that, even though von Neumann was in contact with Atanasoff because of the navy project, he gave Atanasoff a copy of the report or told him about it. Nor did Mauchly and Eckert send Atanasoff a copy of their report, even though his security clearance was higher than theirs.
Although Atanasoff was invited to the February 1946 unveiling of ENIAC at the University of Pennsylvania, and attended, the demonstration of the machine did not clear up any mysteries for him about how the machine worked or the principles behind it. And Mauchly and Eckert were not present. The purpose of ENIAC was to accomplish what Mauchly had originally proposed—the calculation of artillery trajectories. It was so enormous and so expensive that Atanasoff was intimidated. Even so, not long after he saw the ENIAC, Atanasoff called Richard Trexler, the patent attorney in Chicago. Trexler told him that Iowa State had never paid to file the patent application, and so he had not filed it. Atanasoff knew that his moment to patent his ideas was lost—ENIAC convinced him that computers had progressed. Either his ideas were obsolete or they were irrelevant. Computer technology, it was readily apparent, was now established and developing apace.
In Germany, in 1943 and 1944, Konrad Zuse was still hard at it, still undaunted in attempting the impossible. Even the small prototype using vacuum tubes that Herbert Schreyer wanted to build seemed to be impossible—the type of tubes they needed were not being manufactured in Germany. But while a friend at the Telefunken company made ten tubes in his spare time and smuggled them out of the lab, they discovered that they had another sort of access to materials:
Dr. Schreyer was able to get [the German Aeronautics Institute] assigned the task of examining the intended uses of mysterious devices found in shot-down American and British aircraft … After such an examination, a huge number of completely modern components, resistors, small cylindrical capacitors, variable capacitors, the most modern miniature tubes and small batteries, etc. were left over. Never again did we lack parts which we needed ad hoc for developing the computing machine; we had so much left over, we were able to set up a flourishing radio repair shop.
The conditions surrounding the invention of the Z4 were astonishing—every morning, the inventors had to clean up damage and debris from bombings of the night before. One morning, Schreyer decided he needed, as a conductor, a piece of copper-rich bronze. His two assistants decided to find some—and they did so by wandering the bombed-out streets of Berlin looking for a piece of dead streetcar cable. They managed to cut off and steal a fifty-centimeter piece without getting shot for looting. Since the computer was still not considered a government priority, Schreyer had to get a contract for the development of a dud-bomb-detecting instrument in order to have access to other materials. Once he attained first-class status through that, though, his personnel could order almost anything, and one thing they ordered was “a bottle of radioactive material” for painting on the inner surfaces of the diodes they were making. They also painted the faces of their old watches. The watches were soon stolen by invading Russian troops.
One by one, Zuse’s inventions, wherever they were around Berlin—the Z1, Z2, and Z3—were destroyed in the bombing, but work on Z4 continued; it was being built in a basement. And the use of unorthodox personnel continued—Zuse’s first programmer was blind. Watching him work led Zuse to realize that Braille was a type of computer alphabet. Subsequently, he happily employed blind or sight-impaired programmers.
While he was working on the Z4 and trying out designs for the prototype electronic computer mentioned above, Zuse understood that there was a price to pay. He writes, “Our prototype did not have the slightest practical value.” He could not quite solve the old Turing problem—how to mediate between the desirable simplicity of operation and the huge (or even infinite) number of operations required to solve a problem. But throughout the war, Zuse and his workers and programmers pursued their objectives. Reminiscing after fifty years, he writes:
Today when I look back to these days, it seems unbelievable, even to me, that we kept working while the bombs continued to fall on Berlin. We spent a great deal of the night in an air-raid shelter.
All around us, bombs fell and houses burned. More than once after a heavy attack, we thought it was finally over, that nothing would work anymore. We had no water, no electricity, and no telephones, and there was hardly any serviceable means of transportation. But each time, after a few hours, almost everything was working again. And somewhere, all of the employees found ways to pull through.
After Germany surrendered, Zuse heard that Albert Speer had suggested to Hitler that the development of the computer might aid in the war effort. “Hitler is said to have replied that he didn’t need any computing machine, he had the courage of his soldiers.”
But toward the end of 1944, after D-Day, when conditions of every sort were getting desperate in Germany, Konrad Zuse’s savior showed up in the person of a mysterious man named “Dr. Funk.” Dr. Funk was a physicist who had been drafted into the army and was looking for a way to avoid service. Zuse had no illusions—he told Dr. Funk he had nothing for him and sent him to Henschel to ask around for a position. Three days later, Dr. Funk returned with an exemption from military service. His powers only increased from there, Zuse suggests, by means of well-executed forgery. He did seem to know his way around—toward the end of the war, he managed to get Zuse, his assistants, and the machine safely away from Berlin and the encroaching Soviet army. But the evacuation was not without suspense:
The stairway was too narrow for the large relay cabinets; the only way to get them [out] was with the freight elevator. And once again at the wrong moment, the obligatory air raid alarm sounded. The power went out, and we found out just how helpless modern man is without electricity. The elevator had no hand crank, and the only way we could operate the winch was by hand, with indescribable difficulty. Millimeter by millimeter, we raised the device from the cellar to the ground floor. Then the Z4 was on its way for fourteen days on a heavily bombed route between Berlin and Göttingen. It had hardly been unloaded when the freight depot was hit.