The Man Who Invented the Computer
Page 14
Berlin was about 210 miles from Göttingen—John von Neumann and his friends at the University of Berlin had been accustomed to traveling back and forth between the two universities in the 1920s, taking about three hours each way. And Dr. Funk had divined the way to save the machine, as well—for its travels, he christened it, not the Z4, but the V-4 (for Versuchmodell, or “Experimental Model” 4). He allowed those in charge of transportation and evacuation to believe it was a “Vergeltungswaffen” 4, or an advanced version of the V-2 rocket.
In Göttingen, Zuse and his assistants assembled and demonstrated the machine—it still worked—but they were then ordered to take it to “one of the underground ordnance factories,” tunnels where thousands of concentration camp prison workers manufactured weapons and ammunition in appalling conditions. Surprised, shocked, and frightened by what he saw there,1 Zuse managed yet another evacuation, this time to Bavaria. Dr. Funk procured for the journey a Wehrmacht truck and one thousand gallons of diesel fuel.
“For fourteen days we fled along the front, past burning neighborhoods and over bombed-out streets. We usually drove at night; during the day we found makeshift shelter with the farmers.” When they got to their destination, they discovered Wernher von Braun and his team (the designers of the real V-2 rocket). They ended up at the same temporary quarters as von Braun—possibly the most prominent scientist in Germany—thanks to Dr. Funk: “Dr. Funk had free run of the place, and even after we left Berlin, he obtained papers firsthand, whenever it was necessary. How he was able to find us a place in Oberjoch [on the Austrian border] remains a mystery to me to this day.” Zuse did talk to von Braun once—they were close in age and had attended the Technical University of Berlin at about the same time. Zuse was not especially impressed, because he did not get the sense that von Braun foresaw much use for computers in future rocket travel. Von Braun said nothing of his plans to “go over to the Americans. We soon felt it better to keep away from them and to look for our own quarters.” Some years later, though, upon reading von Braun’s memoirs, he saw that von Braun had understood their perilous situation better than he had at the end of the war—an SS man told von Braun that storm troopers had been billeted among the scientists with orders to shoot them “to keep you from falling into the hands of the enemy.” Major General Walter Dornberger, who was in charge of von Braun and the V-2 rocket, managed, with the help of several shots of cognac, to elicit the plan from the commander of the SS, and then to persuade him to abandon it (“And when the Allied troops have learned that you carried out a bloodbath, you will be hanged immediately!”).
Although the war was ending and the French were gaining control, surrendering was a complicated business—first the Zuse cohort used their truck to move the Z4 to the village of Hinterstein, Austria, some 125 miles farther east, where they hid the machine in a cellar. Dr. Funk then tried to make contact with the Americans nearby but was arrested, though he was soon released. In Hinterstein, they encountered a local eccentric, an Indian soothsayer who had a way of knowing, or seeming to know, about everything that was going on, including atom bombs and vast caches of food. He was interrogated by occupying French authorities several times; information he gave them came to nothing, so that when he told them that “a large computing machine—which he [the soothsayer] had invented—was hidden in the village,” the French authorities didn’t bother to investigate. Subsequently, a local Englishwoman, a duchess who had lived in the village for a long time, did report to British authorities that there was a V-4 rocket in the village. When the British investigated and found only the Z4 computer, “they left, disappointed.” Not long afterward, Dr. Funk, Zuse’s mysterious savior, disappeared, too.
Zuse, his wife (he had married one of his employees in January), and his machine stayed in Hinterstein, living as best they could on limited means—they foraged for firewood and food, often eating nettles, spinach, wild mushrooms, and snails. He also managed to sell small paintings of local alpine chamois in a souvenir shop owned by his landlord. For his own pleasure, returning to his love of art, he made intricate woodcuts of the scenery. The scene was more pastoral than Zuse was used to, which led him to think in new ways—he turned his attention to software rather than hardware, spending the next two years on a theory of computer programming that he called “Plankakul,” or “plan calculus,” an “algorithmic computer language” that led him to think about the nature of computer logic. He writes, “This environment did anything but nurture the concept of mechanizing thought processes … the Allgau’s flower-strewn surroundings and—not to be forgotten—the childish laughter of my first son were not exactly conducive to analysing the world into yes/no values.” Like Alan Turing, and at around the same time, Zuse began to think about the nature of the mind, the nature of human free will, and even the nature of the universe. He wrote a paper, uncompleted, that he called “Freedom and Causality in the Light of the Computing Machine.”
In 1946, Zuse moved the Z4 to a stable, where he, his wife, and their two children also rented a room. But the machine wasn’t doing anything—“although we could have taken over fat content analysis for the local alpine dairy.” And once again, there were no supplies for working on the computer—“We joked that the Americans had forgotten only one thing—their soldiers carelessly threw away tin cans. But we really did collect and use such garbage.” In 1947, Zuse and his friends, still living in the Austrian Alps, now in the village of Hopferau, with the Z4 in the stable, began to make contact with the outside world when the trains resumed service (though the trains were so crowded and dangerous that “we were happy just to arrive home safe from our travels”).
Zuse and another friend named Stücken decided to found an engineering firm. Every single item they might need to continue work on the computer was hard to attain, but, he writes, “our courage resulted not least from the fact that we felt we had nothing to lose.” His old friend Helmut Schreyer had a different idea—he had met a South American businessman who wanted him to pursue his computer ideas in Brazil, and Schreyer tried to talk Zuse into joining him. Years later, Zuse was glad he had declined—when Zuse managed to visit him in Brazil, Schreyer was working three jobs, and the suitcase of computer parts that he had managed to salvage after the war had been stolen on the train between Hopferau and the town of Erlangen.
But Zuse’s courage did not extend to believing that his machine had much of a future, and later he deeply regretted that he didn’t bother to file patents on what he had invented. Part of the problem was formulating his insights into patentable ideas—he and a friend who was later to become a patent attorney believed that his thoughts about mathematical and logical relationships would not get through a system that was more geared to devices. He had filed patents in 1937 and 1941. His 1937 patent was granted, but it took so long that it was worthless by the time he got it. His 1941 patent was denied in 1967, with the reason that “the innovation and progressiveness of the object concerned in the main application are not doubted. Yet a patent cannot be granted due to insufficient inventive merit.”
In 1949, Zuse got lucky. One day “an elegant car from Switzerland” drove up, and a man from the Swiss national technical institute in Zurich got out and asked around about a computer he had heard was to be found in the village. The man, a Professor Stiefel, had recently returned from the United States, where he had been shown all sorts of computers “in beautiful cabinets with chromework.” Zuse took him to the stable and turned on the machine. Professor Stiefel presented a problem, a differential equation, and the Z4 solved it. Stiefel then leased the Z4, which stayed in the stable, and Zuse received a small monthly payment for its use.
At the end of the war and right afterward, it was clear that technological advances during the war left research questions related to the war that needed to be answered, but research personnel were quickly returning to civilian life; indeed, Iowa State asked Atanasoff to come back as head of the physics department. For him, though, projects for the navy took precedence ov
er teaching, and Lura and the children were no longer in Ames. While the foremost of Atanasoff’s projects was the plan to build the navy computer, he had not been relieved of his duties in the Acoustics Division. Atanasoff had no choice but to attempt, by working even harder, to run both the Acoustics Division and the Computer Division at the same time. In the Acoustics Division, he had two main projects, the first which was to travel to Bikini Atoll, the scene of atomic tests in the summer of 1946. The immediate purpose was to test the effects of atomic blasts on the junked hulls of ninety-five surplus ships. The assignment for the Acoustics Division was to measure sound waves set off by the tests, with an eye to future detection of atomic tests by other nations. At Bikini Atoll, Atanasoff was put in his usual position of making do, scrounging, repairing, and do-it-yourself, but the tests were both successful and interesting—the column of water discharged by the second, underwater atomic blast rose a mile into the atmosphere and “launched” the aircraft carrier Saratoga (which displaced more than 38,000 tons) almost half a mile. Atanasoff’s acoustic results set a standard for subsequent detection of atomic explosions. It was when he returned from Bikini that Atanasoff was informed that the navy had dropped the computer project. One result of the navy dropping the project was that the “need to know” request Atanasoff had submitted to the navy in order to find out the workings of ENIAC became moot. He would not find out this information until years later.
However, the Acoustics Division at the NOL had another big project. Helgoland Island, about sixty miles north of Bremerhaven, west of Jutland, had served as a German ammunition dump, and the British had decided to blow it up. The navy wanted to take acoustical readings on the shock wave that would be produced, a kind of man-made earthquake. Atanasoff was put in charge of the project. The detonation was to take place in mid-April 1947. Atanasoff had eight weeks to prepare. He subsequently learned through the grapevine that several other scientists had been approached to oversee the project and had refused, thinking that the lead time was too short. He was even advised by a colleague not to accept the assignment, but he did so and accomplished what was asked in his standard way—by noting what was wrong with the preliminary plan, resurrecting old ideas for a seismograph he himself had once designed, then modifying existing equipment to measure seismic waves and sonic waves, no matter how large they might prove to be.
In the meantime, the postwar declassification of ENIAC had other ramifications—when ENIAC’s security was lifted in 1946, the scientific and technological world reacted with oohs and ahs. Tommy Flowers realized that he had invented and made use of a more advanced machine, but he was in no position to protest: Colossus would never be on his résumé. He writes:
With no administrative or executive powers, I had to convince others, and they would not be convinced. I was one-eyed in the kingdom of the blind. The one thing I lacked [for pursuing a computer project] was prestige, which knowledge of Colossus would have amply provided. Personal rivalries also played their part. These were exacerbated, and some were even provoked, by what was considered pretentiousness on my part. Little or none of that would have been possible had Colossus been known.
One person who, of course, knew all about Colossus was Alan Turing. The end of the war meant that Turing had several options available to him. In June 1945, he received an Order of the British Empire for his war work, and then he accepted a position at the National Physical Laboratory with the goal of developing a general-purpose computing machine. The NPL was about thirteen miles southwest of central London, in Teddington. The primary work of the NPL was akin to what was then being done in the National Bureau of Standards (now the National Institute of Standards and Technology) in the United States—it established systems of measurement and standards of quality that would then form the basis for the systematic manufacture and production of goods. The British government had realized in the course of the war that the problem of calculation that had frustrated Atanasoff, Turing, and almost every other physicist before the war was going to be a limiting factor in postwar consumer society, and so a new mathematics division of the NPL was begun and a Cambridge man named J. R. Womersley was put in charge of solving the problems of calculation. The head of the whole laboratory was Charles Galton Darwin, grandson of Charles Darwin and son of astronomer George Darwin.
In spring 1945, right around the time that the order was going out for the ten Colossus machines to be destroyed, Womersley went to the United States and was shown ENIAC (before, in fact, it was unveiled to the general public). When Womersley got back to the UK, he was eager to build a UK version. Since, unlike Mauchly and Eckert, he happened to be quite familiar with “On Computable Numbers” and had even toyed with designing a mechanical version of a Turing machine before the war (his partner, like Mauchly and Eckert’s original partner, was in the horse-racing pari-mutuel totalizer business), he offered Turing £800 per year—£200 more than he had received at Bletchley Park—to come to the NPL. Turing began work on October 1, 1945, and he was ready with plenty of ideas. Many of his new colleagues at the NPL had also been recruited from the war effort, though from the Admiralty Computing Service, not from Bletchley Park. They were doing calculations on analog desktop calculators.
Turing did not reciprocate Womersley’s respect or get along with him; he was openly contemptuous of Womersley’s shaky grasp of mathematical principles and had no appreciation of the political skills that had allowed Womersley to extract the financing for his section from the increasingly parsimonious British government. In spite of the difficulties, though, Turing understood that this was his opportunity to realize the theory behind “On Computable Numbers” in electricity and hardware, and, indeed, the theory that had been realized in Colossus. He set about doing so, writing a report that laid out his theory and design of a computer called “Proposed Electronic Calculator.”
The basic feature of his design was a large memory and the ability to program it (that is, to supply the computer with a set of instructions that the computer could always consult—the program would always contain an instruction for the next step, just as the “computer” in “On Computable Numbers” would always know whether to add or not to add the next number on the infinite tape), so human input would be minimized. And the large memory was to be very fast (no doubt Colossus had shown him how fast a computer could operate).
Turing had a copy of von Neumann’s “First Draft” (who did not?), and he considered his own ideas to contrast decidedly with von Neumann’s, especially in that he expected to construct his memory not like a paper tape, as in Colossus, which would be long and sequential, presenting the problem to the computer of “finding” an instruction somewhere on the tape, but more like wallpaper on a wall, allowing the computer to quickly scan for instructions. The former is called “serial access memory,” the latter, “random access memory.” The shift from one to the other is, according to computer scientist John Gustafson, “almost as big a deal as going from decimal to binary calculation.” Turing’s proposal, in terms of both theory and engineering, was quite specific. According to Jack Copeland, he “supplied detailed circuit design, full specifications of hardware units, specimen programs in machine code, and even an estimate of the cost of building the machine.” It seems likely, comparing this production with his prewar efforts at computer design, that he had learned as much from Tommy Flowers (and the other engineers he had known in the war) as Flowers had learned from him.
When Womersley and Turing made their proposal in March 1946, the meeting went well enough that Darwin granted them £10,000 ($400,000 in 2010 funds) to try out a small prototype, but not so well that they got enough money to build the machine that Turing really wanted to build. Certainly, the same problem obtained with the ACE (as it was called, standing for “Automatic Computing Engine”) as obtained with Tommy Flowers’s efforts in the same direction—so few knew what had been done at Dollis Hill or at Bletchley Park during the war that no one was prepared to give Turing the respect or the benefit of the do
ubt that his experience warranted. Darwin, who knew more than most, did request the Post Office to allow Flowers to help with the computer, but with the ACE, Turing was in much the same position that Atanasoff had been in in 1940 with the Iowa State College Research Corporation—his ideas were so advanced that he had to prove they were worth something to people who did not really understand them. Womersley was his advocate and had some political skills, but Turing himself had none—he needed to be able to refer those who controlled the money to his wartime résumé to convince them, but he was forbidden to do so.
He also had a rival for funding—Maurice V. Wilkes, at Cambridge. Wilkes was almost exactly Turing’s age and he had also gone to Cambridge (St. John’s College, in mathematics). He had also joined the war effort, but in radar development rather than code breaking. In 1945, when Turing was heading to the NPL, Wilkes was returning to Cambridge, to the Mathematical Laboratory. Wilkes also read the “First Draft” when it was published, and he was inspired by it to get to Philadelphia and attend the last two weeks of the Moore School Lectures. He traveled around the United States and investigated as many computer projects as he could before returning to England. Unlike Turing, his goal was not to innovate—it was to supply the university with a working computer as quickly as possible. He visited the NPL at the end of November and wrote to Womersley in December. Womersley apparently either did not understand Turing’s ideas or did not understand Turing, because he passed the letter to him, who wrote back rather sharply: “The code he suggests is … very contrary to the line of development here, and much more in the American tradition of solving one’s difficulties by means of much equipment rather than thought … I favor a model with a control [that is, a CPU] of negligible size which can be expanded if desired.” Turing thought that if the hardware was fast enough and the program detailed and complex enough, roomfuls of processor units could be avoided. However, such a machine would have had difficulties of its own, according to John Gustafson, who maintains,