What is especially intriguing, and even moving, about Atanasoff’s story is that the machine he was trying to create was intended to mimic the brain—it was to be a self-organizing system, with feedback loops. The very mechanism that he pondered most that evening in the tavern was the calculator’s “regenerative memory”—the mechanism by which the capacitors and the vacuum tubes would charge one another, in a feedback loop. And without having a concept of how the human brain works, he also understood that electricity would be the medium of memory and thought, as it is in the human brain. Turing was thinking of a machine-like human process. Atanasoff was thinking of a human-like machine process.
Another way of looking at Atanasoff is that he fits into Malcolm Gladwell’s profile of a maven, that is, a person so interested in a particular field of endeavor that he not only drives himself to become an expert in that field but also is driven to communicate what he learns and intuits about that field to others. Atanasoff was widely considered to be a dedicated and effective teacher: he was good at explaining concepts to his students; he was good at probing their depth of knowledge; and he was good at encouraging them to learn what they needed to know. When his students came up with ideas, he helped them work them out, and he learned from what the students did. Atanasoff also had productive relationships with colleagues like A. E. Brandt. First and foremost, Atanasoff wanted to invent the calculator he thought the world of physics and mathematics needed—he seems not to have given much thought at the time to who might own what piece of the equipment, unlike the IBM executives who were offended when he fiddled with their machine. It also does not seem to have occurred to him that people at IBM could be offended, just as it seems not to have occurred to him that the authorities at his daughters’ school might be offended at his oft-expressed negative views on the science curriculum there. Atanasoff was intent upon innovation.
Through 1938, Atanasoff worked out both the practical and the theoretical implications of the ideas he came up with in the tavern in Illinois. To reiterate, Atanasoff’s four linked ideas were:
Electronic logic circuits (which would perform a calculation simply by turning on and off)
Binary enumeration (using a number system with only two digits, 0 and 1, rather than ten, 0–9)
Capacitors for regenerative memory (a capacitor is like a battery in that it can store electrical energy while not connected to a source)
Computing by direct logical action and not by enumeration (that is, by counting rather than by measuring; the numbers represented by rows of 0s and 1s, or the on-off states of the vacuum tubes, would be directly added and subtracted rather than being represented by points on disks or shafts)
One important consideration was how to stabilize the electrical supply of the vacuum tubes that would be doing the calculating. Atanasoff decided to construct the operating memory (CPU, or central processing unit) and the storage memory in different ways, in this case because vacuum tubes were expensive. He decided to reserve them for the operating memory and use capacitors for the storage memory. The results (including intermediate results) would be charred onto paper cards—still another type of memory. Capacitors (also known at the time as condensers) were (and are) very simple devices that store electricity like a bottle stores water. They store electricity without converting it to anything, using two conductors separated by an insulator. If a charge is applied to one of the conductors, it stays there by electrostatic attraction but cannot jump across the insulator. The charge can be removed very quickly by completing the circuit to the other conductor. In terms of the binary operation of a computer, “charged” can represent a 1 and “not charged” can represent a 0, for example. But insulators leak slightly, so the electric charge doesn’t stay there very long; therefore, both Atanasoff’s design and modern DRAM chips have electronics to refresh the state of the capacitor periodically by detecting its charge and restoring it before it fades.3
Atanasoff was perfectly familiar with condensers—when he was considering the dielectric constant of helium in his PhD dissertation, he was calculating the reduction in electric field strength caused by the presence of helium. Alan Turing was familiar with them, too—when he could make no progress finding the dielectric constant of water—that is, in calculating how effective water is at reducing electric field intensity. In the thirties, the most common insulator in capacitors was dry paper, which has a dielectric constant of 2, meaning that it cuts electric field intensity in half. Most modern capacitors now use ceramic insulation.
The idea Atanasoff had that most vexed and intrigued him over the next year was that the passive capacitors could work with the vacuum tubes. He later testified, “I chose small condensers for memory because they would have the required voltage to actuate the tubes, and the plates … of the tubes would give enough power to charge the condensers.” Atanasoff called this energy reciprocation “jogging,” as in “jogging one’s memory.” He thought that jogging would make both memories more stable while also saving on expense for supplies and on electrical usage. In this context, I think it is important to remember that Atanasoff was by nature and upbringing as frugal as he was ambitious, and also that he had no access to government money or private investment funds. Frugality was part of what drove him to invent a calculator—he didn’t want to waste time calculating using the machines of the day. Frugality dictated what he could try—he and Brandt had to experiment with what they had on hand, the IBM, not a Monroe. And frugality dictated the terms of his invention—it had to be cheap to produce, easy to operate, and cheap to run.
The use of electronic components both dictated the use of a binary number system and was dictated by it. If all Atanasoff needed to indicate a number was “on” or “off,” he was free of the burden of gears, shafts, measuring, and estimating, but once he was freed of those clumsy parts, he was committed to a binary number system, which he justified in two ways at the time—that his device would prove itself by being accurate, and that his device was intended to solve various sorts of mathematical problems including but not restricted to systems of equations,4 which meant that it was most likely to be used by scientists, who were more likely to understand a binary number system. It could also be said that using a binary number system is, as Zuse was pointing out to Pannke at the same time, the frugal choice.
Atanasoff spent a good deal of 1938 thinking about a mathematical system that would enable him to understand how to compute by direct logical operation, the way a person computes
by subtracting 6 from 7 and writing 1, then moving to the left and subtracting 2 from 3, and writing 1 to the left of 2, and then moving to the left and subtracting 0 from 1 and writing 1 to the left of 1, then seeing the answer as 111. Although it looks much more complicated to those used to decimals, binary subtraction would work the same way:
What he came up with was his own form of Boolean algebra (which he, like Zuse, later said that he was unaware of at the time). Here again, Atanasoff and Turing were thinking along the same lines, but Turing, as a mathematician among mathematicians, did not have to devise his own system.
Boolean algebra is a logic system invented by George Boole (1815–1864) that posits that there are only two values in the universe. They are zero and one. On these two values, four operations can be performed: (1) “no-op” (also called identity), (2) “not” (the value is changed into its opposite), (3) “and,” and (4) “or.” The first two operate (i.e., do something to and then return a single outcome value) on a single value. The second two operate on a pair of values and then return a single outcome value.
The values do not have to be read as numbers—they can be read as “true” or “false,” or “green” or “not green,” for example. For the purposes of the computer, both Atanasoff and Zuse realized that large numbers were easier to calculate using a 1 and 0 system, but Boolean algebra also has philosophical implications about the nature of reality and how to discover if something is true or not true that Turing brought to bear on not only breaking Germ
an codes, but also on his theory of how the mind works, and how, therefore, a mind-like machine might work. Working out his own form of Boolean algebra showed Atanasoff, as it showed Zuse, that his system was manageable and would not require rooms full of hardware.
Atanasoff didn’t have the money or resources to try to build any of his components, so most of the work he did was on paper and in his head. However, in March 1939, fifteen months after the revelation in the roadhouse, Atanasoff turned in an application for a grant of $650 to hire a graduate student and attempt to build what he had conceived of. In May, his grant request was approved: $450 was salary for the student and $200 was to go for raw materials.
1. The objects he calls “relays” bear no relationship in either looks or operation to what are now known as relays. They were entirely mechanical.
2. The term “bit” is an abbreviation for “binary digit.”
3. Thanks to John Gustafson, who adds, “Whenever you scuff your shoes on a carpet in dry weather such that you get a shock when you touch something metallic, you’ve made yourself a capacitor. Rubbing shoes on the carpet scrapes electrons from one surface to the other, creating an excess electric charge. The electric charge stays there because the charge cannot jump through the air, which serves as the insulator. If you ‘close the circuit’ by touching a metal object, the charge will suddenly discharge with a painful spark. Or if you stand still long enough, the static buildup will dissipate by itself, because even air conducts a little electricity.”
4. The list he eventually came up with was: (1) multiple correlation, (2) curve fitting, (3) method of least squares, (4) vibration problems including the vibrational Raman effect, (5) electrical circuit analysis, (6) analysis of elastic structures, (7) approximate solution of many problems of elasticity, (8) approximate solutions of problems of quantum mechanics, (9) perturbation theories of mechanics, astronomy, and the quantum theory.
Chapter Four
Perhaps Atanasoff’s greatest piece of luck in inventing the computer was that Iowa State College, unlike Cambridge, Harvard, or Princeton, had an excellent college of engineering, and from his years as a student and as a professor, Atanasoff was familiar with and on good terms with the engineering faculty. One day, walking across campus, Atanasoff ran into a friend, engineering professor Howard Anderson. The two started chatting about the sort of graduate student Atanasoff was looking for—intelligent, motivated, handy, and able to think for himself, as well as familiar with electronics. Anderson suggested a young man named Clifford Berry, who had just completed his bachelor’s degree. He was twenty-one. Anderson thought he was uniquely gifted.
Clifford Berry’s background was not unlike Atanasoff’s own. He had been born in Gladbrook, which was northeast of Ames, about halfway between Ames and Cedar Rapids. His father had owned an appliance store and was an accomplished tinkerer who repaired appliances and built the first radio in Gladbrook. Berry himself built a ham radio when he was eleven. In 1929, Berry’s father went to work for a power company (as Atanasoff’s father had done, and Howard Aiken himself had done) and the family moved to Marengo. Violence, too, had figured in Berry’s life—his father was shot and killed by a disgruntled worker when Berry was fifteen. Although the murder of Berry’s father resulted in hardship for his family, Berry’s mother recognized her son’s talents and sacrificed to give him the opportunity to go to Iowa State, where his abilities were recognized immediately—once again, we see the land-grant system at work, as it had been in Florida with Atanasoff and at Iowa State with Lura.
Physically, Berry could not have been more different from Atanasoff—he was short, slight, and wore thick glasses; his demeanor was self-effacing. When the two met in the summer of 1939, Atanasoff, the experienced teacher, questioned and probed the young man and found himself more than impressed. Berry was knowledgeable, enthusiastic, enterprising. When Atanasoff outlined his ideas, Berry was not shy about making good suggestions. He seemed to grasp Atanasoff’s concepts with no trouble, and the two men began to get excited about the project, though because of financial considerations they could not begin actual construction until September, the start of a new academic year. Temperamentally, Atanasoff and Berry complemented each other. Berry was neither intimidated nor overwhelmed by Atanasoff’s rush of ideas, and he was not subject to mood swings. He combined exceptional intelligence and mechanical dexterity with a steady work ethic and a mild demeanor. He was not the sort of boy who had ever been a pest—at twenty-one, he was only a few years past being an Eagle Scout.
In Berlin, Konrad Zuse was pressing forward in his magpie fashion: the German showing of the film King Kong inspired Zuse’s fraternity to put on a King Kong skit with paper skyscrapers. It starred a young man Zuse had never met, named Helmut Schreyer, as the ape. Schreyer made such an impression on Zuse that Zuse invited him to have a look at the computer. He writes in his autobiography, “I … was of the opinion that whoever was capable of such despicable deeds could also be of use in my workshop.” As soon as he walked into the room where the computer was, Schreyer, who was an electronics engineer, asserted, “You’ll have to make it with vacuum tubes,” but Zuse was hard to convince. He writes, “I never pursued the idea seriously, which may be attributed chiefly to my visual approach to the world. Things that could not be seen were always difficult for me to grasp.” It was during this time, when Zuse and Schreyer were hard at work (sometimes eighty hours per week) on their computer that they discovered Boolean algebra (or, as Zuse calls it, “propositional calculus”). The task was to come up with a mechanism that would switch the relays—or, once Schreyer had convinced Zuse, the vacuum tubes—on and off according to these operations (as Atanasoff also realized).
Zuse and Schreyer managed to build a vacuum tube switching prototype and demonstrate it (sometime in 1938) to a group of people who knew about and were interested in the computer at the technical university. Even as they watched the demonstration, though, the audience felt that what Zuse and his team were attempting to do was impossible. Zuse proposed a machine that would require two thousand vacuum tubes and several thousand of what he called glow-discharge lamps—almost ten times the number of vacuum tubes that the typical electricity transmitting station of the time employed (at least as far as Konrad Zuse knew), and so acquiring so many vacuum tubes would be difficult, if not impossible. He also says in his autobiography that when pressed to think of facilities that used as much power as the computer they were designing, he mentioned high-speed wind tunnels in the aircraft industry—the local power station had to be notified before one of these was turned on. After the demonstration, Zuse and his team decided to resume their former secrecy—they realized that they appeared, at least, to be attempting something that no one believed possible. Schreyer did complete his doctoral thesis about the project, but the computer itself seemed to have no future.
Zuse, not in sympathy with the Nazis and used to secrecy, focused on building his machine. Of the coming of the war, he writes, “It is not true that virtually all news in a totalitarian state is false. On the contrary, most news is completely correct, albeit tendentiously slanted; it is just that certain information is suppressed. One can adjust for the political slanting of the news, but there is virtually no way to fill in the omissions.” If we wonder how Zuse and his friends failed to understand the Anschluss, the takeover of Czechoslovakia, the pact with Mussolini, and the pact with Stalin, well, they were not alone in assuming that Hitler’s intentions were not particularly aggressive and could be ignored, especially since they managed to meet up with some young British men who traveled to Germany in August 1939. All of the German young men felt warm friendships toward the visitors, but when Germany invaded Poland and Britain declared war, the visit was cut short. Soon Zuse was drafted. He expected to be in the army for six months.
Things were quieter in Ames. Once Atanasoff and Berry were ready to build, they had another piece of luck that at first looked a bit like a slight. When Atanasoff sought a workspace in which to
build his computer, he was sent to the basement of the physics building because the first floor was taken up by projects already in progress, and possibly considered to be more important. The basement was full of junk, and Atanasoff and Berry had to set up their workspace in an out-of-the-way, windowless corner (walls with doorways were added later—to the detriment of the computer). Atanasoff appreciated the privacy—later he would also appreciate both the space’s proximity to the machine shop and its steady, cool temperature. In Atanasoff’s life, frugality had often meant that making do was making better. But space limitations and financial limitations also meant that Atanasoff and Berry had to reconfigure their plans into something smaller and more buildable. As Atanasoff later put it, “We did not dare to build everything into our plans. Our skill as inventors depended on how well we chose between these factors, the indispensable and the impossible.” What Atanasoff thought was most crucial were the instructions—that is, how he would set up the sequence of steps, or algorithm, that would perform the mathematical operation he wanted. He was still not quite clear about them when they went to work.
While they were getting ready, Atanasoff did what he could to find out what other computer projects were under way around the country. As far as he could tell, no one was trying what he was trying, in terms of concepts or hardware. Someone he might have heard about was George R. Stibitz (born April 30, 1904, though some sources say April 20). Stibitz had grown up in Ohio, gone to Denison University, and then to Union College. In 1930, at the same time Atanasoff was working on his dissertation at the University of Wisconsin, Stibitz was working on his at Cornell University. Stibitz’s dissertation, like Atanasoff’s, involved extensive and tedious calculations. But Stibitz went to work for Bell Labs in New York City. Bell Labs, a joint enterprise belonging to Western Electric and AT&T, was in both the discovery business and the invention business. In 1932, Karl Jansky had detected radio noise that originated at the center of the Milky Way; in 1933, Bell Labs scientists had managed to transmit stereophonic sound over telephone wires (a symphony recorded in Philadelphia was transmitted to Washington, D.C.). Stibitz, whose doctorate was in applied mathematics, was surrounded by equipment as well as engineers.