The Man Who Invented the Computer

by Jane Smiley

  The trial in Minneapolis began on June 1, 1971.

  The question of whether the ABC had existed and the question of whether Mauchly pirated Atanasoff’s ideas for ENIAC were separate though related. What Mauchly and Eckert had fallen prey to with von Neumann and Goldstine’s 101-page publication of the ideas that led to EDVAC was a question of prior art—in typing up and sending out under von Neumann’s name the ideas underlying EDVAC, Goldstine established them as prior art to any claims that Mauchly and Eckert might make to the same ideas (von Neumann’s biographer, Norman Macrae, sees this as von Neumann’s intentional attempt to preempt the patenting of the ideas underlying the computer). If Atanasoff’s thirty-five-page description of the ABC had had the same sort of distribution as von Neumann’s paper (at least two hundred copies), then it would have stood as prior art. But Atanasoff had made only five copies on the assumption that because Iowa State was planning to patent the machine, it was dangerous to make more copies.

  Much of the case, especially Atanasoff’s testimony, revolved around the question of what ideas he had come up with and how he had come up with them. Because of this, the first part of his testimony was autobiographical—Attorney Henry Halladay questioned him about his childhood and his education in a detailed manner intended to delineate the steps by which he came to a set of concepts so unusual and innovative that other geniuses had not been able to come up with them, including Mauchly. Atanasoff obliged—yes, his fascination with his father’s slide rule had driven all other, more common passions like baseball out of his mind; yes, he had read his father’s books on engineering and his mother’s books on algebra, not because he was required to, but because he enjoyed them. His education at the University of Florida and the University of Wisconsin and Iowa State showed that he was a more-than-exemplary student (and chimed nicely with Judge Larson’s own career at another land-grant university, the University of Minnesota).

  Atanasoff was not the first to testify—Sam Legvold and others set the stage, so when Halladay brought Atanasoff’s testimony around to the subject of his years at Iowa State, it was easy to see that his teaching career in the thirties, and the evidence of not only his own work, but also the work of his students (one piece of evidence was the titles of papers his students had written under his tutelage) showed that he had thought through computing ideas for a long time and in more than one way. This history prepared the way for Atanasoff’s clearly remembered and detailed recollection of that night in the Rock Island tavern in December 1937.

  Halladay pressed him on two ideas, regenerative memory and logic circuits. Of the first he said, “I’m thinking about the condensers for memory units, and about the fact that the condensers would regenerate their own state so their state would not change with time. If they were in a plus state, for instance, they would stay in a plus state; or if they were in the negative state, they would stay in the negative state. They would not blink off to zero. Or if you used two positive charges, they would retain their individual identity and would not leak across to one another.”

  Concerning logic circuits, Atanasoff was honest about the fact that he did not perfectly visualize how the logic circuits would work. He imagined a black box, with input from two memory units—“the box would then yield the correct results on output terminals.” Although he did not envision the contents of the box specifically, he did understand that “since I was going to use condensers, why then I supposed the innards would be electrical in character, and I was well aware that the electrical entities which would be as suitable for such a purpose were vacuum tubes.” He explained that “condenser” was an archaic term for “capacitor.” Atanasoff then described how for the next fifteen months, he worked out these two ideas on paper: the idea for the regenerative memory was fairly simple; what was to be in the black box was much more difficult, but he worked that out for both a binary number system and a decimal number system. When he compared the two, it was evident that the decimal system would be too unwieldy. He declared that he had clarified his ideas by March 24, 1939, when he submitted his two-page grant application, asking for funds. Although the letter was short, it was detailed, describing the three sorts of problems Atanasoff expected his calculator to be able to solve (electrical circuit analysis, approximate solution of differential equations, and multiple correlation). His machine would be able to solve these sorts of problems for many more variables than was then practical with mechanical calculators. The letter also described previous efforts he had made to solve these sorts of problems using already invented methods. He asked for and was granted $650 (some $7,800 in 2010 funds). All of his papers were in order and were presented to the court.

  On the second day of testimony, more papers were presented. In fact, so many papers were presented—letters, notes, papers, diagrams, drawings—that Atanasoff began to weary of the tedium of court procedure, which meant putting descriptions of every piece of evidence introduced into the record. He was pleased, however, with the Des Moines Tribune article from January 15, 1941 (see this page). The importance of the article for the case was clear—the ABC was not a piece of junk that barely functioned, as Mauchly had gotten in the habit of saying.

  The next item on the agenda was Atanasoff’s version of Mauchly’s visit. He was equally detailed. Mauchly had arrived on Friday evening. Over the weekend, they had visited the computer several times and talked about it constantly—with only one small break, during which they spoke of Mauchly’s interest in meteorology. Mauchly had carried around the green-covered thirty-five-page description of the computer. Atanasoff had seen him reading it, and he and Atanasoff had discussed some of the things in the booklet that Mauchly wanted to understand. Atanasoff had explained the binary number system to Mauchly, though he was unsure how clearly the Philadelphian had grasped it. To Atanasoff, Mauchly had seemed eager to understand the ABC:

  “He seemed to follow in detail our explanations and expressed joy at the results, at the fact that these vacuum tubes would actually compute. He was shown addition and subtraction and multiplication and he was also shown the process of punching cards but we only had one unit in operation during his visit and we weren’t prepared to punch all of the thirty ‘Abaci’ simultaneously and no effort was made to fill the entire machine. He was shown the operation of converting base-ten cards to base-two numbers on the system, then the rest of the controls which we planned for the machine to make it operable in regard to solutions of simultaneous linear equations … We discussed logic elements in considerable length with Dr. Mauchly.” Halladay also introduced as evidence a letter Mauchly had written to a friend on June 28, 1941, only a few days after returning to Philadelphia. The third paragraph included the following: “Immediately after commencement here, I went out to Iowa State University to see the computing device which a friend of mine is constructing there. His machine, now nearing completion, is electronic in operation, and will solve within a very few minutes any system of linear equations involving no more than thirty variables. It can be adapted to do the job of the Bush Differential Analyzer more rapidly than the Bush machine does, and it costs a lot less.” The Sperry lawyer tried to get this letter excluded on the grounds that it was hearsay, but the judge allowed it.

  Even though the court procedures were tedious, Atanasoff’s answers were so detailed and self-reinforcing, since he rarely contradicted himself or seemed confused, that when the Sperry lawyer cross-examined, both on technical issues and concerning his relations with John Mauchly, he succeeded only in bolstering Honeywell’s case by giving Atanasoff the opportunity of adding more to the record. At one point, the lawyer asserted that Atanasoff had referred to Eckert as “a high-powered electronics expert.” Atanasoff coolly denied this and said that he had no knowledge of Eckert’s skills. The lawyer asked him why he hadn’t progressed with the naval computer when he was at the NOL. Ignorant of von Neumann’s funding machinations, Atanasoff replied that he had been short of both personnel and time—the navy had promised to relieve him of his ord
nance responsibilities but had failed to do so. His reply made perfect sense.

  When Atanasoff was finished testifying—seven days of direct examination and three days of cross-examination—he had made the best case he could that Mauchly had not only visited the ABC, but he had given every evidence of understanding the principles underlying Atanasoff’s theory of computing, as well as how he had realized these ideas in a piece of machinery.

  One of the star witnesses for Honeywell, who testified at the end of August, was Edward Teller. His job was not to say where he thought Mauchly had gotten his ideas, but to help Honeywell’s prior-use case against the ENIAC patents. According to Teller, the scientists at Los Alamos, thanks to the von Neumann connection, had made use of ENIAC for calculations concerning the feasibility of the hydrogen bomb in late 1945 and early 1946. The calculations were not especially accurate, but accurate enough to show Teller where he was in error and to suggest which direction he might go in when development of the H-bomb was resumed in 1949. The use of ENIAC for these calculations, and their significance as prior use, had not been employed in the previous trial that resulted in Sperry being awarded its patent. Its significance was in the fact that Mauchly and Eckert had not bothered to write up their patent application until August 1947, two and a half years after the machine was employed for the H-bomb calculations. It was a similar argument to the one that had been made about the EDVAC patents after the dissemination of von Neumann’s 101-page “First Draft.” Prior use was the second string to Honeywell’s bow.

  John Mauchly did not testify until November 1971. The Sperry lawyers had already discovered that Mauchly’s depositions were easily challenged: such assertions as the one that he had spent only an hour and a half with the computer, or that he had not seen it running, or that he had not seen it with the cover off were so easily disproved that Mauchly’s story had changed from deposition to deposition. The Honeywell lawyers knew how to press him because he had already given them plenty of ammunition.

  However, the Sperry lawyers did what they could to establish Mauchly’s credentials—like Atanasoff, he told his life story. Like Atanasoff, he outlined what he had done before ENIAC that might have pointed to his computer ideas. Then Halladay cross-examined him. Judge Larson had prohibited witnesses from hearing the testimony of earlier witnesses, so Mauchly did not know what Legvold, Atanasoff, Lura Atanasoff, and others had said about his visit (though he had read their depositions). Throughout his testimony, he persisted in denigrating the ABC and forgetting what was in the thirty-five-page description of the machine. Then Halladay began to cross-examine him, and Mauchly’s inability to remember fairly elementary aspects of his earlier inventions (such as whether his Harmonic Analyzer was mechanical or electronic) worked against him. He could not come up with any drawings or ideas he had made prior to meeting Atanasoff. He could call no witnesses who remembered talking to him about such devices, and he could not point to having invented a digital device—his Harmonic Analyzer was analog. He talked about having discussed electronic computing in his classes at Ursinus but could recall no student who could attest to these discussions. The only papers or notes he had about electronic computing were dated after he met Atanasoff in December 1940 or after he had been to Ames.

  One of the most striking pieces of evidence that Halladay introduced was a paper Mauchly had written in August 1941, two months after seeing the ABC, in which Mauchly had stated “computing machines may be conveniently classified as either analog or impulse types,” appending a footnote that read, “I am indebted to Dr J. V. Atanasoff of Iowa State College for the classification and terminology here explained.”

  At one point, Halladay showed Mauchly the thirty-five-page report, which the Ames people remembered him studying. Mauchly said that he had not read it very carefully, because he was not interested in the machine it described. Halladay pushed him, and he became resentful but finally admitted, in a roundabout way, that Atanasoff had told him that he could not take it back to Philadelphia, and so he must have asked to do so. Throughout the cross-examination, Mauchly quibbled and resisted, but Halladay did eventually establish several points—that after the twenty- to thirty-minute December meeting in Philadelphia, Mauchly had understood that Atanasoff was building a calculator based on different principles from the Bush Analyzer and that if he came to Ames, he could see it and Atanasoff would tell him about it. Concerning the June visit, Halladay established that Mauchly had been there for five days, that he had discussed the computer for many hours with both Atanasoff and Berry, that he had seen the ABC operate and read the report, that he had expressed enthusiasm for and understanding of the ABC and Atanasoff’s ideas after returning to Philadelphia, and that he had asked Atanasoff if he could use some of his ideas in a calculator of his own. It was also established that after he got back to Philadelphia, he had changed his career path and enrolled in the summer course in computing theory, where he met Eckert.

  In Mauchly’s defense, Scott McCartney reports that at the time of the trial, Mauchly was suffering from an illness that damaged his memory. I think we can also infer that Mauchly looked back at the ABC through the lens of ENIAC. There is no disagreement that ENIAC was a more complex and powerful computer than the ABC, and that it also owed some of its design and construction to the Bush Analyzer at the Moore School that had been designed by Irven Travis before he left for the navy. ENIAC was intended to perform a war-related function and had to be put together as quickly as possible, which was why EDVAC was designed—to finally realize the most advanced computing concepts without the pressures of speed or limited funding. There is also no disagreement that J. Presper Eckert and the others who worked on ENIAC contributed to the development of a sophisticated machine that was in some ways advanced (and in some ways not) compared to the ABC. What the Honeywell lawyers endeavored to show was a “sine qua non” or “without which not”—that without Atanasoff, Berry, and the ABC, Mauchly could not himself have come up with the ideas that led to ENIAC. Nothing Mauchly could or could not remember proved that he could have, whereas all of the Honeywell/CDC evidence showed that Atanasoff had done so.

  Mauchly might have had better luck in another country. Because he had to file his patent applications in the United States, he had to deal with a “first-to-invent” system (as opposed to a “first-to-file” system). In the U.S. system, invention is seen as both conception and “reduction to practice”—that is, more or less, making something. In order to get a patent, an inventor can’t just think something up, and he also can’t just make something—he must do both. Once the invention is made (or put into practice), however, the date of the invention is considered to be the date of conception rather than the date of filing. As a result, a patent application filed later can supersede one filed earlier if the inventor can prove both conception and diligence. It was pretty clear from the testimony that Atanasoff had been diligent in conceiving the computer and in “reducing it to practice.” But the United States is the only country that uses such a standard. In any other country in the world, Atanasoff would have entirely lost his chance to claim the ideas behind the computer when Iowa State and Richard Trexler failed to file his application, and Mauchly would have been awarded the patent.

  In this regard, it is also important to note that Mauchly could have avoided patent problems if he had been more careful, as GE lawyer George Eltgroth understood. If he and his lawyers had submitted material acknowledging and documenting what he had learned from Atanasoff in June 1941, as they were required to do, the patent examiner would have considered his claims in light of that material and determined if Atanasoff’s machine (and his thirty-five-page report) qualified as prior art. The fact that he did not do so left him open to having the patent abrogated for what is called inequitable conduct. But Mauchly and Eckert, possibly hyperaware of the commercial possibilities of the computer (for which McCartney, a writer for the Wall Street Journal, specially praises them) were loath to give any credit to others—when they filed their pa
tent, it covered more than a hundred different concepts, even though they were part of a large group working on the machine and Mauchly was also consulting Atanasoff from time to time on technical details of the ABC. It may be that when von Neumann was himself chatting up Atanasoff at the NOL, in late 1945, he was not only getting the benefit of Atanasoff’s ideas, he was also coming to understand that the computer as it existed in 1945 could not be owned by one or two men and was figuring out how to make sure that it would not be. It may also be that it was von Neumann’s insufficient credit to Mauchly and Eckert in his “First Draft” that put them in a possessive frame of mind when they were writing up their own application.

  At any rate, the Sperry Rand defense of having Mauchly forget as much as possible was the best the Sperry lawyers could come up with. It was certainly one that was congenial to Mauchly, however well it was or was not designed to work, and he stuck to it. Another irony of the case, which Charles Call communicated to Kirwan Cox, was that if Sperry had offered to share the patents for the same fee as they asked from IBM ($10 million), Honeywell would not have gone to court; but Sperry asked for $250 million before the publication of R. K. Richards’s book, and then $20 million afterward. Twenty million dollars was still too high for Honeywell, and so they went to court.

  The challenge to Sperry Rand’s patents was lengthy and involved. According to Clark Mollenhoff, it “consumed over 135 days or parts of days.” A total of seventy-seven witnesses had given oral testimony, and an additional eighty witnesses were presented through deposition transcripts. Honeywell had introduced 25,686 exhibits to be marked by the court; and lawyers for Sperry Rand and its subsidiary, Illinois Scientific Development (ISD), had directed the court’s attention to another 6,968 exhibits … The highly complicated trial transcripts stretched to over 20,667 pages.” Honeywell’s brief, filed in September 1972, was five hundred pages long. The key claim in the brief was not that Atanasoff had invented ENIAC, but “that there is no difference between what Mauchly learned from Atanasoff in June 1941, and what Eckert and Mauchly were later to claim to have invented alone.” The Honeywell brief went on to point out that even if the ABC had not worked, under U.S. patent law “one cannot claim a conception derived from another as his ‘original’ invention, even though he may have built the first device based upon that conception.” The Sperry brief, filed in August, rested its case on the fact that Sperry had already been awarded the ENIAC patents and that Atanasoff had not invented ENIAC, and that Mauchly had done a few electronic projects before meeting Atanasoff.