by Kurt W Beyer
After Hopper quickly familiarized herself with flow charts and C-10 code for UNIVAC, her main task during the summer and fall of 1949 was to begin to prepare a library of useful applications for the UNIVAC. Hopper brought to EMCC 5 years’ operational experience that she applied to the task of developing a series of useful subroutines and generic programs. Like the relative code packages developed at Harvard, Hopper and her new team of programmers first constructed subroutines that could generate square roots, sine, cosine, tan, log and other transcendental functions. Each routine had flexible addresses so it could be placed into any specific portion of memory, and accuracy was made available up to eleven decimal places. During this time Hopper also developed a unique package of programs to solve Laplace boundary value problems that was ideal for solving heat and wave equations in engineering.44
In order to validate the efficacy of programs written for a computer that did not yet exist, Hopper employed the recently completed BINAC. The small binary stored-program electronic digital computer was supposed to be delivered to Northrop Aviation on 15 May 1948, but the machine was not operational until a month after Hopper’s arrival. The machine was smaller but similar in design to the proposed UNIVAC, with a fast vacuum-tube processor and 512 words of memory provided by mercury tanks. “I would write in the C-10 code a subroutine to compute the sine and I’d transliterate it into BINAC code and check it out on BINAC so that we would have something ready to run when we got UNIVAC running,” said Hopper.45 Although the machine calculated in binary and could not represent alphabetic characters, UNIVAC I programs written in C-10 could be translated into octal and thus processed.
As Hopper settled into her new position, she took it upon herself to enhance the EMCC staff with talented programmers. “Some of the Harvard people I managed to attract to come down there, like Dick Waltman and Charles Katz,” she said. “I brought a lot of people in down there.” Other key additions included Hugh Livingston who was one of the enlisted operators assigned to Mark I, and most notably, Herb Mitchell, the first person to receive a doctorate from Aiken’s Computation Laboratory.46
Hopper soon realized, however, that as of 1949 the pool of competent programmers was hardly larger then her immediate circle of friends, and that in the immediate future EMCC would have to “create” programmers. To make such a training program less hit or miss, Hopper and Mauchly attempted to identify the traits and abilities that they should look for when interviewing candidates. In the end they agreed on fifteen different characteristics, with three of the fifteen highlighted as essential. These included a systematic nature and methodical approach to new problems, careful reasoning without jumping to conclusions which are unwarranted, and imagination and creative ability. John Mauchly even devised an “aptitude test” for candidate assessment, though the subjective nature of some of the traits made personal interviews necessary. In the end, Hopper and Mauchly’s efforts helped to identify the aptitude boundaries for a new profession, and EMCC began to make programmers, one at a time.47
Unbeknownst to Hopper, her efforts to build a team of programmers in order to build libraries of applications and subroutines were being mirrored by another group thousands of miles away at Cambridge University. Maurice Wilkes was assembling Britain’s first stored-program electronic binary computer, the EDSAC, and in June 1949 he organized a group of programmers headed by David Wheeler to construct a library of programs. “We settled down to work on the library of subroutines and programming research in real earnest,” wrote Wilkes. “The group of programmers—for they could now be called that—met regularly, and each member undertook the production of one or two subroutines.”48 Wilkes, Wheeler, and Stanley Gill eventually published the first work on programming in 1951. Titled The Preparation of Programs for an Electronic Digital Computer, it recorded the fruits of their efforts. It became standard reading for the next generation of computer programmers.
One year before the publication of their seminal work, Wilkes visited the United States to get a sense of developments there. While visiting Eckert-Mauchly, he noticed the similarities in work between EMCC’s programming group and his own. “I do remember some stimulating discussions with the group, led by Grace Hopper, who were concerned with programming.” he recalled. “I found that they had a full appreciation of the importance of programming and of the need to develop organized and disciplined methods. I felt I was among people who looked at things in the same way that I did myself .”49
Hopper’s focus on the UNIVAC I’s library of programs left Betty Snyder time to develop another important yet unheralded contribution to early programming advances. This was the sortmerge generator for a magnetic-tape computer.50 In order for the UNIVAC I to be useful as a business machine, it had to be able to sort through and order data as effectively as existing punch-card technologies. Snyder first began working on the problem during the summer of 1947. With no formal training in logic, she applied common sense to figure out a binary method of sorting. “I remember bringing a deck of cards to the office on 1215 Walnut Street and turning them face down and making decisions and building piles,” she recalled. “I built the logic of binary sorting from a deck of cards without ever having . . . anybody tell me anything about how it ought to be done.”51
Once Snyder had developed the theory behind efficient binary sorting, she created a program that she called a generator. The generator contained the characteristics of the data to be sorted: number of items, length of items, categories or fields of information within the data set, etc. When a specific query was entered, the generator would “generate” another program, which would instruct up to ten different tape drives to take certain data off several reels and feed it back in a specific order onto another set of reels. The process was continued until a final tape contained the desired information in the proper order it was requested.
The coordination and automation of high-speed tape drives by means of a computer program was a monumental task that required a variety of technical problems to be solved. Working with the UNIVAC I engineers, Snyder developed a series of “interlocks” that scheduled the operation of input and output devices. Thus, only one magnetic-tape reel could function at a time, and a reel could not function simultaneously as an input and an output device. Like Hopper, Snyder took advantage of a functioning BINAC to aid in the design of UNIVAC I’s tape system, which represented a new development itself in computer design. “The first time a computer was used to design a computer,” remarked Hopper.52
AN UNTIMELY DEATH
By the fall of 1949, the future of the Eckert-Mauchly Computer Corporation looked bright. So did Grace Hopper’s own future. Hopper had found a new home among the diverse members of a start-up company with the only functioning stored-program computer in America. She had a far-sighted boss in John Mauchly, one of the finest engineers in Presper Eckert, and around her a team of talented engineers and programmers. Betty Snyder had helped her learn to program the first commercial stored-program computer. In return, Hopper shared her 5 years of programming and management experience, and helped build a strong programming team by bringing in some of her former Harvard colleagues.
But just as the company was poised to become the first significant mover in a new and potentially lucrative business sector, the unthinkable happened. On 25 October 1949, a plane carrying Henry Straus, the chairman of the EMCC’s board of directors, crashed just outside of Baltimore. As John Mauchly stated in his annual president’s report to the shareholders of EMCC, “Not only did the Corporation lose a wise and far-sighted leader, but the entire personnel of the company felt keenly the loss of an understanding and sympathetic friend.”53
And sympathy was something EMCC’s senior management needed as the company moved into a new decade. Despite the technical successes achieved in hardware development and programming, cost overruns, poorly negotiated contracts, and production delays threatened the financial solvency of the company. To make matters worse, the provisions of the contract between Straus’s c
ompany and EMCC were no longer adequate to see through the construction of the first UNIVAC. “It is now apparent that, even if the American Totalisator Company were to continue to supply operating capital in return for preferred stock up to the limit contemplated in their contract with us, we would not have adequate funds to carry through the test period of the first UNIVAC,” stated Mauchly in his report to shareholders.54
Mauchly’s appeal for more funding after Straus’s death could not have been more poorly timed. The accident killed the person at Totalisator who believed most in the potential of the electronic computer. The Munn brothers, the majority owners of Totalisator, had little interest in or understanding of digital computers, and instead of answering Mauchly’s plea, immediately placed EMCC up for sale. To make matters worse, Totalisator demanded repayment of a $62,000 advance by January 1950.55
Both Hopper and Snyder remembered the shock of losing such a “marvelous man,” as well as the malaise that came over the company. Instead of concentrating on the UNIVAC, Eckert and Mauchly were now forced to focus on the immediate need for an infusion of funding. They initially turned to loan companies and research foundations, but with little success. With their backs against the wall, the neophyte businessmen hoped to sell EMCC to a major corporation that specialized in calculating equipment (National Cash Register, IBM, Burroughs), or to a firm that had a need for advanced calculating technology (Hughes Aircraft).
If there was any glimmer of hope, the successful demonstrations of BINAC during the late summer were attended by executives and engineers from many of these companies. Among the attendees were representatives from International Business Machines Corporation, and soon after Straus’s death Eckert and Mauchly were invited to meet with the aging Thomas Watson Sr. and his heir apparent, Thomas Watson Jr. In his memoirs Watson Jr. vividly recalled his meeting with the two young inventors, and his impression of them would stay with him from that point on:
I was curious about Mauchly, whom I’d never met. He turned out to be a lanky character who dressed sloppily and like to flout convention. Eckert, by contrast, was very neat. When they came in, Mauchly slumped down on the couch and put his feet up on the coffee table—damned if he was going to show any respect for my father.56
Eckert explained the technology behind their work, while Mauchly shared his vision of the computing future, a future that jeopardized the technological monopoly held by mechanical punch-card machines. Because of the slow speeds and limited storage of punch cards, they were not an optimal input-output system for electronic computers. For instance, a handful of UNIVAC’s magnetic-tape discs could store the policy records of an entire insurance company, while an individual punch card was needed for each policyholder. Moreover, those individual punch cards could only be sorted and processed at mechanical speeds. Electronic circuits, combined with Betty Snyder’s sort-merge system, could process, store, erase, move, and save magnetic records quickly and securely.57
At the end of the meeting, Thomas Watson Sr. thanked Eckert and Mauchly for their presentation, but frankly told them that, for anti-trust reasons, IBM lawyers advised that IBM’s purchasing the Eckert-Mauchly Computer Corporation could have negative legal ramifications. According to Thomas Watson Jr., the real reason that Watson Sr. rejected Eckert and Mauchly’s offer ran far deeper:
Having built his career on punch cards, Dad distrusted magnetic tape instinctively. On a punch card, you had a piece of information that was permanent. You could see it and hold it in your hand. Even the enormous files the insurance companies kept could always be sampled and hand-checked by clerks. But with magnetic tape, your data were stored invisibly on a medium that was designed to be erased and reused.58
An aging CEO had built his career and corporation on mechanical punch-card systems, and he wholeheartedly believed that Eckert and Mauchly’s electronic computer was an interesting but unproven business machine destined for the technological ash heap. IBM’s engineers at Endicott, Watson Sr. believed, should focus on building a better punch-card system, not on experimenting with marginal technologies.
Up until the 1949 meeting with Eckert and Mauchly, Watson Jr. had shared his father’s sentiments. In fact, he had attended the formal demonstration of the ENIAC in February 1946 with the number-two man at IBM, Charles Kirk, and had walked away unimpressed by and unconcerned with the temperamental monstrosity. “The truth,” Watson Jr. recalled, “was that I reacted to ENIAC the way some people probably reacted to the Wright brothers’ airplane: it didn’t move me at all. I couldn’t see this gigantic, costly, unreliable device as a piece of business equipment. Kirk felt the same way.”59 But by 1949 the younger Watson had begun to reconsider. Besides the increasing attention the popular press was giving to computers and electronics, Watson Jr. acknowledged the unexpectedly robust sales of a marginal IBM product. “It was the success of the 604 Electronic Calculator that convinced me that electronics was going to grow much faster than anyone had anticipated,” he recalled.60 The IBM 604 was an electronic version of a popular IBM desk calculator. It had been developed by a small group of electronics experts working in Poughkeepsie rather than in IBM’s main laboratory in Endicott. The group, headed up by Ralph Palmer, had applied experience gained with radar during the war. Standard IBM plug-in boards were mounted with vacuum tubes and supporting circuitry in such a way that the new electronic innards could be mass-produced at a reasonable cost. “Palmer and his men,” Watson Jr. recalled, “had produced an amazingly elegant design that made it easy to cope with vacuum tubes, which were constantly burning out or otherwise going haywire.”61
Although the unexpected sales of the 604 were seen as an anomaly by many of IBM ‘s top executives, Watson Jr. viewed it as a sign that, at the least, IBM had to learn more about the new technology. He offered Presper Eckert a prominent position at the Poughkeepsie laboratory, where he would be able continue work on his envisioned computer. With so few options, and with bankruptcy looming, Eckert seriously considered the offer, but in the end he decided to remain with Mauchly. “I certainly was influenced by him [Mauchly] and by my wife,” Eckert recalled. “My own feelings about doing it [alone] were much less stronger than his. . . . I don’t think Mauchly recognized the difficulties in doing it ourselves. I think I had a better sense of that.”62
Since EMCC was not joining the International Business Machines Corporation, they were competitors. “The IBM Company,” Mauchly recalled, “was telling its customers that it’s absolutely ridiculous for the tape machines they’re talking about down there, because you know you can’t sort data on tape. You can only sort it with cards because you move the cards around.” Watson Jr. confirmed Mauchly’s suspicions in his memoirs: “Dad told the marketing men to call on Prudential and persuade them that the UNIVAC idea was not sound.”63
A PERSONAL SHORT CIRCUIT
With the decision to keep EMCC intact for the time being, Eckert and Mauchly spent the remaining months of 1949 seeking financing. Hopper remembered the period as one of uncertainty, scarcity, and low morale. Not only was the future in question; there was no money to pay the bills in the present. Hopper and others accepted salary reductions, and purchases of equipment for the UNIVAC were put on hold. Instead of dreaming about the coming computer revolution, discussions now turned to what to do if EMCC were to fold.64
Hopper had thrown herself wholeheartedly into Eckert and Mauchly’s business venture, and its deterioration sparked a crisis in her life. The 43-year-old programmer’s furtive addiction to alcohol resurfaced, and on a cold Saturday night in November she was arrested for public drunk and disorderly conduct. Released to Philadelphia General Hospital for treatment, she was eventually placed into the custody of her friend Edmund Berkeley. It was far from the first time Berkeley found himself covering up his friend’s growing addiction. The incident prompted Berkeley to write a passionate “intervention letter” to Hopper, copies of which he sent to John Mauchly and to other close friends and relatives. Berkeley emphasizes from the onset that the l
etter was not written to hurt his friend’s feelings, but rather out of deep love and sincerity. He justifies distributing copies of the letter under the pretext that there is nothing to hide, because so many people know Hopper’s “secret”—even marginal acquaintances who, according to Berkeley, routinely ask “Has Grace stopped hitting the bottle?” Though Berkeley emphasizes that “a person changes not for rational reasons but for emotional reasons,” the rest of the letter provides a detailed and rational analysis of Hopper’s worsening addiction.65
Grace Hopper’s drinking, according to Berkeley, was the result of a variety of factors, many of them shadow sides of her most admirable traits. For instance, Hopper’s ability to persevere in the face of obstacles, be they intellectual, organizational, or cultural, was a characteristic that Berkeley and others deeply admired. Her career thus far had been punctuated by personal choices that demonstrated this trait: the first woman to graduate with a doctorate in mathematics from Yale, one of the first women to become an officer in the Navy, the third programmer of the first modern computer, and now a senior manager at the first commercial computer start-up company. Even her divorce could be seen as a break from the norm, given the cultural climate of the 1940s. According to Berkeley, it was the intensity of Hopper’s strengths, such as her perseverance, that paradoxically fueled her addiction:
What are the sources of these dammed-up emotions, worry prohibitions, and proud self-imposed laws? These ways of behaving are chunks of immaturity, of infantile habits, baby ways of dealing with grown-up situations. The biggest chunk of all the immaturities I think is your determination to be your own boss, decide for yourself, and to hell with doing anything you don’t choose to do. Oh, I don’t mean that you are consistently head-strong; of course you often do things you do not choose to do, but you do them in cases where you have not involved all your personality, all of yourself. This biggest chunk of immaturity I believe probably derives from revolt against family constraints while you were young, coupled with your innate tendency to be stubborn.66