by Kurt W Beyer
Berkeley truly believed that Hopper did not intend to become an alcoholic, but that her potent intellect worked against her and kept her from seeing her problem for what it was: a disease that was slowly destroying her ability to work and to interact with family and friends. “Your alcoholic habit however is of such long standing that it has warped most of the intellectual processes that you would ordinarily use to attack your alcoholism,” he wrote. “Your fierce passion to reject authority has enlisted very many of your intellectual capacities to defeat all intellectual approaches.”67
Berkeley strongly condemned Hopper’s growing tendency to use her drinking as a means of manipulating friends and family for attention. After one of her “episodes,” Hopper was in such a desperate physical and psychological state that she typically had to persuade friends to stay with her until she recovered. Since most of her friends were also hard-working people, her drinking problem began to interfere with their productivity. What had started as a personal problem, according to Berkeley, had become a community issue, and one that threatened the very thing that Hopper craved most: love and attention from those around her.68
More disturbing, according to Berkeley, were Hopper’s recent threats of suicide, which he believed were cries for help and attention:
Committing suicide cannot solve the problem you have taken into your feelings, of more financing for EMCC; but several hours of sober, resolute intellectual effort might result in your giving some very useful suggestions to John [Mauchly]. To fill up one’s head with vague pleasant ideas about committing suicide and other people being sorry is also a lot easier, more childish, and more “ego-stroking,” than concentrated thinking right on the point of an objective problem.69
Berkeley truly believed that Hopper could cure herself. She needed to enlist her powerful mind to overcome the addiction that gripped her more with each passing day. She needed to find a different way to deal with her emotions and her worries, openly communicate with her friends and family, and maybe even find a less stressful job with more stability. “The EMCC strain is building up greatly. I think it is wiser for you to avoid that strain,” wrote Berkeley. “One definite and good possibility for you is another job soon in some other company, where there will be less strain.”70
Berkeley’s motivations for helping Hopper to defeat her addiction went beyond friendship. A pioneer in the computer industry himself, Berkeley recognized Hopper’s talent:
I and many other people know full well what a wonderful intellectual and emotional endowment you have. Even when you function properly only 70 percent of the time, you are worth to Eckert-Mauchly Computer Corporation all of the $6500 salary they pay you. . . . I can see in my mind’s eye the marvelous things you could accomplish with the 30 percent of the rest of your time (now wasted), such as writing, teaching, living, and any other of the other things at which you are so competent.71
Grace Murray Hopper had played a pivotal role in the early years of the computer industry. Edmund Berkeley sensed rightly, however, that her most important contributions to the field were yet to come.
A POTENTIAL SAVIOR: THE REMINGTON RAND CORPORATION
As winter set in, the future looked bleak. While Hopper and other EMCC employees fought their own personal demons, the company as a whole headed steadily toward bankruptcy. With few options, Presper Eckert and John Mauchly accepted a less than generous offer from James Rand, the aging president of the Remington Rand Corporation. At the time of the purchase, in February 1950, Remington Rand was an established leader in the office equipment business, selling everything from file cabinets to punch-card calculating machines. Rand reimbursed American Totalisator $70,000 less than it had paid for its 40 percent share in 1948, even though EMCC had already produced the BINAC, was in the process of constructing the UNIVAC, and had six pending orders. Eckert and Mauchly received another $100,000, which was promptly disbursed to employees and suppliers for salary and back payments. Rand also agreed to pay EMCC 59 percent of monies received from its patents, with a minimum annual guarantee of $5,000.72
Unlike Henry Straus, the aging James Rand had little interest in computers. Though an inventor himself ,73 he had only a limited knowledge of electronics. Rand’s special assistant, Arthur Draper, had persuaded him to purchase the struggling Eckert-Mauchly Computer Corporation, but Rand viewed the acquisition as a niche product line to add to his company’s diverse product catalogue. Lou Wilson, an EMCC engineer, recalled Rand’s first visit to the laboratory. After being briefed on the technical aspects of UNIVAC and the potential of computers, Rand responded “That’s very interesting, but why is that an IBM typewriter?” Rand was pointing at the proposed UNITYPER, the output source for the UNIVAC. The UNITYPER was a reconfigured IBM typewriter, and Wilson’s new boss wanted to know why he was not using a Remington Rand. “Take that label off that machine! I don’t want it seen here!” said Rand on his way out.74
In the reorganization that followed, the Eckert-Mauchly Computer Corporation became a subsidiary of Remington Rand and was listed as a separate division. The new EMCC division was far from autonomous, however, and Remington Rand’s management was grafted onto the organization. “We were just peons,” Betty Snyder Holberton recalled. “When [Remington Rand] came in, they set up this hierarchy, and I finally had a boss that I worked for. That was a disaster too, because I’d go to him with decisions and he’d make the wrong decisions and I had to live with them.”75
Part of the problem, according to Hopper, was that the majority of the Remington Rand management had little understanding of computing in general and programming in particular. Hopper and Snyder found themselves defending the technological choices made by EMCC in the face of management schooled in mechanical punch-card machines. Typical arguments that they had to deflect were “We don’t think those tapes are ever going to work” and “We didn’t put tapes on our machines; the card is the thing.” Sadly, according to Hopper, Remington Rand management was looking at information processing from a different mental paradigm. Theirs was a world of specialized mechanical machines dedicated to a specific processing task: one machine for accounting, another machine for inventory, a third machine for billing. All the input and output used by these machines could be found on paper punch cards. Punch-card information was “real” because it could be seen and verified with one’s own two eyes. According to the experts, there was no place for bits, Boolean algebra, logic gates, internal memory, subroutines, and machine code in the office machine marketplace.76
A difference of opinion concerning UNIVAC’s potential also existed between Eckert and Mauchly and their new boss, General Leslie Groves. Mauchly in particular had a vivid vision of a commercially viable general-purpose machine that would help transform science and business. Groves, the former head of the Manhattan Project, saw an unreliable, unproven, and costly piece of equipment, which was now his headache as Remington Rand’s director of advanced research. As the man in charge of overall operations of the Eckert-Mauchly division, Groves had the final say on strategic business decisions and on funding for research and development.
For Groves, the first thing to do was clean up the financial mess that Eckert and Mauchly created. The novice businessmen had gravely underestimated the development costs of UNIVAC, which now approached $1 million per unit, and Groves sent Remington Rand lawyers to renegotiate the UNIVAC contracts. Even though the three government agreements for $300,000 each could not be changed, Remington Rand successfully broke contracts with Prudential Insurance and A. C. Nielsen Inc., who expected to purchase UNIVACs at the implausible price of $150,000 each.77
Next, Mauchly was transferred to the sales department when a Remington Rand security check revealed that the co-founder of EMCC did not have proper security clearance. According to Eckert, not only had Mauchly attended Communist meetings during the 1930s; there was evidence that his secretary had Communist affiliations as well. “[The secretary] was sleeping with a guy who was a Communist,” Eckert recalled. “As a matter
of fact, he would take her to Communist meetings. Therefore they decided maybe she’s a Communist, and maybe John’s a Communist.”78
Not only did EMCC’s new bosses look unkindly on potential Communists; it also appeared that Remington Rand was no place for women. “Women, as far as I could see, had absolutely no future under Remington Rand, absolutely none,” Betty Snyder Holberton recalled. EMCC’s programming expert remembered how poorly Remington Rand salesmen and management had treated her because her salary rivaled their own: “You felt it. There was some resentment against a woman because you had essentially moved up all the way from the bottom.”79 Grace Hopper too sensed the tide turning against women after the takeover: “There were not the same opportunities for women in larger corporations like Remington Rand. They were older companies, and the jobs had been stereotyped.” When Hopper had joined Eckert and Mauchly, the sky was the limit because the field was new, the company was small and flexible, and there were no gender-based roles. “Eckert and Mauchly were singularly unprejudiced, but also they were trying to gather together a team to build that first computer, which no one believed in,” she said.80 Betty Snyder Holberton also recalled a tolerant, flexible company with few traditions or defined job classifications. “Some days,” she said, “you would be programming, sometimes you would be doing logic, sometimes you would go out with Mauchly selling. In fact, I can’t remember anybody really telling me what to do. I just did what I felt had to be done and then went to somebody with what I had done.”81
But by the summer of 1950, Snyder was so disillusioned with her new circumstances that she contemplated leaving Remington Rand and computers for good. She felt that her new masters were milking her for information, and that eventually she would be kicked to the curbside, especially since she did not have a doctorate. That summer, she decided to marry John Holberton, the man who in 1944 had interviewed her for the ENIAC programming position. After an intimate ceremony with friends and family, the couple went on an extended honeymoon to England, where they visited Maurice Wilkes at Cambridge and inspected the EDSAC. Upon their return, Betty Snyder Holberton decided to leave Remington Rand rather than be transferred to New York. She eventually accepted a job at the Navy’s David Taylor Model Basin in Annapolis. (Her husband worked in Washington.)82
Although Betty Snyder Holberton continued to work with computers at the David Taylor Model Basin, her brilliant career at the first computer start-up company had come to an unexpected end. During her time at EMCC, she had worked with John Mauchly to develop C-10, the object code for the world’s first commercial computer, and her pioneering work with sort generators would have important ramifications throughout the 1950s. Despite her accomplishments, the weight of Remington Rand’s misogynistic culture, combined with her newly perceived obligations as a wife, tempered her enthusiasm for computing.
Betty Snyder Holberton’s departure and John Mauchly’s political struggles put the new company’s programming burdens on Grace Hopper’s shoulders. Hopper, however, had only recently recovered from her struggle with alcohol addiction. The unexpected death of Henry Straus and the resultant near bankruptcy of the Eckert-Mauchly Computer Corporation had thrown Hopper into a deep depression that nearly ended her career. But sometimes a person’s darkest moments in life serve as the catalyst for change and establish a foundation for future success. Paradoxically, the takeover of EMCC by Remington Rand would mark the beginning of the most productive and creative chapter of Hopper’s career.
8 THE EDUCATION OF A COMPUTER
One of the more difficult concepts for historians of technology to explain is that of invention. From patents and other written evidence, it may be simple to determine when and where a certain technology came into being. What remains difficult for the historian is to determine how and why a certain technology was created. Answering how requires insight into the process of invention. If we believe Thomas Edison’s maxim that invention is 99 percent perspiration and 1 percent inspiration, then it follows that invention is not a completely serendipitous event. The inventor patiently prepares for inspiration to strike in a variety of ways. These include being well versed in the field of inquiry, having access to the essential equipment, and acquiring skill in empirical learning through long hours of trial and error.
Even more difficult for the historian is determining the moment and source of inspiration. Often the historian is left with just faint reflections of the moment as it is recorded in memoirs and interviews after the event. The accuracy of such reflections may be questionable, in view of the passage of time between the event and its recording. Moreover, errors can be magnified by the mind’s ability to organize the past into a coherent progression of events.
Tied to the question how is another question: Why was a certain technology created by a particular person at a given time and place? Understanding the motivations for why the inventor added her perspiration to the process of invention goes beyond altruistic notions of improving one’s society. Personal impulses are conjoined with a variety of external forces generated by a multitude of actors, be they individuals, organizations, or social groups. Since these groups often provide the knowledge, the funding, and the demand that propel the process of invention forward, it is necessary to understand the underlying milieu in which the inventor works.1
THE INVENTION OF THE COMPILER
It is generally accepted by historians of technology that Grace Hopper wrote the first compiler during the winter of 1951–52. According to interviews and speeches she gave years later, she began working in her spare time on the programming invention in October 1951. Hopper’s recollections point to motivations ranging from an altruistic desire to allow “plain, ordinary people” to program to dealing with her own laziness. Naturally one must be skeptical of such claims, for they were made years after the fact. In 1951 it was difficult for even a visionary like Hopper to imagine the eventual ubiquity of computer technology, and one can be pretty confident that Hopper was not a lazy person.2
Likewise, the inspiration leading to Hopper’s compiler design seems to be blurred by the passage of time. On occasion Hopper has credited Howard Aiken’s Mark III coding machine as the source of her inspiration; at other times, Betty Snyder’s sort generator has been mentioned as providing the inventive spark. In one speech, Hopper went so far as to credit her experience playing basketball in college as the inspiration for a program that “passed” subroutines back and forth.3 To get a better sense of the compiler’s roots, let us first take a look at conditions at the Eckert-Mauchly Computer Corporation just after the merger with Remington Rand in 1950. We will then turn to the articles and papers Hopper wrote on compilers during this same period.
“Captain Marvel and the Incredible Calculator.” Courtesy of Archives Center, National Museum of American History, Smithsonian Institution.
REMINGTON RAND AND THE CHALLENGE OF SALES AND CUSTOMER SUPPORT
As was discussed in the previous chapter, the purchase of EMCC by Remington Rand brought with it a multitude of organizational challenges. EMCC, a small computer start-up focused on innovation, did not mesh well with the culture of a large, corporate organization concerned primarily with quarterly revenue targets. To get a sense of the conditions at EMCC, let us turn to Hopper’s own assessment and to the views of more than 30 computing pioneers associated with UNIVAC computers during the 1950s.4 From their collective memories emerges a narrative that highlights Remington Rand’s difficulties with constructing a sales and customer support structure for UNIVAC during the first half of the 1950s. These difficulties weighed heavily on Hopper and her staff, and generated the conditions that propelled Hopper to search for a radical technological solution to ease EMCC’s mounting programming crisis.
From the first days of the merger, Remington Rand senior management was reluctant to invest heavily in what they viewed as an unproven technology. Instead of embracing a strategy to increase UNIVAC sales in order to bring down unit costs, the director of advanced research, General Leslie Gro
ves, decided to cut EMCC’s costs and to reduce the production quota from twelve units per year to six.
With such low sales expectations, there was little incentive to educate Remington Rand’s sizable sales force about the new technology. According to Cecil Shuler, at the time a Remington Rand tabulating machine salesman in Nashville, the company did not establish a training program for tabulating machine salesmen who were interested in learning about the UNIVAC.5 Moreover, Remington Rand failed to define a commission policy for sales of UNIVACs, thus depriving salesmen of a financial incentive to sell the computers. Hopper’s former Harvard colleague Fred Miller remembered the difficulty he had inquiring about purchasing a UNIVAC for the Naval Proving Ground: “You had to find a salesman and buy him a drink before he’d talk to you.”6
Eventually, in the summer of 1950, Remington Rand’s senior staff established a UNIVAC sales training program. David Savage, referred to as “the most brilliant person in the Remington Rand Sales organization” and “comparable to Professor Eckert” in the company’s internal newspaper, the Remington Rand News, was placed in charge of training the sales staff and preparing UNIVAC marketing materials. To the amazement of EMCC personnel, Savage developed the programming and applications course material and marketing handouts during the fall of 1950 without ever contacting anyone familiar with the UNIVAC, and when Hopper and others offered assistance, they were viewed as “meddlesome.” In an angry memo to senior Rand executives, John Mauchly wrote: “It is not to [Savage’s] discredit that he has had no previous experience in programming automatic computers, but we are at a loss to understand how anyone, having accepted a position demanding experience which he does not have, would fail to seek all possible help in preparing himself to discharge his duties.”7
