by Kurt W Beyer
After some deliberation, Hopper narrowed down her choices to two. Option one reunited Hopper with her former Yale mathematics professor Dr. Howard Engstrom. Engstrom and William Norris had founded Engineering Research Associates (ERA) with financial backing from the Navy after the war. During the war both Engstrom and Norris had been commanders specializing in cryptography at the Navy’s secretive Communications Annex. Engstrom and Norris transferred their development teams to St. Paul, Minnesota after the war and founded ERA with the intent of designing computing machines for cryptological purposes. The company had substantial expertise in magnetic drum memory and in 1947 began to construct a digital computer called ATLAS for the Navy.78 The second option was also a private venture. The Eckert-Mauchly Computer Corporation was a computer start-up based in Philadelphia. Hopper first met its founders during her visit to see the ENIAC at the Moore School of Electrical Engineering at the University of Pennsylvania in 1945. After a dispute concerning patent rights with the University, the two ambitious inventors struck out on their own in the spring of 1946. By the time of the Oak Ridge Conference, the Eckert-Mauchly Computer Corporation had signed contracts with the Northrop Aviation to construct an electronic digital computer, BINAC, as well as contracts with the Census Bureau, National Bureau of Standards, and the Army Map Service to construct a general-purpose electronic digital computer, UNIVAC.
What made both companies attractive was the fact that they were to have working machines available within the year. Hopper thrived in the Harvard Computation Laboratory’s operational environment, and she hoped to find this again at the two start-up companies. In fact, Hopper had applied for a position at IBM, but walked out of the interview after experiencing what she perceived as the suffocating environment of Thomas Watson Sr.’s oversized corporation. “I couldn’t take it,” Hopper recalled, “because that was back in the days when they still had flags, an IBM flag, and sang songs about it.”79 In the end, Hopper chose the Eckert-Mauchly Computer Corporation. “The thing that finally tipped my decision,” Hopper recalled, “was the fact that they had BINAC running and UNIVAC I was well underway and would be running within a year whereas it looked as though the work out at St. Paul was going to be quite a bit longer. Plus the fact that John Mauchly looked like a wonderful person to work for, and I would be working for him.”80
Hopper’s time at Harvard was instrumental for her development as a programmer, a manager, and a leader within the emergent computing community. The pace of the war years catalyzed her transition from a college professor to that of a computer programmer, a term that would not be applied until 1949. During that time she helped to define what programmers were, what they did, and how they did it. Along with Richard Bloch she developed a methodical system of coding and batch processing that turned the experimental Mark I from a mechanical curiosity into a useful mechanical tool. This achievement was far from trivial, for it proved that computation could be automated on a large scale.
After the war, Hopper was responsible for recording the Computation Laboratory’s tacit organizational knowledge so that it could be widely disseminated. She authored the Mark I Manual of Operation and later the Mark II manual that documented the laboratory’s technical accomplishments and placed the Harvard achievement within historical context. Her three articles concerning hardware and programming development at Harvard were widely distributed to those interested in the new field. As interest in computers grew through the 1940s, Hopper gave seminars and escorted hundreds of visitors through the Computation Laboratory. Furthermore, her close association with Howard Aiken, coupled with her growing expertise, afforded Hopper continued access to the elite of the field at a time when many women were retiring from public life. But despite her successes, Grace Hopper’s career in computers had only just begun.
7 STARING INTO THE ABYSS
On a cold night in November 1949, only 6 months after leaving Harvard and joining the Eckert-Mauchly Computer Corporation, Grace Murray Hopper found herself behind bars at the central Philadelphia police station. The programming pioneer was arrested at 3 a.m. for drunk and disorderly conduct. She was eventually placed in the custody of Pennsylvania General Hospital for treatment. Hopper’s life was unraveling. At the age of 43 she had accomplished much, yet her growing dependency on alcohol was jeopardizing her career and her relationships. As winter approached, Hopper even contemplated suicide by drowning herself in the Schuylkill River, something she had supposedly tried unsuccessfully in Boston’s Charles River weeks before.1
Eight years had passed since Grace Hopper chose to reinvent herself. During those years she left her husband, quit her secure teaching position at Vassar College, joined the Navy, and aligned her future with the embryonic computer industry. The long hours and late nights she endured at the Harvard Computation Laboratory transformed the former mathematics professor into the world’s premier computer programmer. She helped to prove both the viability and the utility of the new computing technology during the war, and she wrote and lectured about the Harvard Computation Laboratory’s achievements after the war. Hopper’s accomplishments during her eight-year transformation, however, came at a considerable personal cost. Pioneers such as Hopper are faced with far more than technical conundrums. They must deal with a variety of social and psychological pressures associated with the very act of exploring uncharted intellectual waters. In the process of creating that which does not exist, the technical pioneer must manage not only his or her own doubts but also the doubts of colleagues, investors, managers, end users, and a skeptical public. Being an inventor is in many respects an act of faith: faith in one’s own technical abilities, faith in those who work alongside, faith in the ultimate vision and purpose of the project.
The pressures associated with inventing the computer age at the Harvard Computational Laboratory were exacerbated by the special circumstances associated with the wartime environment. The war weighed heavily on Hopper and the Harvard crew, for they truly believed that their work was instrumental to the successful conclusion of the conflict. Despite the experimental nature of the Harvard machines, deadlines had to be met at an accelerated pace. On a more personal level, the war interrupted the peacetime rhythms of life: travel restrictions, food rationing, and material shortages became the norm.
But even after August 1945 and the country’s gradual return to normalcy, the Computation Laboratory’s overbearing director, Howard Aiken, attempted to maintain the sense of urgency associated with the war. It was during this period that Hopper’s drinking problem worsened. No longer was drinking a social diversion; now Hopper found herself drinking during the week and even on the job. Flasks were hidden in closets and desks. Hopper’s drinking binges and periods of recovery eventually amounted to two days per week wherein she could not function, thus undermining her productivity and placing an ever growing strain on friends and family.2
According to her friend and former Harvard colleague Edmund Berkeley, Hopper turned to alcohol during this period as a way to deal with the compounding pressures at the Harvard Computation Laboratory. She had dedicated herself fully to the overwhelming task of bringing Howard Aiken’s machines to life. She used the machines to solve critical military problems, including one that resulted in an explosion over Nagasaki. As the psychological strains became increasingly pronounced, alcohol seemed to serve as an effective outlet, freeing Hopper to express emotions and to temporarily forget obstacles real and imagined. According to Berkeley, the expiration of Hopper’s Harvard research contract was the best thing that could have happened to her, although in the short term unemployment added to the stress. During the last week of May 1949, the 43-year-old programmer packed up her belongings, headed to Philadelphia, and bet her future on two younger men who believed they could create the first commercial computer company.3
THE ECKERT-MAUCHLY COMPUTER CORPORATION
Leaving the caustic environment of the Harvard Computation Laboratory may have made sense to Edmund Berkeley in light of Grace Hopper’s gro
wing dependency on alcohol, but the environment of the Eckert-Mauchly Computer Corporation presented challenges too. The world of the start-up company is not for the weak of heart. Lacking the organizational constancy and financial stability of established firms, start-ups can only guarantee employees grueling hours, scant benefits, marginal pay, and a volatile future. As a rule they are the domain of the young and restless, for the vast majority fail within the first year of inception.
Even so, as of 1 June 1949 the future looked bright for Eckert and Mauchly’s venture. Hopper’s first days on the job were actually quite pleasant, and Berkeley noted the positive change in his friend’s mental condition during her first 5 months in Philadelphia. Hopper had been ardently recruited by the company’s leadership, and her reputation as a programmer, mathematician, and manager preceded her. Besides being one of the most experienced programmers in the new field, Hopper came from an operational environment. She had been instrumental in turning the Harvard Computation Laboratory into a highly efficient data-processing factory—a feat that Presper Eckert and John Mauchly had never matched with the ENIAC, given their creation’s temperamental vacuum-tube technology and programming limitations.4
In contrast to her initial experience at Harvard, Hopper found EMCC to be very accepting of women. In fact, the majority of the programming staff as of June 1949 consisted of women. During the ENIAC project, Eckert and Mauchly arranged to have six women transferred from the Ballistic Research Laboratory (BRL) at Aberdeen to the Moore School of Electrical Engineering at the University of Pennsylvania to serve as ENIAC operators. The six had been members of a staff of about 200 mathematically inclined women who computed firing tables for artillery and rockets with desktop calculators. Three of the original six, Francis Elizabeth “Betty” Snyder (later Holberton), Betty “Jean” Jennings (later Bartik), and John Mauchly’s future wife, Kathleen “Kay” McNulty, remained with Eckert and Mauchly after the two left the Moore School to form their company.5
The same charm and charisma that persuaded Snyder, Jennings, and McNulty to continue working with John Mauchly caught Hopper’s attention when she chose EMCC. “Mauchly was one of the grandest people I have ever met,” she recalled.
John Mauchly. Courtesy of John W. Mauchly Papers, Rare Book and Manuscript Laboratory, University of Pennsylvania.
“He was very broadminded, very gentle, very alive, very interested, very forward looking.”6 The man who was about to become her new immediate boss was quite different from Commander Howard Aiken. Where Aiken was overbearing, Mauchly was accommodating. Where Aiken demanded, Mauchly convinced.
The sharpest distinction between the two computing pioneers, however, was between their visions of the future. While Aiken saw the computer primarily as a scientific instrument, Mauchly’s field of view was far more expansive. “He was looking way ahead,” said Hopper. “Even though he was a college professor he was visualizing the use of these computers in the business and industrial area.” In fact, an inquiring customer could bring up almost any field in business and Mauchly could ad-lib detailed application descriptions, ranging from marketing research to industrial process control. “He saw much more of the future of the computer than just as a scientific tool or for the solution of mathematical problems,” Hopper recalled. “I think that’s what was exciting.”7
Hopper marveled at the extraordinary relationship between Mauchly and his youthful partner, Eckert. The two had met in 1941 at the University of Pennsylvania. At the time, Eckert was a talented graduate student in the Moore School of Electrical Engineering and Mauchly was teaching a war-related electronics course. Two years before they met, Mauchly had experimented with the notion of an electronic calculating machine, partly because of the intense calculating work he was involved with for the Army Signal Corps concerning radiation patterns for antennas. The inventive physicist shared his ideas with the young engineer and even demonstrated a simple mechanical counter that he had constructed with gas tubes.
Eckert was drawn to the man and his computing ideas, and the two began a dialogue concerning the feasibility of reckoning at the speed of electricity. “Those conversations and things in 1941 gave me some confidence and assurance that I wasn’t just on a wild goose chase,” Mauchly recalled.8 Eckert felt that Mauchly’s theoretical concepts were plausible, but that Mauchly’s engineering acumen was marginal. Mauchly saw in Eckert a no-nonsense practitioner who could turn his ideas into reality. Thus was formed a partnership that would construct the first all-electronic digital computer and establish the first commercial computer company.
According to Mauchly, the two did not expect to be business partners, for they were first and foremost academics. But with the appointment of Irvin Travis as the University of Pennsylvania’s Dean of Research in January 1946, the inventors of ENIAC were forced to make a fateful decision. Travis offered Mauchly and Eckert permanent positions at the university on the condition that they sign a patent release. Travis believed that faculty and staff should not be permitted to benefit financially from research generated while working for the university. Given the potential of electronic computers, both inventors refused to sign and submitted their resignations effective 31 March 1946.9
Undeterred, the ambitious partners decided to build ENIAC’s replacement privately, and in the spring of 1946 they formed the Electronic Control Company. “I think,” Eckert recalled, “John and I wanted to get the computer out there and being used by people of every type and description for every way and every purpose, come hell or high water. We didn’t care whether you called it scientific or engineering or mathematical or non-mathematical—we wanted to get this job done.”10 Their Electronic Discrete Variable Automatic Computer (EDVAC),11 later renamed the Statistical EDVAC or Universal Automatic Computer (UNIVAC), followed the theoretical design set out in John von Neumann’s June 1945 report, “First Draft of a Report on the EDVAC.”
But according to both Eckert and Mauchly, the majority of the ideas in von Neumann’s paper were reiterations of concepts that Eckert and Mauchly had fostered for years before the publication of the well-known mathematician’s report. Von Neumann, they said, had been a beneficial sounding board because he quickly grasped concepts, and his prominence within academic and government circles permitted him to effectively broadcast ideas, but he had made few if any original contributions to the theoretical design of the EDVAC. Mauchly maintained that Eckert had written his own EDVAC report (dated January 1944), but because of government security concerns he had not been permitted to publish or even distribute it. “I was not allowed to make speeches about it, and he [von Neumann] went out and made them anyway without clearance and got out of it because nobody wanted to come down with the Espionage Act on a prestigious guy,” said Eckert. “If I had done it, they would have come down on me with a ton of bricks.”12
To add insult to injury, during the early spring of 1946 John von Neumann and Herman Goldstine (Eckert and Mauchly’s trusted Army Ordnance representative to the ENIAC project, who had joined von Neumann at Princeton after the war) attempted to wholly claim the patent rights to the EDVAC. The two cited the “First Draft” as evidence of priority. Eckert and Mauchly fought their claim intensely, but in the end, on the technical grounds that one must apply for a patent within a year of publishing the results, neither party was granted a patent. Since von Neumann’s “First Draft” had been distributed widely, it was considered to have been published in the legal sense; thus, both parties had exceeded the time limit.13
Despite the controversy surrounding the theoretical concepts for the stored-program computer, the fact remained that Eckert and Mauchly were the first Americans to bring those ideas to fruition. The UNIVAC design called for vacuum-tube technology that counted thousands of times as fast as Harvard’s electro-mechanical relay machines, Mark I and Mark II. The proposed computing circuits added directly, and they replaced the subtraction operation with the addition of complements (the same numerical technique used in today’s computers). Multiplicatio
n and division were achieved by successive additions—because of the tremendous speed of the processor, there was apparently no perceived advantage to applying short-cut multiplication and division methods.14
Similar to the architecture described in von Neumann’s “First Draft,” the UNIVAC’s electronic circuitry processed information in a serial fashion, unlike the parallel design of the ENIAC. Since data and instructions passed one after another through the system’s logic gates, circuit design was simplified and programming became less complicated. The serial flow of data also made it easy to debug hardware, for all information and instructions in a given operation had to pass through the same circuits.15
The UNIVAC had an experimental internal memory that could store 1,000 twelve-bit words. The technology, known as mercury delay line memory, had first been used in wartime radar systems. Electrical data from the UNIVAC was sent into a tube of mercury, where it generated small waves that bounced indefinitely until the data was reacquired by comparing the waves with a master clock. Despite the system’s dependability, delay line memory had significant drawbacks. In addition to the overall size and cost, information could only be accessed sequentially, a loss of power wiped the memories clean, and the mercury had to be maintained at an uncomfortable 40 degrees Celsius for optimal performance.16
Electronic circuits combined with a relatively large internal memory that held both operating instructions and data would make the UNIVAC the most powerful calculating machine ever proposed. Its closest rival, the Harvard Mark II, performed 30–40 operations per second and had an internal memory of 200 words. The UNIVAC would perform more than 5,000 instructions per second with an internal memory of 1,000 words. Yet all that computational speed meant little without efficient input-output (I/O) mechanisms for entering data into internal memory. The ENIAC, for instance, was limited by the mechanical speeds of its IBM punch-card input system. Mark I and Mark II, despite their slow processors, achieved a higher I/O rate by means of paper tape.