I first learned about ERMA while researching the future of cashless societies. ERMA (Electronic Recording Machine, Accounting) is a pioneering technology that automated bank bookkeeping and check processing. The nonprofit research institution SRI developed ERMA between 1950 and 1955, when the discipline of computer science was just emerging. Today, payments experts say that ERMA established the foundation for computerized banking, magnetic ink character recognition (MICR), and credit-card processing. In reading about ERMA’s principal engineers on SRI’s website, I wondered who could work together on such an extensive and complex project—and how. After all, the developers created a truly pioneering computer: Automated banking had never existed before this machine, and it fundamentally impacted industries for decades to come.
On an afternoon in July, I called Jack Goldberg, a logic designer on ERMA, at his home in Palo Alto. I’d read that Goldberg joined SRI in 1951, when the Menlo Park nonprofit organization was still known as the Stanford Research Institute. I knew that Goldberg later managed SRI’s Computer Sciences Laboratory, which today is one of the most respected institutions of its kind in the world. So I assumed that when he and his team built ERMA’s logic design—i.e., the organization of its circuitry—they were already experts in computer science. But Goldberg and his colleagues were figuring it out as they went along. “We were just learning about how to use computers ourselves,” he told me.
The “whiz kids” in their early careers
Goldberg was one of ERMA’s “whiz kids,” a nickname given to the project engineers by SRI’s clients at Bank of America (BofA). I was connected to him by Donald Nielson, a former SRI vice president and author of A Heritage of Innovation: SRI’s First Half Century. (He joined SRI in 1959, after the ERMA project had already been completed.) Nielson’s book explains that BofA became interested in electronic banking applications to prepare for a looming paper-handling crisis. The post–World War II commercial boom in the United States resulted in a staggering increase in the volume of bank checks deposited. The manual process of bookkeeping and check-clearing was so labor-intensive that the banks’ operational capacities were on the brink of being overwhelmed.
To learn more about ERMA’s engineers, I called James D. Skee, a history of technology researcher who has written about SRI’s commercial consulting. Skee pointed out that at the time, SRI was a relatively new research institution without an endowment.
“The way the founders got around [the lack of funding] was through obtaining [private] contract project work, of which ERMA is an example,” Skee said. He noted that in its early days, SRI had difficulty finding staff, but ERMA—as a large, funded contract—helped the organization attract talent. “What you saw was a lot of people from the [Stanford] MBA school going over to work for SRI once they graduated, or people who had some kind of aviation-industry background coming out of the war,” Skee said.
I imagined SRI as a fledgling institution with a team of young engineers who may have been eager to prove themselves in their careers. Skee said that ERMA was a “big win” for SRI. I wondered if that meant that ERMA’s project leads felt like they didn’t have the option to fail.
According to the Silicon Valley Historical Association, SRI was founded in 1946 by Stanford University and a group of business executives, as a way to support scientific innovation and economic development on the West Coast after World War II. On SRI’s website, I found a photograph of ERMA’s whiz kids, along with a list of their names. As I looked into their biographies, it was evident, by and large, that their careers were just beginning.
I learned about their lives from academic papers, professional biographies, and published oral histories. Some of ERMA’s project engineers were still PhD candidates at Stanford. Others had worked as junior scientists in research laboratories that advanced U.S. war efforts during World War II, including the Radio Research Laboratory (RRL)—a secret, 800-person research laboratory located at Harvard University, which contributed to the development of radar—as well as the Manhattan Project, a research and development project that produced the first nuclear weapons.
Alfred E. Kaehler was a mechanical engineer who worked on the Manhattan Project before joining the ERMA team at SRI. In a 2016 oral history for the Stanford Historical Society, he described how SRI’s facilities were housed in a former Army hospital. The work culture was similar to a university, where everyone was free to state their opinions and write research papers. The culture at SRI differed from other, old-fashioned companies that he had worked for, where there was a division head and “everybody knew their place.” The stories that Kaehler heard from his cohort at SRI included their contributions toward the Allies’ victory on D-Day. Many of them had worked on the radar technology that helped hide Allied ships from German radar. At SRI, Kaehler also met Doug Engelbart, the inventor of the computer mouse, whom he described as casual and good-natured.
Thomas H. Morrin was the head of SRI’s Engineering Group, and, according to the organization, he played a key role in managing the success of the ERMA project. Morrin had worked as a Naval engineer during World War II, and had served in the Office of Naval Research, where he oversaw the work of the RRL. After Morrin joined SRI in 1948, he recruited several of his former colleagues to help with ERMA’s development, including technical team lead Jerre Noe. During World War II, Noe had conducted radar research in Europe. It seems that SRI’s project leaders assembled ERMA’s talent base from the engineers they knew, and built upon the experience the team members had acquired from backgrounds in military research.
Over email, Nielson wrote that ERMA’s “leadership form was clearly hierarchical . . . but the style had to provide the patience for innovation since the world was in great technological flux.”
He also outlined the size and complexity of the ERMA project, which required a broad variety of talent and a multitude of teams. He wrote, “Keep in mind the disciplines involved, all the way from magnetic ink development and the ability to read it reliably, to computing logic (the basis for which was embryonic), to the mechanical design of a sorter that could handle nonstandard check sizes, to magnetic drum storage, to the grand design of the world’s first real time–processing business computer.”
In other words, ERMA required the work of a series of teams to meet its demands for redevelopment and redesign. It was one of the first computing projects to charter the iterative development process that is codified among technology teams today.
Despite the formidable backgrounds of many of ERMA’s engineers—and the presence of women in that photo of the whiz kids—I noticed that none of the women who worked on ERMA were identified by name on SRI’s website. I got in touch with Troy Eller English, the archivist at the Society of Women Engineers (SWE), who searched its directory to see if one of their members had worked at SRI at that time. While no SWE members surfaced, English wrote in an email that SRI’s picture of the ERMA team included “an unsatisfying list of contributors that appears to omit the names of all the women in the photograph. I suspect the women were the . . . invisible labor behind the design project.”
When I asked Nielson about the photograph, he wrote that there were some women on the technical staff who helped implement the design, working on ERMA’s wiring-defined components. “Mostly the role for women would have been in administrative help,” Nielson wrote. “Computer programming, probably the first substantial role for women in computing at SRI and elsewhere, soon appeared as machines gained that capability. But ERMA was not a stored program machine.”
The process of redevelopment and redesign
On the phone, Goldberg spoke vividly about his memories of ERMA. “We were just aware of new computer technology at the time,” he said. “We had an initial survey of the bank’s needs, and how computers might be used. And we were excited about this because we wanted to learn about using computers.”
During the years it took the engineers to create ERMA’s concept and build its prototype, advances in computer technology progressed at a rate that changed ERMA’s functionality as it was being designed. Goldberg said that by the time the prototype was completed in 1955, the developers “didn’t use the original technology that started the project.”
To understand ERMA’s development chronologically, I looked up a paper from 1993, “The Development of the ERMA Banking System: Lessons from History” by Amy Weaver Fisher and James L. McKenney, published by the Institute of Electrical and Electronic Engineers (IEEE). It states that SRI completed ERMA’s first, exploratory study with a feasibility report in September 1950, which named the machine ERM (Electronic Recording Machine). In the feasibility study, the engineers identified a problem: Checks were filed via an alphabetical system, which meant that they’d be reshuffled any time a new one was added. The engineers convinced the bankers to transition checks to a numerical filing system, which eventually paved the way for future project innovations, like the development of magnetic ink character recognition—the curlicue numbers and symbols that appear in different configurations at the bottom of bank checks.
SRI’s second study for ERM extended from November 1950 to April 1951. The engineers focused on learning more about banking procedures and on developing ERM’s logic design. Nielson’s book reveals that SRI did not plan to actually build the machine themselves. Although building the machine’s prototype was ordinarily beyond the scope of a research institution, BofA failed to find any industrial manufacturers interested in producing the machine. Morrin was reluctant to have SRI’s engineers take on the task of construction because they were a staff primarily oriented toward research. Nevertheless, in January 1952, SRI signed a contract to construct the prototype.
Goldberg said that it took several years to plan and build a working model of a machine that could store check information, make appropriate calculations, keep records, and respond to user demands. As team members developed the prototype, they worked in close proximity to a bank representative.
“Mr. Charles Conroy resided with us, teaching us about banking and guiding us with our planning,” Goldberg recalled. Nielson also suggested that the ERMA team was uniquely galvanized by the banker’s participation. “It was essential that the bank had representatives that almost lived in [SRI], so that they could experience the challenges and assess the quality and pace of work,” he wrote in an email.
I read more about Goldberg’s work in the IEEE paper, and learned that he, along with Bart Cox and William Kautz, built the prototype’s electronic logic with vacuum tubes as stand-ins for transistors. (Transistors were newly available, but the team considered them inconsistent in quality due to their recent introduction to the market.)
In learning about Goldberg’s experience, I became curious to hear more about SRI’s style of leadership, which may have allowed ERMA’s engineers to experiment and find creative solutions. Nielson shared that the leaders were primarily concerned with keeping the immense project on track within the bank’s time constraints: “Leadership, particularly of the subsystems, had to define cutoff points both for the kind of technology available and for performance adequacy. This process would have been fraught with compromise and doubtless no designer wanted to quit innovating as new technology surfaced.”
The IEEE paper also emphasizes the intense time pressure on the project engineers. The bankers were anxious to announce ERM’s completion, and they grew restless with the engineers’ iterative development process— they saw it as continual tweaks to a product that was already functional. In the spring of 1955, the bankers announced that ERM’s design was complete “as is,” and planned to debut the prototype that September. The engineers worked shifts at all hours to build the first ERM through carefully drafted plans—and haphazard engineering. The name ERM was also changed. The bankers wanted the machine to have a more familiar and approachable name—thus ERM was renamed ERMA.
When BofA presented ERMA to the banking community, press, and electronics manufacturers that might bid for its production, the computer weighed 25 tons, contained more than a million feet of wiring and 8,000 vacuum tubes, and required cooling by an air-conditioning system. Through strategic planning by BofA’s public relations office, journalists from leading national publications attended the first public demonstration. Clark Beise, BofA’s president at the time, called ERMA “the biggest single advance in bank-account bookkeeping in the history of banking,” and the press highlighted the achievement in similar fashion. ERMA was a success.
From pioneer to institution
ERMA was the first major computer research program at SRI. General Electric won the bid to manufacture the product version of ERMA in 1956. ERMA was one of GE’s first projects in computer manufacturing. Today, both SRI and GE are multinational companies and leaders in scientific research and development. In a phone conversation, Goldberg recounted how GE redesigned ERMA almost in its entirety with newer technologies. A computer scientist named Joseph Weizenbaum was the programming lead at GE, and under his supervision ERMA shifted from a hardware project to more of a software project. (After ERMA, Weizenbaum became a professor at MIT and developed ELIZA, an early computer program that performed natural language processing.)
Goldberg said that, back at SRI, his team had to “look around to see what interesting things using computers might [we] develop, might we learn about.” Along with several other SRI engineers, he worked on a contract project for NASA to design computers that could tolerate failures, systems reliable enough to be used for space missions—including trips to the moon.
ERMA’s automation was so successful that it established BofA as a leader in the banking industry. BofA built upon ERMA’s technology to offer a general-use credit card in 1958, BankAmericard, which was licensed to other banks around the nation in 1966 and renamed Visa in 1976. The last ERMA computer was replaced in 1970 by newer equipment, but the impact from both the project and the efforts behind it lives on.
I was surprised to read that, in a 1985 interview with MIT’s newspaper, The Tech, Weizenbaum expressed regret over how computers fortified the big-banking system. He said that they revolutionized banking in only a superficial way:
If the computer had not been invented, what would the banks have had to do? They might have had to decentralize, or they might have had to regionalize in some way. In other words, it might have been necessary to introduce a social invention, as opposed to the technical invention.
Weizenbaum said that he was not concerned by the impact of automation when he was working on ERMA, due to his focus on the project’s difficult technical challenges. He described the work as fun: “In the act of programming you discover new ideas,” he said, “and most particularly you discover that there are deep holes in your knowledge that you have to fill before you go on.” It was over a decade later, when he began writing down his thoughts on technology’s impact on society, that he began to reflect on the repercussions.
This pride in a feat of engineering, coupled with the philosophical discomfort around the societal repercussions of an enduring technical legacy, made me wonder how working in teams can keep us accountable. As creators, we can iterate endlessly as new methods and new ideas present themselves through our work. But having a sense of accountability toward others—whether that’s reflected in meeting a deadline, sharing our accomplishments, or considering the way our work may ripple out into the future—may be what pushes us to reach standards that we couldn’t imagine achieving on our own. Now, when I follow the development of new, foundational technologies, that’s the part of ERMA’s legacy that I remember.