What percentage of programmers were women? | The Computer Boys Take Over

One question I get quite often about my work is “how many programmers were there in [year X]?” As I have written about briefly on this site, and more extensively in the book, this is actually a very complicated question. We had electronic digital computers years before we had a well-established vocabulary to describe them, and now commonplace words like “software” and “programmer” had to be invented. Even then, what exactly counted as “programming” and therefore who counts as a “programmer” were never well settled, and to this day continue to be disputed (see, for example, Miriam Posner’s Javascript is for Girls article). The Bureau of Labor Statistics did not start tracking computer work until 1972, and have adjusted their categories several times since then.

A version of this question that has been coming up more frequently is “how many programmers were women?” This is an even more complicated question, because as little as we have solid data on programmer labor as a whole, even less of this data is broken down by gender.

That being said, in The Computer Boys Take Over I do state the following:

Contemporary estimates suggest that throughout the 1960s at least thirty percent of working computer programmers were women. One study puts the figure closer to fifty percent.¹

A close reading of the sentence in question suggest that it is a highly qualified statement about contemporary perceptions, and not an absolute claim about percentages. But it since these percentages were so outside of the (at that time) contemporary wisdom about women in computing, it is worth breaking down the evidence I used to justify this sentence.

My immediate source in The Computer Boys for this claim was a 1974 report by the industry analyst Richard Canning. For more than a decade Canning published the influential EDP Analyzer newsletter, and often wrote about the growing labor market for computer programmers. In his 1974 discussion of “Issues in Programming Management,” Canning references Joel Aron, the General Manager of IBM Programming Systems Development, and says the following: “Aron says that about one-half of the programmers are women, and that the number of women managers is rising rapidly.”² Canning, one of the most informed industry observer of the era, repeats this claim without comment or skepticism.

Canning’s source for the Aron quote is actually another contemporary report from the industry journal Infotech. A year previously, Infotech had published a massive 526 page report on “Computing Manpower” that included presentations and invited papers from eighteen industry leaders. The 50% estimate actually appears in this volume three times. The first is in the report from Joel Aron quoted by Canning, of which the relevant sentences are:

Of the programmers, about half are female; and the number of female managers is rising rapidly. At one time, when I had a line management division in the organization, two of the five key managers were women and, under them, about a third of the managers were women. Our experience, which I imagine is fairly typical, is that there is absolutely no difference in the performance of men and women in similar jobs.³

While it is perhaps not necessary to defend the knowledge and experience of an IBM Vice President who had been with the company since 1954 and who was, in addition to having been the manager of several large software development groups at IBM, also the head of the IBM FSC Programming Laboratory and the editor-in-chief IBM sponsored Systems Programming Series (published by Addison-Wesley through at least the early 1980s), there is an earlier reference to the 50% figure. In a 1964 article in the NY Times entitled “Computers Are Getting Ideas From Women: I.B.M. the Leader in Employing Girls as Programers,” Vartanig Vartan suggests that 1200 out of IBM’s 2500 computer programmers were women. 1200/2500 is not quite 50%, but as Joel Aron might argue, close enough for Federal Systems work…

The second reference to a 50% figure in the Infotech report comes from a report by Suzette Harold on “The Role of Women in Computing.” The specific quote is from page 199 of the report, and reads:

Many installations now have more than 50% of women in their programming teams and indeed, commercial programming is in danger of becoming another “women’s job.” I say “in danger” because employers in the past have seen in women a source of cheap labor and once women are segregated into what thereafter becomes known as women’s jobs, equal pay and conditions are inhibited.⁴

Like Joel Aron, Suzette Harold was well-situated to comment on the role of women in computing. Harold served on the Council of Software Houses Association, a U.K.-based software industry organization. She had previously worked on software development at the Atomic Energy Research Establishment, the Rutherford High Energy Laboratory, and the Atlas Computer Laboratory. She was on the scientific staff at the Science Research Council. Like Aron, she had broad experience in the industry.

At the time she presented her findings on women in computing in the 1973 Infotech report, Harold was an executive at F International, which was a software services consulting firm that eventually would become a multi-million dollar firm with hundreds of clients spanning both the U.K. and the European continent. F International employed almost exclusively female programmers.

F International was not the only early software development company to take advantage of the large number of women working in computing in this period. It was itself a subsidiary of Freelance Programmers, a company founded in 1962 by Stephanie “Steve” Shirley. Freelance Programmers was one of the first consulting software development firms, and one of the most successful. When by-then-Dame Stephanie Shirley sold the company in 2007, it was worth £400 million. In 1962, when Freelance Programmers was helping to establish the software industry, it employed only women programmers. It only hired its first male programmers in 1975 when, ironically, it was forced to by the passage of Sex Discrimination Act in the UK. At that point there were 300 women working as programmers at Freelance. So, in 1962 we have a company in which the actual, well-documented percentage of female programmers is 100%.⁵

The third reference to 50% figure in the Infotech report is by the editor of the report C.J. Bunyan, who in his lengthy (150 page) introduction/summary to the report highlights both Aron and Harold’s assessments. Bunyan, who edited and authored numerous industry reports from this period, seemingly found nothing to disagree with in these estimates.

Although I do not cite them in The Computer Boys, there are other references to the 50% figure outside of the Infotech report. A 1964 article in the Guardian by Maureen Eppstein argued that 50% of programmers in the UK were women.⁶ I first discovered this reference in Janet Abbate’s 2012 book Recoding Gender, which is based on both historical research and a series of fifty-two oral histories she conducted with women who had worked as programmers in both the US and UK in the early decades of computing.⁷ On the very first page of the first chapter of that book, Abbate quotes Paula Hawthorn, who worked as a programmer at Texaco in the mid-1906s. According to Hawthorn,

It never occurred to us that computer programming would eventually become something that was thought of as a men’s field. At that time … it was at least half women.⁸

Note here both the (again, contemporary to the late 1960s) claim that 50% of all programmers were women and Hawthorn’s expressed surprise at her realization that programming was anything other than women’s work.

I should note that Janet Abbate herself does not endorse Hawthorn’s claim as being a true reflection of the actual percentage of women in computing in this period. To the degree that she even attempts to provide numerical estimates, she follows the Bureau of Labor Statistics numbers from the early 1970s.

But Hawthorn is not the only woman cited by Abbate who clearly believed that the percentage of programmers in the 1950s who were women was significant. As Elsie Shutt, another pioneering innovator in the software development consulting industry expressed in her interview with Abbate,

It really amazed me that these men were programmers, because I though it was women’s work.”

In this quote Shutt appears to be talking about her time as a programmer at Raytheon in the early 1950s. By 1957, Shutt had been forced to leave the company because she had gotten pregnant, and had founded Computations, Inc., which like Freelance Programmers employed 100% women. Shutt ran her company successfully for more than four decades. As far as I know, none of the employment records that would definitely document this claim remain in existence, but this is typical of the corporate records of smaller companies, and see no reason to doubt Schutt’s assessment of the state of the programming industry in the era that she helped define it.

And so we see that there are multiple, plausible claims from the 1960s that the percentage of computer programmers who were women was as high as (or even more than) 50%. There are also lower estimates. In a 1966 study of programming proficiency, Raymond Berger and Robert Wilson of the Electronic Personnel Research Group at the University of Southern California identified 28% of the 366 government civil service programmers as being women.⁹ This is the source I relied on for my lower abound of roughly 30%. Of course, without knowing exactly how Berger and Wilson selected their sample, it is uncertain how well representative their data represents the industry as a whole, but that is the best data I had to go on at the time, and their report was published in the Association for Computing Machinery (ACM) SIGCPR (Special Interest Group in Computer Personnel Research), the most relevant and prestigious professional publication in this area. When I was researching The Computer Boys, it was not as easy to access historical materials as it is today. And even if it had been, there are simply not many sources of reliable, quantified data from the 1950s and 1960s.

The first official government statistics on programmer employment comes from 1956 report by the US Labor Department on “Employment Opportunities for Women Mathematicians and Statisticians.” “It was a fact,” according to this report, “that many industrial laboratories employed only women in their computing groups; others employ a high percentage of women.” (from Abbate, 65, emphasis mine) The focus of this study on college-educated mathematicians and statisticians seems to suggest that it is talking about higher-level computing occupations such as programming and systems analysis, but it is also true that the term “computing group” potentially covers a variety of types of work, some of which were almost exclusively feminized (such as key-punch operator, or the now obsolete category of “computer operator”), and none of which we would today consider to be aspects of computer programming. That being said, the report also cites specifically one example of a laboratory in which “67% of the programming staff were women.”

The United States Bureau of Labor Statistics did not begin tracking programmer labor until 1970, and their estimate of the percentage of women in computing was 23%. As I (and others) have argued elsewhere, there are numerous reasons to believe that their data under-represents the number of women working as programmers (and, for that matter, the number of men)¹⁰ Bruce Gilchrist, who wrote extensively about labor issues in the computer industry and who held executive positions both in the Association for Computing Machinery (ACM) and the International Federation of Information Processing Societies (IFIPS), reviewed the 1970 BLS report and noted that the BLS numbers were significantly lower than earlier industry estimates.¹¹ Even prior to its publication, he had criticized the BLS report as being “based on relatively little firm data” and noted particularly the “lack of widely accepted standard definitions for occupations such as programmer, systems analyst, computer operator, etc.”¹² These fundamental ambiguities, combined with a reliance on self-reporting and the tendency of some workers to inflate their job titles, meant that some forms of work systematically misrepresented. He notes particularly errors in the categorization of keypunch operators and computer operators, which were roles particularly associated with women.¹³

Over the course of the 1970s, Gilchrist and his co-author Richard Weber attempted to develop a better data set on programmer employment, consistently arguing that the BLS data was insufficient. But even as they developed alternative and more inclusive measures of programming labor, they acknowledged that their own data sources also under-represented the “significant and probably growing amount of programming” that we being performed by people and organizations who were not fully visible in the formal statistics.¹⁴ Often they relied on a different data set from the BLS Area Wage Studies which excluded, among other groups, employees of educational institutions, scientific programmers, agricultural employees, and the self-employed. As far as I am aware, Gilchrist and Weber make only one attempt to estimate the percentage of female programmers, and they highly qualify this in the same way that they do all of their estimates. In that study, their figure of 21% corresponds closely to the BLS statistic.¹⁵ But as they point out in their 1974 attempt at “Enumerating Full Time Programmers,” the lack of standard categories and structural biases in the data meant that, at best, they could “paint only a hazy and incomplete picture of the employment situation in computer related occupations.”¹⁶

The fact that even basic terms like “software” and “programming” were being actively defined and negotiated during the 1950s makes it difficult to rely on job titles or self-descriptions of work. Consider the female operators of the ENIAC, the so-called “ENIAC Girls” (the use of such gendered language, while unpalatable to modern sensibilities, is essential to understanding the gender dynamics of early computing): as both Janet Abbate and Thomas Haigh have correctly noted, both “coder” and “programmer” are anachronisms when applied to the work these women did. Such concepts and categories did exist at the time. And yet the work these women did undeniably helped shape the practices that would become programming, and many of them would go on to work in roles that would become labeled as “programmer.” And they most definitely saw themselves as pioneering programmers.¹⁷ Do we count them as programmers? I would, but I can also see the argument for not doing so. If we do consider them to be computer programmers, than for that brief moment in history, at that particular (but highly visible and influential) place, the percentage of programmers who were women was 100%. As we have seen, much depends on which moment and which organization one chooses as a representative snapshot.

All this being said, the “between 30-50%” figure so often cited from The Computer Boys was never meant to be a definitive estimate of the percentage of working programmers who were women. In my research I generally resist trying to precisely enumerate the number of computer programmers (male or female) working in the period covered by the book (which is roughly 1946-1968). In fact, a key argument of the book is that it would be almost entirely pointless to even attempt to do so, since this is the period during which the very categories of what constituted “computer work” were being actively negotiated. Who “counted” as a computer worker was a highly contested question. Much of the scientific and informational work that today we associate with computers did not originate with computing (data processing was a well-established industry long before the advent of electronic computing, and IBM was a globally dominant information technology company decades before it manufactured its first computer), and many of those who happened to use computers in their work still identified primarily as engineers, scientists, systems analysts, or professional managers. Not only were there no industry-standard job categories for computer work in this period, practitioners themselves used a variety of terms to describe what they did and who they were. Trying to be too precise about the total number of “computer workers” would be a pointless exercise, and in my written work I try to be careful to avoid doing so. In fact, I embrace this ambiguity, in large part because I believe is more authentic to the experience of my historical actors. But that computer programming was seen in the 1950s and 1960s as uniquely accessible to women — in stark contrast to later periods —is undeniable. (For more on my historical work on gender and computing, see the posts YOUR CAREER IN COMPUTER PROGRAMMING and THE COMPUTER GIRLS? as well as my articles “‘Beards, Sandals, and Other Signs of Rugged Individualism’: Masculine Culture within the Computing Professions“¹⁸ and “Making Programming Masculine.”¹⁹). See also the work of Janet Abbate and Mar Hicks, among others.

1
Ensmenger, The Computer Boys Take Over: Computers, Programmers, and the Politics of Technical Expertise (MIT Press, 2010), p. 237.
2
Canning, Richard. 1974. “Issues in Programming Management.” EDP Analyzer 12 (4): 1–14.
3
Aron, Joel. 1973. “Manpower Control in Large Computer Projects.” In Computing Manpower : International Computer State of the Art Report, edited by C. Bunyan, 327–47. Maidenhead: Infotech Information Ltd.
4
Harold, Suzette M. 1973. “The Role of Women in Computing.” In Computing Manpower : International Computer State of the Art Report, edited by C. Bunyan, 195–215. Maidenhead: Infotech Information Ltd.
5
Shirley, Stephanie, and Richard Askwith. 2013. Let It Go: The Story of the Entrepreneur Turned Ardent Philanthropist.
6
Eppstein, Maureen. 1964. “The Computer Women.” Guardian, January 31, 1964.
7
Abbate, Janet. 2012. Recoding Gender: Women’s Changing Participation in Computing. History of Computing. London, England: MIT Press.
8
Abbate, p. 1
9
Berger, Raymond M., and Robert C. Wilson. 1966. “Correlates of Programmer Proficiency.” In Proceedings of the Fourth SIGCPR Conference on Computer Personnel Research – SIGCPR ’66. New York, New York, USA: ACM Press.
10
See Ensmenger, The Computer Boys Take Over, as well as Loseke, Donileen R., and John A. Sonquist. 1979. “The Computer Worker in the Labor Force: New Occupations and Old Problems.” Sociology of Work and Occupations 6 (2): 156–83; Greenbaum, Joan. 1976. “Division of Labor in the Computer Field.” Monthly Review 28 (3): 40.Kraft, Philip, and Steven Dubnoff. 1986. “Job Control, Fragmentation, and Control in Computer Software Work.” Industrial Relations 2 (25).
11
Gilchrist, Bruce, and Richard E. Weber. 1974. “Enumerating Full-Time Programmers.” Communications of the ACM 17 (10): 592–93.
12
Gilchrist. 1969. “Manpower Statistics in the Information Processing Field.” Computers & Automation.
13
Weber, Richard E., and Bruce Gilchrist. 1975. “Discrimination in the Employment of Women in the Computer Industry.” Communications of the ACM 18 (7): 416–18.
14
Weber, and Gilchrist. 1972. “Employment of Trained Computer Personnel: A Quantitative Survey.” Communications of the ACM.
15
Weber, Richard E., and Bruce Gilchrist. 1975. “Discrimination in the Employment of Women in the Computer Industry.” Communications of the ACM 18 (7): 416–18.
16
Gilchrist, Bruce, and Richard E. Weber. 1974. “Enumerating Full-Time Programmers.” Communications of the ACM 17 (10): 592–93.
17
Bartik, Jean, Jon T. Rickman, and Kim D. Todd. 2013. Pioneer Programmer. Truman State University Press.
18
Ensmenger, Nathan. 2015. “‘Beards, Sandals, and Other Signs of Rugged Individualism’: Masculine Culture within the Computing Professions.” Osiris 30 (1): 38–65.
19
Ensmenger, Nathan. 2010. “Making Programming Masculine.” In Gender Codes: Why Women Are Leaving Computing.

Leave a Reply Cancel reply

Computers, Programmers, and the Politics of Technical Expertise