Every other professional field that women entered in significant numbers during the twentieth century shows the same shape: a low base, a rise through the 1970s and 1980s, and continued growth or a plateau at a high level.
Medicine, law, biology, veterinary science, accountancy — all follow that curve.
Computing does not. Women's share of computer science degrees in the United States rose to a peak in the mid-1980s, then declined for roughly two decades before beginning a partial recovery. It is one of the only fields in which women's participation went meaningfully backwards.
That anomaly is analytically valuable. Whatever caused it was specific to computing and specific to a period, which narrows the candidate explanations considerably.
What computing looked like before
Early programming work was not high-status. Hardware was the prestigious part; instructing the machine was regarded as clerical, and it was staffed accordingly. Several of the foundational figures in the field's early decades were women, and this was unremarkable at the time precisely because the work was not considered important.
As the field's status rose, its demographics changed. This is a well-documented dynamic in occupational history, and it runs in both directions — occupations that become female-dominated tend to lose relative pay and status, and occupations that gain status tend to become male-dominated.
That process explains part of the long arc but not the specific inflection in the mid-1980s.
The home computer hypothesis
The most-cited explanation for the timing is the arrival of the personal computer in domestic settings.
The argument runs: home computers arrived in the early 1980s, they were marketed overwhelmingly as toys for boys, and within a few years a cohort arrived at university in which a substantial proportion of male students had years of unstructured machine time and a much smaller proportion of female students did.
Introductory courses, encountering students with wildly divergent prior exposure, calibrated to the experienced group. Students without prior exposure interpreted their relative struggle as evidence of unsuitability rather than of a head start.
The supporting evidence is circumstantial but reasonably strong. Advertising from the period is unambiguous about the intended market. Survey data from the era does show large differences in home computer access by gender of child. And the timing of the enrolment reversal follows the diffusion of home machines with roughly the lag a cohort effect would predict.
It has not been demonstrated causally, and it probably cannot be. It remains the best-supported single account of the timing.
The prior-experience effect in classrooms
The mechanism the hypothesis proposes — introductory courses calibrated to students with prior experience — has been studied directly and does exist.
Institutions that split introductory computing into separate tracks by prior experience have reported substantial improvements in retention of students without background, and several have reported markedly better gender balance as a result. This has been documented at a number of universities that made the change deliberately and published the results.
The intervention is not complicated. It amounts to not putting a student who has never written a line of code in the same room as one who has been writing code since she was twelve, and then grading them on the same curve in week three.
That such a simple change produces measurable effects is itself evidence about how much of the attrition was about experience rather than aptitude.
The culture argument
A second explanation focuses on the field's self-image: the emergence during the same period of a strong cultural association between computing and a particular obsessive, socially narrow personality type.
Studies manipulating the physical environment of computing classrooms — removing science-fiction posters and technology paraphernalia, changing nothing else — have found effects on stated interest among women. These are small experimental studies and should not be over-interpreted, but the direction has replicated.
Surveys of students' images of computing professionals find the stereotype is strong, specific and widely held, including by students who go on to enter the field.
What the recovery suggests
Women's share of computing degrees has been rising again in recent years, and the rise is not uniform. It is concentrated in specific institutions and in specific sub-areas — data science, computational biology, human-computer interaction, and applied programmes framed around problem domains rather than around the machine.
Institutions that restructured introductory sequences and framed the discipline around applications rather than around programming as an end in itself have reported the largest gains, several of them dramatic.
This pattern is consistent with both leading explanations and inconsistent with any account based on fixed differences in interest, since fixed differences would not respond to curriculum design.
Why the history matters now
Because the field's current composition is routinely presented as a natural equilibrium, and it demonstrably is not one. It is the outcome of a specific set of events in a specific twenty-year period, most of which had nothing to do with anyone's aptitude.
A discipline that has already gone up, down and partway back up in living memory is not one whose composition is fixed by anything intrinsic. That is the most useful thing the history tells us.