3 ms·

I said engineering, and I meant engineering. Not computer science. See https://www.aei.org/carpe-diem/table-of-the-day-bachelors-degrees-for-the-class-of-2016

by btilly 7y ago

I said engineering, and I meant engineering. Not computer science. See https://www.aei.org/carpe-diem/table-of-the-day-bachelors-degrees-for-the-class-of-2016-by-field-and-gender-oh-and-the-overall-25-6-college-degree-gap-for-men/ https://www.aei.org/carpe-diem/table-of-the-day-bachelors-de... to verify that in 2016, just under 21% of engineering degrees were given to women. The figures for the mid 1980s were almost the same. But if you can find figures, in the mid-1960s, engineering degrees were almost exclusively given to men. The ratios have been fairly constant in recent decades, but vary a lot from degree to apparently related degree. For example around 40% of degrees in mathematics go to women, and that factor of 2 difference from engineering has not changed appreciably since Reagan was President. Now you are right that women were a large part of early software development. "Computer" used to be a job that was almost exclusively held by women. Who then made the transition from actually being the ones who did the calculations to writing code. (Note, the preponderance of women can be explained by mathematical ability being fairly evenly distributed, but most careers using that ability having been barred to women historically.) See Grace Hopper and Margaret Hamilton for justly famous examples. However I have noticed that the gender ratio seems to be strongly dependent upon language. Some years ago I worked on a project with a COBOL and a Perl group. The COBOL group was largely women, but Perl was almost all men. I got to see what happened in Perl up close. You had a pretty good balance of men and women getting into web development. But a lot of men went from web development to back end web development into programming. Women were more significantly likely to go to front end, design and product management. And neither I nor the women that I talked to saw discrimination as having been the driving factor. So why does it happen? Here is my best guess. Between COBOL and more modern languages we have introduced many language features that are hard for most people to get their heads around. For example recursion, pointers, closures, and class hierarchies. These barriers are sufficiently hard that many teachers privately say that only some people are able to learn to program. My suspicion is that one or more of these concepts is more of a barrier for women than men.

fzeroracer 7y ago
My point was that this article is specifically using the term engineering in the sense of computer science. Using engineering as a whole is misleading because as I have demonstrated, software engineering had a historically larger amount of women which faced a decline post-80s. It's burying the lede to make a point about something as a whole rather than the specifics of the article. Also, your point about those language features being harder is also, objectively wrong. All of those features you mentioned existed before the 80s and were used at the time. If anything software engineering has gotten easier to understand over time, not harder thanks to more robust tooling, debugging and support. Back then you didn't have access to easy resources and had to do everything yourself, so you would have to provide a better explanation than 'things are harder now than they were before'.
- btilly 7y ago
  Also, your point about those language features being harder is also, objectively wrong. All of those features you mentioned existed before the 80s and were used at the time. Existed, yes. Widely used? That was a period of transition! The period from the early 1980s to the late 1980s coincided with a host of changes. Including the rise of personal PCs and Unix, less VMS and mainframes, a decline in COBOL, a rise in C and Pascal, etc, etc, etc. In 1980 it was reasonable to take introductory programming courses in COBOL (the most commonly used language in business) or FORTRAN (the most commonly used language in science). At the time neither language supported recursion, closures, or object oriented programming. Pointers existed, but were not as heavily used as they were later. That cohort graduated 37.1% women in 1984. If you were taking your introductory programming course in 1986 on a personal computer in C or Pascal, you would have dealt with both recursion and pointers heavily. That was about when I took that course. This cohort graduated 29.9% women. Object oriented programming didn't break into the mainstream until the 1990s. (The original Code Complete written in the 1980s doesn't even mention it.) It existed, but was not widely adopted "by industry". Closures began to spread out from Scheme but remained a fairly niche concept until people got serious about them in JavaScript in the 2000s. So the timeline is not evidence against my suspicion. Of course so many other things were in flux that it isn't evidence for either.
- hackinthebochs 7y ago
  >software engineering had a historically larger amount of women which faced a decline post-80s. How much resemblance does software engineering pre-80's and post-80's have with each other? Furthermore, how much generalizations can one make when the sample size is small? Presumably when computer programming first became a distinct activity, the pool of players were significantly smaller than it currently is.
danans 7y ago
> For example recursion, pointers, closures, and class hierarchies. > My suspicion is that one or more of these concepts is more of a barrier for women than men. Seems like quite a stretch, given that those are all concepts created by mathematics, which women have long excelled in - after all, women were the original "computers" before we had machines to automate calculations. Add to that that the hard sciences like chemistry are full of women, and many of the concepts they involve dwarf the complexity of concepts like classes and pointers. What changed for women between the time of early programming languages and C++/Java wasn't pointers and classes, but rather the culture of tech itself moved from academia to business, and big business had a heavy male bias.
- hackinthebochs 7y ago
  > given that those are all concepts created by mathematics, which women have long excelled in - after all, women were the original "computers" before we had machines to automate calculations. Doing rote arithmetic is nothing like doing mathematics. Doing manual calculations is nothing like being a programmer. The activities have zero resemblance. Also, advanced mathematics degrees (i.e. Masters and above) have the same gender parity as computer science. Undergrad mathematics degrees has a more balanced gender parity, but this is explained by a large proportion of those women going on to become teachers.
  - danans 7y ago
    > Doing rote arithmetic is nothing like doing mathematics. Doing manual calculations is nothing like being a programmer. The activities have zero resemblance. Understanding the function of pointers, classes, and most programming itself are also nothing like advanced mathematics. Even ML is largely stats and calculus in its technical aspects, not cutting edge math. There is zero at all evidence to suggest that women are less capable of understanding the concepts behind programming.
    - hackinthebochs 7y ago
      >most programming itself are also nothing like advanced mathematics. This is certainly true. But they do overlap in requiring a heavy amount of abstract reasoning. While "abstract reasoning" is a bit vague, one proxy for this is spatial reasoning which shows strong gender disparities and has implications for science and math ability. https://link.springer.com/chapter/10.1007/978-3-319-44385-0_10 https://link.springer.com/chapter/10.1007/978-3-319-44385-0_...

by btilly 7y ago

fzeroracer 7y ago

My point was that this article is specifically using the term engineering in the sense of computer science. Using engineering as a whole is misleading because as I have demonstrated, software engineering had a historically larger amount of women which faced a decline post-80s. It's burying the lede to make a point about something as a whole rather than the specifics of the article. Also, your point about those language features being harder is also, objectively wrong. All of those features you mentioned existed before the 80s and were used at the time. If anything software engineering has gotten easier to understand over time, not harder thanks to more robust tooling, debugging and support. Back then you didn't have access to easy resources and had to do everything yourself, so you would have to provide a better explanation than 'things are harder now than they were before'.

btilly 7y ago
Also, your point about those language features being harder is also, objectively wrong. All of those features you mentioned existed before the 80s and were used at the time. Existed, yes. Widely used? That was a period of transition! The period from the early 1980s to the late 1980s coincided with a host of changes. Including the rise of personal PCs and Unix, less VMS and mainframes, a decline in COBOL, a rise in C and Pascal, etc, etc, etc. In 1980 it was reasonable to take introductory programming courses in COBOL (the most commonly used language in business) or FORTRAN (the most commonly used language in science). At the time neither language supported recursion, closures, or object oriented programming. Pointers existed, but were not as heavily used as they were later. That cohort graduated 37.1% women in 1984. If you were taking your introductory programming course in 1986 on a personal computer in C or Pascal, you would have dealt with both recursion and pointers heavily. That was about when I took that course. This cohort graduated 29.9% women. Object oriented programming didn't break into the mainstream until the 1990s. (The original Code Complete written in the 1980s doesn't even mention it.) It existed, but was not widely adopted "by industry". Closures began to spread out from Scheme but remained a fairly niche concept until people got serious about them in JavaScript in the 2000s. So the timeline is not evidence against my suspicion. Of course so many other things were in flux that it isn't evidence for either.
hackinthebochs 7y ago
>software engineering had a historically larger amount of women which faced a decline post-80s. How much resemblance does software engineering pre-80's and post-80's have with each other? Furthermore, how much generalizations can one make when the sample size is small? Presumably when computer programming first became a distinct activity, the pool of players were significantly smaller than it currently is.

danans 7y ago

> For example recursion, pointers, closures, and class hierarchies. > My suspicion is that one or more of these concepts is more of a barrier for women than men. Seems like quite a stretch, given that those are all concepts created by mathematics, which women have long excelled in - after all, women were the original "computers" before we had machines to automate calculations. Add to that that the hard sciences like chemistry are full of women, and many of the concepts they involve dwarf the complexity of concepts like classes and pointers. What changed for women between the time of early programming languages and C++/Java wasn't pointers and classes, but rather the culture of tech itself moved from academia to business, and big business had a heavy male bias.

hackinthebochs 7y ago
> given that those are all concepts created by mathematics, which women have long excelled in - after all, women were the original "computers" before we had machines to automate calculations. Doing rote arithmetic is nothing like doing mathematics. Doing manual calculations is nothing like being a programmer. The activities have zero resemblance. Also, advanced mathematics degrees (i.e. Masters and above) have the same gender parity as computer science. Undergrad mathematics degrees has a more balanced gender parity, but this is explained by a large proportion of those women going on to become teachers.
- danans 7y ago
  > Doing rote arithmetic is nothing like doing mathematics. Doing manual calculations is nothing like being a programmer. The activities have zero resemblance. Understanding the function of pointers, classes, and most programming itself are also nothing like advanced mathematics. Even ML is largely stats and calculus in its technical aspects, not cutting edge math. There is zero at all evidence to suggest that women are less capable of understanding the concepts behind programming.
  - hackinthebochs 7y ago
    >most programming itself are also nothing like advanced mathematics. This is certainly true. But they do overlap in requiring a heavy amount of abstract reasoning. While "abstract reasoning" is a bit vague, one proxy for this is spatial reasoning which shows strong gender disparities and has implications for science and math ability. https://link.springer.com/chapter/10.1007/978-3-319-44385-0_10 https://link.springer.com/chapter/10.1007/978-3-319-44385-0_...