The fall of women’s share in computer science

“Engineering the gender gap: Fall of Women’s Share in Computer Science”

Download Job Market Paper (PDF)

No college major is inherently male or female. In this paper, I explore how gender traditions in the U.S influence women’s academic preferences today. I make two arguments. First, the scientific fields which involve more women today coincide with those science subjects included in home economics, an exclusively feminine field. Second, the percentage of women in computer science decreases when this major relocates to the Engineering School, a traditionally masculine domain. I argue that shaping computer science into an engineering subject has constrained women’s ability to reallocate their human capital in response to the technology shock brought by personal computers.

That is from Yiling Zhao, who is on the job market this year from Northwestern.


In general, women these days prefer the life sciences. Women like life. Few women ever thought "I am become Death, destroyer of worlds" was such a cool thing to say that they wanted to go into the death sciences.

A lot of women got into working with computers when it was a 9 to 5 white collar job back in the 1960s and 1970s. But then the personal computer came along and opened up the door to a lot of guys who _really_ loved computers.

Women never had a share in computer science. Data entry used to be classified as part of computer science prior to the widespread use of word processing software. Women were getting training in that field for assistant jobs.

The market also decided to pay women less for computer science for it makes sense for them to look elsewhere.

But to the extent that the market pays women less, it does so in *every* field. So where is this "elsewhere"?

So essentially the argument is that since there is so much math in computer science the little girls can't compete and thus they choose basket weaving or something. Is that really what you want us to believe?
I would love to here this guys explanation as to why there aren't enough male nurses and school teachers

Maybe the argument is that little girls aren't that interested in how computers work so they become nurses and school teachers.

I don't think one should ever let a comment like this pass without pointing out that it is a lie. Many people actually believe it, since many people who know better continue to say it.

So, one trusts that James Damore was cited as authorative in this subject.

I read all of Damore's original "screed". It was not a screed.

I watched part of an interview with him a week ago or so and he made the point that men and women test about equally well in math but woman do much better than men in language arts. Women have more options. I don't know why people glamorize computer science. The bulk of the jobs are in software engineering - coding and debugging - and it's hard and lonely work. You would really have to a nerd or on the spectrum to choose software development when you have the ability to leverage your interpersonal skills.

I think women are happy with their choices.

Agreed, Damore's thoughtful paper was well written and he was right on everything but made one relatively minor omission. He should have included a bell curve showing the distribution of male and female IQs - with a greater gap between men and women on the right side of the curve - rather than a straight line bar graph of average IQs.

"I don't know why people glamorize computer science"

It is not necessarily glamorized for those in the middle ranks but at the top of say the Forbes list you have an impressive run of tech billionaires and young billionaires tend to be from tech. For non-billionaires, again not glamorized but it pays some of the highest average salaries for someone who doesn't want to go further than a bachelor's degree.

Those people arent tech people , they are charismatic leaders with a fair understanding of what is possible with computers.

Gates and Zuck and Woz and Page and the rest could all code as well as most.

And they stopped coding as soon as they could financially afford to.

"Oddly, the people saying that tended to have no interest in women joining the bar in the first place."

I'm doing paragliding. There is something like 95%men/5% women in the sport.

We DO have even special programs for women - the categories in competition are 'Open' and 'Women' (not Man and Women), so if you are a women and would like to compete, you can get a prize quite easily (i.e. out of ~ 130 competitors, we can barely get over the minimum 5 in the women category). If a women wins the Open category (very rarely happens in individual races), everybody cheers. We would actually like very much if more women went into the sport.

The women I spoke with generally said they are happy not to do that sport. Like 'we are clever enough not to do that'.

"Though clearly not out of hostility to women, but simply because the natural order pre-ordained men to be lawyers, unlike women"

Yes, I think women are happy with their choices. Yes, the psychologist have measured that on average women seem to be more risk-averse. Paragliding is a risky sport. Yes, it does seem that 'the nature' made women made - on average - different choices than men. Do you dispute that? Do you suggest that I'm saying that out of hostility to women?

How is that different to Compute Science?

Biking anecdote notwithstanding, it is in the Scandinavian countries, where women have been most encouraged and supported in following their hearts when choosing their careers, that we find the largest gender differences in career choices.

"Obviously I dispute such a simplistic perspective using 'nature' as an explanation for why there were so few female lawyers in the U.S. than 1969"

I didn't gave that explanation...

"those barriers were effective until dismantled seems beyond discussion, based on actual empirical data involving law school admissions or those taking bar exams. "

So what barriers we do have in paragliding? I would be very interested to see such discussion as it seems to me you are offering a very simplistic explanation.

"Maybe not all that significantly, with so many people arguing that women are not suited, because of 'nature,' to be attorneys ... oops, programmers."

Errr... that's not what Damore wrote. It looks like the only people who oppose his memo are those who didn't read it. Right?

Does the following summarise your argument?

Because in the past there were barriers for women to enter some fields, it follows that all the differences in participation in different activities today are results of such barriers today and none are caused by natural differences between men and women.


I read the Damore's memo and I always wondered, how could anyone draw any conclusion with respect to 'underlying hostility to women' from it. Many people did - so I wonder, what does trigger that? Could you explain? Is it something particular in the memo?

I don't recall the details now, but I do recall me reading his (in)famous memo and saying to myself "ok, ok, hmm what? oh boy..." I'd say that if you have to ask what's wrong with it, then you should avoid writing about the topic of gender discrimination. If I recall correctly, and I might not, he had some specific complaints and then wandered off into making some broad generalizations. On average we're all dead. Or to put it another way, averages and generalizations aren't hired. Individuals are hired. Try to put yourself into the shoes of a woman peer of Damore. It isn't a stretch to think that she'd likely to have experienced some 'hostility' or 'sexual objectification' during her education and career. Try reading his memo from that viewpoint and see if you can identify where he went off the rails. If you can't ask a female do HAVE female peers, right?

"If I recall correctly, and I might not, he had some specific complaints and then wandered off into making some broad generalizations"

You don't recall that correctly.

Try reading it and underlining all the “weasel words” like “might” or “suggests that”. You end up with a sea of red ink. I’m not sure he made a single “specific complaint” in the whole essay. As far as actual conclusions, it says almost nothing.

As far as the conclusions, it says that the gender disparity in Google is very unlikely a result of gender discrimination, and thus google's diversity programs are wrong-headed.

As for 'weasel words' - so do you subscribe to the line of thought I have mentioned?

The prevailing argument I saw was like this: "he obviously doesn't like women -- therefore he meant the opposite of what he wrote/he makes this particular argument to cover his hatered -- therefore the whole memo is anti-women and the women are rightly offended".

I have not found any weasel words. Can you be specific?

"Or to put it another way, averages and generalizations aren't hired. Individuals are hired."

He writes almost exactly that in the memo. That's why I was asking what triggered the people.

The prevailing argument I saw was like this: "he obviously doesn't like women -- therefore he meant the opposite of what he wrote/he makes this particular argument to cover his hatered -- therefore the whole memo is anti-women and the women are rightly offended". I find it hard to concur with such way of thinking.

Wasn't the point of the memo to show that Google "hiring individuals" predictably leads to the current employee demographics at Google. Whereas the premise of the diversity promotion meetings seems to be that the demographics of Google employees is a result of Google "hiring groups". He is arguing that Google *isn't* discriminating and they fired him for it.

He said that Google's diversity efforts were wrong-headed (because they were unrealistic and based on false assumption).

But, there's a line of thought that *all* arguments are actually just post-hoc rationalizations of what you already wanted. Which is fairly popular with the "everything is about power and oppression" crowd.

So his *reasons* don't matter, all that matters is that he argued against google's diversity efforts. Which clearly must mean that he's a bigot, because that's the only reason anyone would ever look for a an excuse to argue against diversity. And once that's established, it's fairly easy to spin the 1/3 or whatever of his facts weren't actually true -- IIRC level-headed analyses said he got the science about 60% correct -- as proof that he was just digging for excuses without much care about what was actually true.

Wasn't he actually arguing for more effective diversity efforts that would actually result in more women and less men?

> I'd say that if you have to ask what's wrong with it, then you should avoid writing about the topic of gender discrimination.

I'd say that if you can't form an argument or remember anything you pretend that you read, then you should avoid writing about the topic of gender discrimination.

"On average we're all dead."

What does this mean?

Her paper is not available on her website. In recent years the percent of women in computer science has started to rise, and I wonder if her model predicts or explains that.

It's possible that the explanation is simply that STEM has been hammered into people's brains as the only worthwhile set of majors, and concomitantly students are streaming out of the humanities. I.e. a phenomenon that's about higher education and not computer science per se. But maybe things have been changing in computer science itself.

Yes, STEM programs are considered the only worthwhile majors because STEM careers are seen as having more openings and better financial prospects. That's what it's all about in the US of A, making as much money as possible and retiring so one can do what one really wants to do, which isn't working.

....want to live well, and for that you need money.

Or, the STEM fields are seen as the only worthwhile ones because the left-wing 'march through the institutions' has destroyed the value of the Humanities, which are now producing ideologues instead of thoughtful, well-rounded people who were exposed to a wide variety of intellectual thought - which used to be the goal of a liberal arts education.

They can't easily get away with it in STEM, because in STEM facts still matter. So the are trying to introduce new STEM programs like 'Engineering Ethics' where they can import their poisonous ideology without fear of having to learn math.

Where is the paper? Couldn't find it.

Back after a month of self-imposed exile. Nothing to add here except this is a 'past as prologue' theme that makes this assumption as a prior. The heavy hand of history as Fernand Braudel once called it. That's why you still get poverty in and around the Potasi silver (slave) mines, 500 years later. Nature or nuture? Keep in mind Pb is found in Ag mines which affects the childhood mind, so it could be a bit of both, same with this paper, maybe testosterone helps in math? But then you have super GMs Judith Polgar and Hou Yifan as counterexamples (nuture).

Bonus trivia: I added value, being in the 1%, the old fashioned way the last month overseas. In a certain south European country that values old olive trees, with "California style" environmental regulations that make it mandatory you make an environmental impact study before you chop down said trees, which takes forever, is sometimes not granted, I seized the bull's horns, chopped down the trees, made the lot into a parking lot (it's in a city, and logical to do so), and increased the value of my real estate by at least double in one month. I'll pay a small fine years from now, if at all, if any tree hugger complains (one already has promised to report me, pfft). That's how the rich add value. They skirt the boundary of the law, bending it. It's what the 1% do, y'all. Peace.

Yesterday, the Treasury Department announced that it was repealing regulations adopted during the Obama administration to discourage "inversions" (in which a U.S. company merges with a foreign company in order to avoid U.S. tax). Treasury officials said the regulations were no longer necessary because of the corporate tax cut. If the regulations are no longer necessary, why repeal them? So the Ray Lopez of corporations can do an inversion. Why don't all U.S. corporations do an inversion? Chaos (i.e., the Treasury officials' explanation for the repeal of the regulations would prove to be a lie).

From Bloomberg: "The new rule would give companies some leeway to prove that two transactions aren’t related and don’t deserve adverse tax consequences." Sounds good so far.

"Businesses resisted the original rules, published in 2016, arguing that the IRS was overstepping its authority. " Hallmark of Obama, overstepping authority.

"The regulations, commonly called tax code Section 385 rules, were some of the most controversial tax rules ever written at the time"
Pass just as he was leaving office, they we just an FU to businesses on the way out. Typical.

Section 385 has nothing to do with inversions; it's the Code section authorizing the IRS to issue regulations that help distinguish equity and debt. Yesterday's Treasury announcement also "streamlined" those rules. The rules for distinguishing equity and debt are sometimes referred to as the Milken Rules, named for felon Michael Milken who became very rich by blurring the distinction. Milken has been in the news recently, having exploited his relationship with the Secretary of the Treasury to obtain favorable treatment of an investment in an "opportunity zone" in what was, ironically, an industrial park but converted to just another shopping center and office park by Milken.

Investment in "opportunity zones" (sometimes jokingly referred to as the "O zone") qualify for favorable tax treatment. The purpose is to induce investment in areas that are economically depressed. In Milken's case, he already made the investment in the land that would later become part of an "opportunity zone" (raw land that had been an industrial park). So what did Milken do? He persuaded the Treasury to issue a rule that one who owned the property before "opportunity zone" designation could still qualify for favorable tax treatment if the taxpayer (Milken or his affiliated entity in this case) "leased" the property to someone else who would develop the property. Milken had blurred the distinction between equity and debt, and I suspect his "lease" blurs the distinction between ownership of fee title and ownership of a "lease". Here is the NYT expose: More significantly, here is Cato piling on:

Trump is chaotic? Look at the farce that is the House today. What a joke. It has spent 2 years trying to undo a valid election at any cost. Look at the reaction from the NYT and others when Trump retweets a cute joke pic of the dog. God help us, but Trump is calm in the wild sea these days.

Can't really tell if you believe your bullshit. Are you trolling or just dumb?

Trolling the dumb? Seems likely.

So you don't believe this crap, you're just trolling. Carry on.

Can easily be both. Nothing in the comment was controversial, but I did know it would trigger the double digit IQers. Thanks for participating.

It's just total bullshit, which even you yourself don't believe. And that's fine, trolling is a noble profession. But you're pretty transparent.

Dude, I'm not saying Trump is not half crazy, just that he's the sane one in the room full of Democrats. Easily defendable, if not painfully obvious. The troll part is that there's going to be some retard reacting to this obvious observation. Again, thanks for doing your part.

Surrender accepted.

The tech start-up bubble single-handedly pumped by Soft Bank is deflating.
WeWork and Wag crashed before IPO, Uber and Slack had gone down since IPO.

It will be fun to visit reddit to read the sob stories of nerds similar to James Damore after all this ends. Sorry guys, you thought your coding skills were developing a great product for customers while the real product was the company valuation.

My impression is that software engineers (coding and debugging) like to work alone and be left alone. Is that an accurate description? My sample size is very small. My niece works in IT for a large health care system. She is mostly self-taught (she was a business major in college well before computer science became a thing). From time to time they suffer a software crash, and when they do I've noticed that she often complains that it's difficult to contact the male debugger on which she/they rely. He is a loner. Women are not inclined to be loners.

Liking to be left alone while working on a job that requires collaboration is a problem.

If an individual successful and loner, quite probably succeeded in spite, not because of being a loner. Just because someone is successful, it doesn't mean any defects of character become praise-worthy features.

Being somewhat of a loner is a good thing for coding. You really do need quiet time to code. In the time of crisis as Ray's example, the debugger is probably better off no answering as it would be, 'STFU and let me fix this'.

Around 20% of my billing hours are coding.

Being perseverant and focused is much better. It's tempting to fall in the self-indulgence of believing that one of my character flaws is a positive thing. Albeit, one day you're at the client's site, with the project manager behind your back while you're trying to fix something. 'STFU and let me fix this' is not an option.

Everyone has a different tolerance to stress, noise and disruptions. If someone needs 15 min to focus, fine we're all different. But it's not a requisite, it's a personal quirk and certainly not a feature.

All true, but especially in troubleshooting mode, having someone get status every 5 min is disrupting.

Is being a loner a good thing? If yes, how dare you say women can't be loners! If not, men need to stop being toxic loners!

"Autism is significantly more common in boys than in girls. This skewed sex ratio has been recognized since the first cases of autism were described in the 1940s. The exact reasons for the ratio remain unclear. It could be rooted in biological differences between the sexes. Or, some experts say, it may be an artifact of the way autism is defined and diagnosed.....The most comprehensive analysis of autism’s sex ratio, published in 2017, drew on data from 54 prevalence studies worldwide. That analysis estimated about 4.2 boys with autism for every girl."

The percentage of software engineers with autism is approximately zero.

They're bloody good at imitating it, then.

It typically takes about 15 minutes for someone to achieve real focus on a complex task like programming. So if you interrupt them every 5 minutes they never reach the high productivity zone.

As a manager, that focus is an asset to be protected. Offices with doors, phones to voicemail (unless your job is support), etc. A need or preference for continuous chit chat is a bug, not a feature.

More like 30 minutes on SQL coding (it's brain burning). And if you aren't in the zone you are just puttering around.

I worked with some of the best sql programmers in the world(*). The first thing they said, and I later said, to anyone suffering was "no gui." Learn to be expressive at the naked sql prompt. It's easier to build a commend up, or tear one down for understanding there. Then you can later embed that stuff into another framework, another language with bindings.

And I hope anyone coding has a good scratch copy of the database to work with. It's good engineering to have tools to create those as needed.

I've had to invoke commands on a big live customer database, and that isn't fun. But it should certainly be a rare event. In all but the most desperate circumstances that should be done with scripts developed and QA'd on database copies. That's engineering again.

* - not to be too specific, but I worked with an Oracle development partner back in the day, and went through a few series of Oracle training.

My impression is that software engineers (coding and debugging) like to work alone and be left alone.

Actually, most developers work in teams. Much of the work is pretty collaborative. It's common for developers to want to talk through particularly sticky problems with colleagues. Some organizations formalize it with 'design reviews' and 'code reviews' where particular approaches and solutions are discussed and critiqued. Also, depending on the particular domain, there may also be a lot of personal interaction with customers, users and managers.

"Actually, most developers work in teams."

+1, the example of a lone coder doing complete software packages without collaborating with anyone else is rare and mostly a product of the movies. I know a couple of guys like that, and a few dozen who aren't.

I think most coders I know prefer to spend time coding, and everyone once in a while talk over an issue or the best approach with someone else. Sometimes that talk is just Skype or email, but it's still not fundamentally a solitaire approach.

Coders work in teams, but not the way you'd think. In Agile programming there is a lot of effort made to lower the 'friction' of meetings and other group work. So meetings are often 15 minute 'standup' meetings to encourage fast resolution. Most people then go to their desks and work quietly by themselves, consulting others only when they have issues they can't solve themselves.

There was a fad for 'Extreme programming', part of which involved working all the time in pairs. Almost every developer I've known hated it and much preferred working alone.

Within that broad spectrum you find some people that put on headphones and shut out the world as much as they can, and others who are constantly talking, asking questions of others, seeking help, etc.

In my experience, the ones who just hunker down and get the job done solo are about 10x more productive than the ones who are more social.

In software, there can be as much as a 100x difference in productivity between good programmers and bad ones. When an organization takes its eye off the ball and starts looking for diversity hires rather than the absolute best people it can get, it's devastating for productivity. One below-average programmer can kill the effectiveness of a team, or introduce bugs that delay releases and cost many multiples of their salary.

This is not to say that women can't be as good as men. I have worked with excellent female programmers, and worked under excellent female managers, But when schools are only graduating 25% women but your company has a mandate for 50/50 male/female split, you have to hire pretty much every woman who applies, including some whose school records and experience would never otherwise get them hired. Add too many of these weak people to a programming team, and you will destroy productivity,

"In my experience, the ones who just hunker down and get the job done solo are about 10x more productive than the ones who are more social."

Except that the only ones who can spend all their days 'hunkered down' are code monkeys who take in specs and grind out code and play no role in design or planning and have no contact with customers or users.

My larger point is that software development isn't a particularly solitary pursuit. Not compared to a number of other occupations including 'novelist', 'composer', or 'artist' or to say plumbers or electricians who spend their days driving through a series of jobs and whose only human contact is a brief hello and goodbye to the homeowners who let them in to work.

"I argue that shaping computer science into an engineering subject has constrained women’s ability to reallocate their human capital in response to the technology shock brought by personal computers."Does her research document the harm that has been caused to consumers?

"Second, the percentage of women in computer science decreases when this major relocates to the Engineering School"

Generally, moving computer science to the Engineering School means adding some much more difficult math and advanced programming classes. That adds value to the degree. However, it will preclude some people who could have gotten the degree when it was lighter on the advanced math and programming classes.

Are we to understand from this item, Tyler, that Northwestern U. does NOT offer graduate degrees (yet) in Culinary Engineering?

I would posit that the endeavors that would attract women have certain characteristics.
Reasonable hours. No death march projects, 80 hour work weeks as part of the expectation.
An individual is replaceable. A women over her working life would expect to be able to leave for periods of time and return, and the timing cannot be scheduled. So the job needs to be able to be done by someone else.
Physical demands must be moderate. Brute strength jobs, where your production is measured by tons of steel installed won't attract women.
People not things. If there is a choice, fewer women will chose occupations that are about things, although many of those occupations have important jobs of managing people in some way.

In my experience women cluster in occupations like this, and make arrangements to fit their life goals of having children. With exceptions of course, but they work to live, not the other way around.

So the question is how computer science or engineering fits. I know I chose another field even though computer engineering was extremely attractive to me, simply because at the time the industry ate people and was very unattractive if you had a family. I think that has changed, but I appreciate the nature of my job where I'm not chained to a desk.

In conversations with people in health care, the work by doctors is changing as women take over the field. Older men who are nearing retirement age can't find replacements because no women will work the hours they work. And the men won't either because the women in their lives won't put up with it. One internist/oncologist has for years carried an extraordinary burden in this remote area, and he is still working because no one is willing to take on the work load and responsibility he carries. Maybe quite logically, but that was the expectation of men entering the field for decades.

There are opportunities for both men and women to step up, but neither seem willing.

They could open up more places in medical schools to increase doctor supply. Although medicine is a challenging subject and you do want good people the sky-high requirements to get into medical school are probably not needed.

Let's not forget the research that shows that in societies that are scored most open to women, which have the highest female participation in the labor force, which strive for sex equality in child rearing, education, and in government (Scandinavia usually tops these charts), women opt out of Engineering and Computer Science and into nursing and teaching in even higher proportions than in societies scored less open to women. If women with the greatest opportunities are opting out of these fields, what is the problem that we are fixing by actively trying to change their minds? Why should we be making an effort beyond equality of opportunity?

I am a Chemical Engineering professor and former practicing engineer. Both my Junior and Senior classes this term are 50/50 male/female, but that is not true of Mechanical or Civil Engineering. We talk a lot about this 'problem', but is it really a problem? I'm not anti-feminist, I have a 20-year-old daughter (she wants to be a doctor), I am aware of the locker room atmosphere that you can find in some Engineering offices (which should be discouraged), but how hard should we be pushing young women into these fields?

"how hard should we be pushing young women into these fields" Tom, how in the world did you write this? No, I'm not a millennial offended by every unintentional slight. But "hard" and "pushing" in one sentence about women? Since you are a professor at a college, I suggest you have your TA (a woman, maybe?) review what you write before you publish it. I'm old and am not part of a college community, but I know that such communities are dangerous, as the "take offense police" are out in full force. I'm a lawyer and am accustomed to being offended; indeed, if my adversary gives me a compliment, my first reaction is that I have overlooked something and she has the advantage.

"No college major is inherently male or female."
Goodness. What advisor would let their JMC write such drivel.

I did a career in computer science, without actually having a degree in computer science. Based on 30 years in the trenches, I'm reluctant to agree that "shaping computer science into an engineering subject" was a decision by schools alone. The applications of computers and software are huge, but a great deal of them require precision. And once you have precision, isn't that just semantically engineering?

Sure, you can probably find a private college who will do a B.A. Computers in Art for you, without much emphasis on reproducibility of results.

But if you are going to be hired by Kaiser Permanente or Fidelity, to program the big computers?

" I argue that shaping computer science into an engineering subject has constrained women’s ability to reallocate their human capital"

As so often in identity politics, not doing something is identical to being prevented from doing something. Just don't ask why- that makes you a Bad Person.

As an aside, any women out there find it insulting that this assumes that women are simply incapable of "reallocation their human capital" into an engineering discipline?

Every now and then he goes on one of his obsessive diatribes, and gets his thread deleted.

He was fired from GMU a couple of decades ago and holds a grudge.

I think the moderators are finally tired of his act.

Good Lord, what had he done? (Or left undone?)

Comments for this post are closed