How CMU Increased the Proportion of Women in Computer Science (2013) | Hacker News Reader

How CMU Increased the Proportion of Women in Computer Science (2013)

blog.play-i.com

36 pointssuperlucy46 comments

I think that the biggest one is number 2 (giving more emphasis to CS potential rather than CS skill in the admissions process).

Somehow we've made a society in which getting into computer programming at a young age is generally not something young girls are taught to want. Consequently, the huge majority of young programmers (below college age) are boys.

I started programming when I was 12. The advantage this has given me in the tech world (academia and industry) cannot be overstated. There is a giant gap between people who start programming at a young age due to desire and curiosity and people who learn to for coursework at uni.

So while not giving preferential treatment to the boys whom society has not discouraged from programming at a young age (and seems to discourage girls from it) is definitely going to increase the number of girls in the department (which is a good thing), I also think we should really figure out what element of society is making programming an undesirable hobby for young ladies.

I started programming when I was about 11. I did it on my own at home, and also at school, where our classroom had an Apple II computer. At school, I was joined by (especially) two others strongly interested in programming, one boy and one girl.

Looking back, I could not identify anything whatsoever that would have made our programming activities especially appealing to either boys or girls, nor especially undesirable to either boys or girls. Really, I had the impression that pretty much everyone else -- boys, girls, and teachers -- thought our interest in programming was kind of strange, and being a boy (or a girl) had nothing to do with it.

An anecdote, of course, and I have no idea how, or if, it fits into the bigger picture.

It's definitely hard to see. That's part of the problem. But if you look at the number if programmers under age 18, I suspect the gap will be even wider than > 18. Something about society is pushing young women away from it.

Some of my suspicions:

1. The internet is generally more hostile to girls. If you stumble into the wrong chat room or forum, you get "show your tits". When I started programming, I got on the internet for help and met some amazing friends who helped me become a better programmer.

I absolutely hate when gender topics come up on HN because I generally like this website, but gender topics tend to bring the assholes out of the woodwork. I got a female CS student friend of mine reading HN regularly, but when she ran into one of these sorts of threads she was appalled. And rightfully so. And HN isn't nearly as bad as the internet at large.

2. It's tied to video games being male majority. I played computer video games a lot more than any girls I knew as a kid. This exposure to computers (and config files, etc) set the stage for programming.

The video game realm is making a lot of progress in getting girls playing (I think more or consoles than PC, but the general trend is up everywhere), and I've seen this positively influence girls' predispositions towards programming.

3. There is a positive feedback loop in social circles, particularly in formative years (middle school). Kids encourage the status quo. So it's hard for a kid to develop hobbies outside of their clique. I think this relates to the gender problem because girls at that age tend to be even more exclusive than boys (at least in my experience and everyone I've asked).

But those are just suspicions and the real problem could be more fundamental.

I start programming below college age as well. I also have some friends who learned programming by themselves. The major difference I can tell between those (used to be) young programmers and those who started programming in college is their motivation and passion. Those who started programming as boys/girls usually learned it by themselves and were motivated by curiosity. Knowledge can be learned, experience can be accumulated. But the passion of learning things almost defines one's behaviour. I saw too many people don't want to learn new things because they simply don't like to learn.

Sorry, a little bit OT.

Your comment is self-contradicting: If there's a "giant" gap between boys who start young and girls or boys who don't, then acknowledging that gap at the application stage is not preferential treatment. Ignoring it because it doesn't favour girls, on the other hand, is preferential treatment.

Edited to add: The time to intervene, if it to be equitable, is around the age of 12. But getting the interest of 12 year olds who wouldn't otherwise participate is a more difficult task than changing admission criteria at colleges.

Where is the contradiction?

I was pointing out essentially what your edit suggests. Somehow we've made a society in which young boys favor programming more than girls.

I wasn't saying that people who have been programming since age 12 should be on equal footing in terms of the admissions process. As I said in some comments below, I think that the non-preferential treatment is instead geared towards people who have dabbled in programming a bit (AP Computer Science or a similar course in highschool). So at face value, the person who has 1 year experience in Java will appear to be a better fit for the program than someone with no experience, but by graduation time, there will probably be no correlation between those who came in with 1 year experience and those without any.

Someone who has been programming as a hobby since age 12 will probably be highly correlated with success by graduation time. I don't think CMU is disputing that. Rather, they are disputing the first case of someone with some experience vs. someone with no experience.

So I think there was just a misunderstanding when you read my comment, since I was pretty much saying (longer-windedly) what you did in your edit.

> Where is the contradiction?

You referenced, approvingly I assume, emphasizing "potential" over skill in admissions.

You'd like for it not to unfairly impact boys who've had a genuine interest since a young age. Unfortunately CMU's policy seems to be to disregard all levels of prior experience. They contrast the newer polices with the old, which gave preference to "lots of previous programming experience". (not just a little bit) They target aptitude and "non-academic strengths". (a lot or a little previous experience doesn't fit here). Finally, if you are actually giving weight to experience, explicitly stating "no prior experience required" is a funny way to go about it.

One last thing: a corollary of my edit is that an intervention that changes requirements however they need to be changed, just to get more girls in the door, is inequitable.

I do think that girls at that age are less likely to be interested in solitary hobbies like programming, as opposed to more social hobbies. That said, I also think this is one of the key distinctions between hacking around as a teenager, and programming as a professional: the latter is a deeply social activity.

Girls engage in other solitary activities like literature, music, art, crafts etc.

I'm a CS major at CMU graduating this year and I think I can help elaborate on the "potential vs. current level" thing a bit. Things have changed somewhat since I entered the school, but the ideas are mostly the same.

For my class, there were three intro-level programming courses, each a prereq for the next: introductory programming for those with no experience, introduction to data structures, and fundamentals of algorithm and data structure analysis. A pre-test given over the summer was used to establish which class you were put into. According to the department, roughly one-third of the class ended up in each one. I had no programming experience coming in, so I was put in the first class.

The final course of that intro sequence was also the major fundamentals course for one of the three tracks that the curriculum branched out to after the first intro class, focusing on algorithms and their implementations. The other two tracks focused on systems programming and mathematical interaction (via functional programming).

Every student, regardless of their intro class, typically completes these tracks by the end of their second year or early into their third year, which opens up basically all of the 300- or 400-level specialization courses in the department. Which is why the school doesn't need to focus on people with lots of experience coming in--the structure of the curriculum means that everyone gets to the same level of important fundamentals pretty quickly, and once you've done that you get to branch out and specialize into whatever you want. The more experienced newcomers just have more time to do that.

And anecdotally, the approach has definitely worked. I've got a great job lined up starting this summer that I'm excited to start. I am not a woman so I can't speak to its effectiveness in attracting women specifically, but in my experience the school has done a great job accommodating a wide range of experience levels.

As a CMU alumn, I have to add that part of the reason that the CS department has the 3 levels is that most majors at the school require a programming course.

I'd guess that your intro course had more non-CS majors in it than the other 2 freshmen courses which tend to have a higher proportion of CS majors.

I had a chance to talk to the undergraduate dean of SCS about #2. Essentially they found that "geekiness" (as he described it) had no correlation with performance in the program, but a huge correlation with gender.

Experience, if not geekiness, still counts. Lots of boys get into coding in high school, with or without formal training. My daughter, who is in SCS at CMU now, was fortunate to transfer from her high school to Mass Academy, adjacent to the WPI campus, where she took WPI's intro course that's somewhat SICP-based. I believe that helped significantly, compared to other women in SCS who are certainly intelligent and in all other ways capable enough, but were starting without experience, and hence without informed expectations of what they were about to embark upon.

[EDIT:] The SCS curriculum is designed to minimize the academic effect of coming in "cold," and all majors, especially ones where outsiders have little notion of what's current have this risk, but there's still a relatively larger potential among women for the shock of not knowing what exactly you signed up for.

I've recently been toying with the idea that the "brogrammer" trend is in reality little more than a term used to vilify male programmers who, more than anything else, share the common trait of having rejected "geek/nerd/gamer" culture or fashion.

The term is male specific because female enrollment has been suppressed for the past few decades, making males the most prominent members of just about any fashion/culture trend in tech. The term has become synonymous with sexism because many (but by no means all, or even most) "brogrammers" are indeed sexist (as are other programmers). Because "brogrammers" are different, this sexism is recognized for what it is more frequently than sexism from "non-bro programmers" (who, for most programmers, are part of the in-group).

If we stop selecting for "geek/nerd/gamer", I think that we might see the evaporation of the concept of "brogrammer" as a particularly problematic gender specific trend.

(One of the assumptions that I make is that most people who are labeled as "brogrammers" by others do not self-identify as "brogrammers".)

>If we stop selecting for "geek/nerd/gamer"

I think that is the key problem. I don't know if it is just a western cultural issue but the association between social awkwardness and technical competence is the real problem. My personal experience is that if you don't fit that archetype people assume (erroneously) that you aren't technical enough.

This is a problem not just affecting women but men as well, as someone who's highly technical but doesn't come across as "nerd/geek/gamer" people universally assume I am not as competent, sometimes even when undeniable evidence is present I get treated as a magician with a bag tricks :)

I've seen friends get labeled "brogrammer" just because they wore polo shirts instead of t-shirts with gaming stuff printed on them, and because in university they belonged to a fraternity instead of the anime club. These people don't spend all day in the office doing kegstands, they work just as hard and just as effectively as anybody else, but they are derisively labeled "brogrammer" and the depth of their skill is routinely questioned.

Above a certain age, 30-35 or so I'd estimate, not being "nerd/geek/gamer" seems to lose it's stigma as "family man" becomes a more common occurrence (but of course with that comes age discrimination).

(I believe the tie-in to sexism is fairly complex. Selection for "nerd/gamer/programmer" probably causes a gender imbalance in the industry, but I think that genuine sexism and not just prejudice against lifestyles/fashions also plays an important role. Evening out the gender imbalance would likely have the effect of loosening prejudices against lifestyles/fashions, even if those prejudices were the cause of the gender imbalance in the first place. I believe the issues are related in a more complex manner than "one caused the other".)

You have talked about who is labeled a brogrammer, but can you clarify who is doing the calling? You seem to be saying it is the geeky types themselves. And where are you geographically, roughly?

I am not in Silicon Valley. I spent the first 11 years of my career in the Midwest in the corporate world where you had to dress professionally, didn't get to put up posters in your cube or shoot each other with nerf guns, and sexual harassment would get you instafired. Being a "brogrammer" would not fly there.

For the last couple years I am surrounded by "geek" types and can wear a t-shirt and jeans to work, but I still haven't seen any disdain for "non-geek" types. The only place I hear the term "brogrammer" is here, Twitter and a subset of the media, and it seems primarily by feminists/progressives. So my experience is quite different from yours.

I am emphatically not saying it doesn't exist, I just wonder if it's a "thing" once you get beyond Silicon Valley/Mountain View/San Francisco.

I've noticed it primarily in SF and Seattle. I've seen it to a limited extent in Philadelphia, but it's been my observation that most tech jobs on the East Coast have sufficiently strict dress codes to make lifestyles/hobbies relatively non-apparent at first glance. (the limited extent that I witnessed it in Philadelphia was at a startup that I interviewed with. One of my "geek" friends warned me that the startup was filled with "brogrammers").

Although the term "brogrammer" was not invented yet, I also witnessed it in university on the East Coast, where CS kids in fraternities were frequently given a hard time by other CS students.

It definitely is "geeks" doing most of the name calling, though the twitter/SJW sphere has certainly adopted and perpetuated it as well.

Um, how CMU:SCS has a gender ratio higher than average is because they accept more women. In 2000 the acceptance rate for women was 39%, for men it was 9% [1].

This is completely boring, MIT could make it so its CS class was 100% women if it wanted to, and because loads of very intelligent people apply to MIT I am sure that class of 100% women would do well. But, this is not a tactic that your average school who struggles to get qualified applicants at all can utilize.

[1] https://www.cs.cmu.edu/~lblum/PAPERS/women_in_computer_scien...

I take it from your comment that you infer that it was 39% because they had lower standards for women. Therefore, they are affirmative action babies who shouldn't be there.

I think it is more likely that their outreach programs (through CS AP teacher training etc.) that encouraged the best and brightest women to apply were successful, so that the pool of women were significantly more qualified.

Also, I wonder what the more recent numbers look, as 2000 was a long time ago.

> I think it is more likely that their outreach programs (through CS AP teacher training etc.) that encouraged the best and brightest women to apply were successful, so that the pool of women were significantly more qualified.

That is entirely implausible. Those straws are miles away, dude.

That could perhaps explain a 20% difference, not a 330% difference.

See also: "How Harvey Mudd College Increased Its Percentage of Women in CS", a ten-minute interview with Harvey Mudd's President, Dr. Maria Klawe.

https://www.youtube.com/watch?v=8Ssh6ZcOorw

Prior to now, many of the approaches I'd seen targeted women specifically. The tactics listed in this article instead focus on trying to abstract out what could make a good developer _in the future_.

This is elegant to me because it approaches the problem in a bottom up way: "how can we more equitably target candidates so that we don't miss potentially great fits?", instead of top down: "how can we include more of these specific groups in our candidate pool?"

It's really wonderful to see schools like CMU lead the charge in this area, and I hope industry gives them the support encourage them to keep pushing forward. Programs like these will always be criticized as being "affirmative action" but I think in the long run, industries are better for them. Industry says we have a shortage of programmers, yet are content to maintain their focus on the sorts of people, overwhelmingly men, who started dabbling with it as teenagers. This is supremely short-sighted. It is in the interest of industry to bring people into the talent pipeline who might not have had the inclination, often for social or economic reasons, to spend their teenage years "hacking" but who demonstrate aptitude and could be excellent programmers if trained.

The real test is in how many decide to stick with it.

Edit: the comment thread on the site discusses this.

There were 2 women in my class, my first year at CMU. Really helped me to focus on my studies.

Anyway, it's hard to do better. I'm just glad that they are.

No offence. But I would be more interested in how they increased diversity without lowering the bar.

Maybe you should read the article, then.

It sounds like the only thing they did with admissions was to not look at previous programming experience as a major factor.

I was admitted to a Chemical Engineering college program, despite the fact that I had no previous high school experience in fluid mechanics, thermodynamics, and the rest. No one thought that was strange.

Why is CS so different that you have to be a hot-shot programmer before you enter?

Where is the hot-shot requirement?

> In the past, admissions gave preference to people with lots of previous programming experience.

Right. Lots of exerience doesn't equal "hot shot" though.

CMU is a top tier school. Why should you deny a hot-shot programmer admission for someone who has a passing fancy? Judging "potential" instead of a candidate's past work and experience reeks of affirmative action.

You are totally using the candidate's past work and experience to judge them. I'm sure those that get in have outstanding AP tests in math and science, great test scores, etc. A hot-short programmer would have that experience counted as well. The point is that programming is not the only type of experience they looked at. That's the same as it is for most other technical degrees (like my ChE).

And there isn't any preference based on gender, so this is nothing like affirmative action. It is just saying that bright technical people can learn programming, and don't need to come in as an expert.

Also, computer science is no more about computers than astronomy is about telescopes [1].

[1] http://en.wikiquote.org/wiki/Computer_science

They aren't giving them negative weight, they are just looking forward.

I know plenty of guys who went into uni with a bit of java experience (AP CS in highschool or something similar). So yeah, they know your basic control flow and OOP. But when it comes to how they will perform in N years, it means nothing. Any intelligent student without experience could (and did!) surpass the "more experienced" programmers by graduation time.

If someone is the next programming wiz of that year, I'm sure they will be viewed as having plenty of potential and get in.

If someone has a year or two of bad Java, they aren't viewing that as any better than someone with a good quantitative background. That's not affirmative action, it's foresight.

Anecdotes, not data, but I know plenty of people who learned Java, C#, or JS in highschool, some even did internships, yet someone who hadn't coded before 2nd year uni soared past their skill level within a year because he was a generally smart guy with good math and logic skills.

So should the people with a bit of experience be assumed to have more potential? No. Should somehow who's been coding their whole life and it's their number 1 hobby be assumed to have more potential? Yes. Not because of their experience, but because of their passion.

This. When you do something in the name of political correctness (and a completely economically (and risk free) biased one - i.e. almost nobody cares about the lack of female truck drivers) you still have to measure what are you inflicting damage upon.

- Is rejecting someone more fitted for this career path something good?

- Are you assuming everything is a societal result and there are no biological differences between the female/male psyche that makes them look for different hobbies and careers?

- It is know that women care more about socializing than men, and sitting in front of a screen for countless hours is pretty much the opposite of that; even when you work in a team, meetings are usually not something you look up to.

But we're not going to be able to measure those things until we try it and find out.

Also, many computer projects are done socially and aimed at socializing. Isolation is an old stereotype.

Not all women are extroverts? I'm a woman and could go without seeing people for a very long time. I think this is a huge misunderstanding of what it means to have a preference for the "social." It's about the purpose behind the project, not face time.

CMU is a top tier school. Why would you expect hot-shot programming experience to be relevant to their computer science program?

CS, much like art and music, is something that any sufficiently motivated individual will start doing far before they turn 18. (They differ from chemeng in this manner.) Both CMU art and CMU music require a portfolio review or audition.

http://admission.enrollment.cmu.edu/pages/art-required-portf...

http://www.music.cmu.edu/pages/audition-portfolio-review

Art and music are not unusual things to get into. But computer programming is often not presented as an option to anyone, especially women. So plenty of motivated and talented people might not even be introduced to it, or take it seriously.

Anyone with a Facebook account could have typed "learn to program" into Google rather than poking people in the farm while eating candy (or whatever people do on facebook these days).

If a person chose not to take programming seriously, that's on them.

So, women who get into CMU but disproportionately haven't been exposed to programming are in that situation because they were too busy playing Farmville?

I have no idea what they were doing - the point is that a Facebook account indicates the opportunity to use the internet. There has literally never been more information/tools available to potential programmers than there is today - free online coding schools, tutorials, stackoverflow, wikipedia, anything you want. Not to mention free compilers - when I was a kid, Borland C++ cost money.

I mention Facebook somewhat sarcastically because it's a service with no shortage of women. If women can be on Facebook, they can also be on Github if they choose to be.

There was never an easier time to learn programming, for free, while sitting at home, without even putting on pants, than today.

The secret of the patriarchy: http://bit.ly/w0iWlY

So what are people supposed to do, read up on every single topic on the internet? There was never an easier time to learn about alchemy, either.

The link goes to lmgtfy.com/?q=learn+programming

Not lowering the bar, changing the metrics and possibly even raising the bar.

I had some conversations with a couple professors there before I graduated in the mid-1990s, they saw it as a two fold problem. The geek culture makes it hard for anyone not within it regardless of gender, second they saw that MIT, Stanford, and some other top schools had more alumni donations and also tended to have more technology startups and such closely related to the school which also sort of affected the school's income but also affected future placement of current students which has a pretty large impact on getting the best students in the future. The thought at the time was that the geek culture was just bad for the school, period, they produced a lot of talented engineers but maybe not a lot of really successful people in the bigger picture.

I don't remember the exact numbers but I want to say something like over 50% of the SCS students switch majors out of SCS or left CMU, maybe even more. I want to say 250 students were pre-SCS and something like 50 graduated, but 250 seems way too high though. I can also say that I was a mid-tier geek, I had some experience and some chops but there were dudes who had been paid programmers when they started as freshmen, it was at times intimidating. I do personally know a half dozen folks that changed majors because they decided they couldn't hack it before things really got difficult; good smart people. It can be a very intense pressure cooker, say 50% leave SCS, of the 50% that stay, maybe 90% have had a fair amount of experience and then the 10% that don't struggle with tools and just basic things before they can even do the work. Life's not always fair but that environment was definitely self shaping. I have no doubt in my mind that some of the ones that switched majors could have made it and were maybe a bit psyched out.

When I heard this, in a way I felt a sense of pride, I had finished the "Real CMU" before it got watered down. I also felt a bit of loss, like I had spent the time and money to get their name on my resume and they were going to water its value down. As I've matured, started a family, etc.. I've grown to believe that CMU had (maybe still has) a fairly unhealthy emotional culture about it, I learned a lot, and I really don't have a lot of bad things to say about the place but I meet a bunch of damaged people there and some that might not be damaged so much as just not healthy emotionally. I'm not a mental health expert though, it's just my feeling, there are also some great people I meet there too. I'm in far more contact with people not of SCS though. I've worked at a number of startups, including a couple that were very much on the brink of death and I don't know that I've ever encountered any professional stress close to some of the times at CMU.

Reply on news.ycombinator.com