Testing for gender differences in Python programming style and quality on GitHub
academic.oup.com
academic.oup.com
Maybe I'm just of a younger generation, but I've literally never heard this in my life. The assumption among people my age has always been that women just generally aren't interested in software development at the same rate that men are; not that they're naturally worse at it!
The comment that started this thread did no such thing. In fact, the entire point of that comment was to argue that this is itself an unfair stereotype of the community (i.e. that such a belief is not prevalent).
It instead quoted a claim from the article - which indeed implies this in a vague, hand-waving way, by citing another paper. When I look at the abstract for that paper, in turn, I don't actually see the claim asserted; the closest it comes is
> Challenges that women face in OSS are mainly social, including lack of peer parity and non-inclusive communication from a toxic culture
Also, that paper in turn describes itself as "a survey of the literature".
The whole thing comes across to me as an act of citogenesis (https://xkcd.com/978/).
> One follows from the other
There are two ways to interpret this.
Either women are naturally worse programmers and that leads to less interest, or they are less interested in programming which makes women worse programmers.
Unless I am missing another interpretation, both imply women are worse programmers.
Second, maybe you got into it as a solitary activity as a child (I did too, long ago), but unless you work at a very, very unusual job, it's not at all a solitary activity these days at any normal company.
My workplace even had a sexual assault in the workplace training course where they coached us to stop co-workers from speaking if they started to disclose being victimized at work, I assume to avoid being legally responsible to respond appropriately.
That happens about as much in any field as long as you mix genders, so that is no reason to not want a software job.
Maybe it's just the company I'm at but be aware this was also the case in college. It was hard to make friends when it seemed everyone was cosplaying an anime character while simultaneously bragging and trying to make me feel bad for my achievements, or lack thereof.
to anyone.
Keep in mind also the media they might consume. I mentioned anime cosplaying in real life - and I mean it. Now, look at how anime depicts women. They're sexual objects for men's enjoyment and entertainment.
But perhaps the biggest reason is men tend to get... overly confident in male-dominated spaces. This is something I've noticed since I was a child. All bets are off in locker rooms or at the poker table or in an online game. It's an opportunity to say the vilest things imaginable.
Yep, this is one big reason I hate being in all-male environments or workplaces. Men are usually much better behaved when there's some women around.
Men also experience more social pressure to be successful at work, or else are seen as not as worthy of respect as people or as romantic partners regardless of their personal goals and values.
https://en.wikipedia.org/wiki/Google%27s_Ideological_Echo_Ch...
https://www.independent.co.uk/news/business/news/gender-ineq...
I remember reading about a large study of pre-college students in OECD countries that found that while the top students in STEM fields were about equally split between boys and girls, for the boys it was more likely that they were only top in a STEM field. The girls were more likely to be top students in multiple fields.
Students tend to go on to the field they are best in. For the top boy STEM students that tends to be STEM. For the top girl STEM students the field they are best in often turns out to be one of the non-STEM fields they are also top in.
The research contradicts you (https://en.wikipedia.org/wiki/Gender-equality_paradox).
>We have tons of data on that, including experiments where using a feminine or masculine name changes how contributions are received.
Your link is to reporting of a viral anecdote shared by an editor for a movie review site. It's also about how clients responded, not coworkers.
Aside from the "experiment" being N=1, easily confounded by any number of factors (such as the simple fact that, from the client's perspective, a supposedly troublesome contact was being replaced at all), and at least partially an excuse for Mr. Schneider to complain publicly about his boss, all of this clearly has nothing to do with "STEM", let alone programming. Historically it has been common in programming circles to receive contributions under pseudonyms (or, for that matter, first-initial-plus-last-name usernames) that don't disclose gender.
The argument you present here is simply not intellectually honest (and I have seen very similar arguments countless times over the years).
It takes a special kind of gall to accuse another of intellectual dishonesty by using a Wikipedia article as proof and in under thirty minutes in the same thread say that Wikipedia cannot be trusted.
https://news.ycombinator.com/item?id=42011586
Hard to find a better example of intellectual dishonesty than using a source as proof when it supports your argument and decrying it as unreliable when it doesn’t.
Furthermore, the article I linked to is clearly distinct along the lines that I described in the other post: it's not trying to paint a historical narrative, but to document a phenomenon noted by researchers.
I had thought this was a place where people don't stalk other users to try to attack them personally. (Also, your attempt to try to identify some kind of hypocrisy is ignoring the actual critique I offered of your own source.)
My initial response was different, and I even thanked you for the correction, but I deleted it and posted a new one after suspecting (as per the above) that you might not be arguing in good faith.
Please don’t try to play the victim when you’re the one who initiated the conversation with personal attacks.
Is this a product of nurture or culture? I don't know.
HN is in complete denial of this simple fact. It's hilarious how seemingly intelligent people can be so badly influenced by identity politics.
But yeah, I'm 39 and have never heard anyone serious claim that the gender imbalance is due to innate ability.
This isn't going to give accurate results. It's common for those who are female to conceal this fact, with neutral usernames and profiles, to avoid harassment from predatory males.
Similiarly, there are many males who present with a feminine username and profile for various reasons, such as a desire to be female.
It is likely that this is more than anecdata.
And trans are also over represented because of this connection.
>There is an elevated co-occurrence of autism in trans individuals, with recent meta-analyses suggesting that 11% of trans individuals are autistic. The presence of autism in trans young people can create clinical challenges by adding complexity to the presentation, assessment and management of those presenting to gender clinics. Although many trans young people display traits of autism, how these traits relate to the nature of their gender diversity is unclear.
https://sites.asee.org/se/wp-content/uploads/sites/56/2021/0...
>Despite the study’s limitations, the initial pilot data gives reason for additional research to be conducted. Similar to Wei, Christiano, Yu, Blackorby, Shattuck, and Newman’s 9 findings, this study also supports the commonly accepted assumption that individuals with ASD are very likely to major in Computer Science.
--- And to clarify, I did not vet these, it was a very lazy google search.
If you see lots of trans people where you work, they may very well be hiring or at least referring each other. I've seen that phenomenon with gay people too. Once in a while you'll run across a business with a lot of one kind of person.
My cohort is the category theorists / functional programming crowd which indeed has a higher percentage of trans people in it than even other programming cohorts. But this is not due to a hiring ring, they lack the numbers.
As far as work quality, I think assuming trans people are better in some way is the same kind of fallacy as assuming autistic people are geniuses like Rain Man. Most of them probably aren't.
By my observations geniuses are very likely to be on the spectrum. At least 50%.
I really don’t know what the normal programmer experience is like, I exited the labor market before the DEI stuff really got traction. My rule of thumb is, of all the code that I’ve seen that I’ve been impressed by, it’s far more likely to be written by a transwoman than a non-trans woman by a significant margin. A disappointing observation for a straight male who had aspirations of dating a hacker-girl. Additionally the more impressive the code the higher percentage likelihood of transgenderism, I do think high IQ, autism spectrum / ADHD, and transgenderism are rather correlated.
No, I'm not.
>By my observations geniuses are very likely to be on the spectrum. At least 50%.
The problem is, you are disproportionately likely to notice someone being odd and then incidentally being a genius, than someone normal who happens to be a genius. There is also an issue with the "continuum" nature of the "spectrum" that makes people assume that any bookish person with lame social skills has a neurological affectation instead of just a lack of social experience.
>My rule of thumb is, of all the code that I’ve seen that I’ve been impressed by, it’s far more likely to be written by a transwoman than a non-trans woman by a significant margin.
So you're really saying, the code is more likely to be written by a biological man than a biological woman. That is statistically true. Most women have better things to do than slave away on some obnoxious code.
>A disappointing observation for a straight male who had aspirations of dating a hacker-girl.
I was lucky enough to date a cute engineer girl once. Unfortunately it didn't work out. It's like trying to date someone with 10x more options than a normal woman, who has like 10x more options than you by default. I could probably accuse her of being on the spectrum because of how she ended the relationship, but I think she just doesn't have enough common sense to accept the few minor flaws I have. Then again, idk who else she had waiting to date her in the background. It could be as simple as, she met someone with more money. She was rich and I wasn't, and that was a problem. As most women require a partner who makes more money than they do, having a girl who does the same job as you can be kind of a problem.
>Additionally the more impressive the code the higher percentage likelihood of transgenderism, I do think high IQ, autism spectrum / ADHD, and transgenderism are rather correlated.
I think all of those things are correlated because of social media, except for IQ. But at this point we are just speculating.
Again you're not understanding conditional probability, the code I've been impressed by has a strong selection criteria bias as I'm only impressed by a very very tiny percentage of code that I've read, additionally I don't read every line of code written. This selection criteria is so strong that the fact that men in total write more code than women is almost completely irrelevant.
A large part of science is dedicated to not finding obvious things out about IQ which is increasingly hard to do as statistical tooling and data has become widespread. Most academics are stuck with old data and tooling and a existential requirement to find only politically correct results. If you think DEI is bad then you should see the anti-racists pledges that must be made by academics for access to data. My findings are from a mountain of data and sophisticated tooling but unfortunately I can't make appeals to authority when posting anonymously.
If you could actually get accurate data identifying which programmers are actually male or female, and which male programmers desire to be female, I wonder if you'd see a quantifiable difference in their code compared to regular males.
ie. Stylistic differences likely arising from in-group dynamics, cross reading, pair programming, etc.
As such there's likely no implicity { transgender, male, female } style, more various styles that are more pronounced in groups that cluster, be that male programmers, female coders, transgender hackers, google devs, etc.
...or more often to capitalize on the short term, early benefits (attention, leniency, opportunity, etc) often offered to women. This (males using a female name + character) is a common practice in multiplayer games.
Perhaps I am obtuse, but I've been in FOSS for 15+ years and I have never heard this as a common topic. When it is infrequently mentioned, its almost always shouted down quite quickly in the FOSS community as sexist nonsense.
The first is that the measure of quality doesn't actually capture quality. It's like measuring the quality of a writer by looking at only their grammar and spelling.
The second is that it implicitly makes causal claims, but the design makes it impossible to infer any causal relationship between sex and code.
Curious what others' take is on just the empirics of the paper itself.
The senior developer is made manager on the team. He begins to require all javascript code to have C-style curly brackets. (the opening bracket goes below the function declaration, not right next to it). The reason for it: he is the only one on the team with a C background and prefers to read everyone’s code that way. Overnight he goes into all repositories for multiple customer projects that have nothing to do with each other and changes the code to his preference. The next day the individual developers who work on each project, lose hours sifting between meaningful changes and style preferences and adjust their code. The next night he changes everyone’s code again. Control and desire for dominance over the code base when the code is essentially multiple consulting projects for multiple different customers done by multiple developers at the same tech consulting firm. His treatment and outbursts were worst against the female developer. The female developer among others on the team left for more senior roles elsewhere soon after.
The female software engineers I have worked with have been diligent and thoughtful and often more skilled than male peers, but the treatment of them has always been different in one way or another. I am not surprised if most hide the fact that they are female in open source code contributions. One of the perks of crafting software is you don’t need permission to make stuff, and you certainly don’t need to tolerate bad behavior distracting from the work. The easiest way for teams to lose female developers is to let an insecure bully chase them out.
You can't imagine how that can result in different code?
It's the least remarkable claim imaginable. Though I'm not sure this study quite proves it.
I don't think it is impossible that there are differences. I just don't have it as my prior.
There IS a pattern of group similarities. It's not some prejudice or discrimination or anything bad, it's just a simple fact that makes no sense to pretend to be blind to.
The same is also true for any number of other common properties like culture and language and age. But the question was not about that. the question was "I'm totally boggled, why would sex change anything about how people tend to code?"
I don't rule out the possibility. But I do place a low likelihood on it.
Also, I was not claiming any nefarious reasoning here. You introduced that verbiage.
There are an infinite number of ways to solve any problem, be it in the world or in code.
A common stereotype example (whether it's true or not, it still serves as an example of a difference in approach to a problem) would be how a male would more often choose to deal with some problem directly while a female will more often find some indirect way. Some people might say that as honest vs manipulative, or simple vs smart depending on their own bias and which approach they want to portray as more admirable.
I say the "manipulator" (female) way has a lot in common with what I as a male admire in the most "elegant" engineering solutions. Getting something to happen by itself from a small input vs a more direct brute force less flexible way. It can be bad, because one form of this indirect action is to manipulate people, but that is just one expression of it. The same thing also just means seeking and being willing to employ compromises and bargaining, being willing to give an complete asshole what they want to get what you need and not caring so much about your own pride or whatever. I myself am far less likely to do that. I judge and I don't compromise or bargain unless forced, and I rationalize it as not necessarily toxic because I don't think it makes the world a better place to let people get away with some things, I think it's important and worth it in the long run to absorb some difficulty to push back, precisely because many won't and someone should.
But intellectually I can recognize that there are these two different ways to attack the same problem, and they both get results, and so they are both valid, including the approach I myself don't usually choose.
Neither is better or worse or has a better or worse value necessarily. They are different ways to successfully get from the same A starting point to the same B ending point. But they are different.
In code there is practical infinity. Like preferring functions vs objects, implicit vs explicit, imperative vs declarative... countless different paths to arrive at a desired final result.
And they are not all equally random. Males and females absolutely have bell curves of how they are more or less likely to interpret some data or problem or process, and how they go about breaking the job up to get from a to b.
Other things like culture would have it's effect too, but that's just it's own seperate thing. We have studied ourselves thoroughly enough for long enough that even with all the bad and biased studies, we still just have overwhelming and ridiculously obvious data that there is a pattern of difference attached to sex regardless of all other factors.
None of the rest of what you mention here changes my priors, though. I expect that some differences could be seen between different sexes in many things. Outside of childbirth and physical strength, I fully expect most would be driven by culture far more than any brain difference.
You obviously have different priors. Seeming to expect that there will be differences in thought work.
And you know what? That is fine. I'm surprised by your priors here, yes. You are obviously surprised by my priors, as well. That is fine.
So, to that end, I'll note I was not trying to say this shouldn't be studied. I do not dismiss that it could be, out of hand. I do fully expect that it is not the case. But it would be far from the first time my expectations were irrelevant.
???
Yeah there's still some sexist assholes around like this, but this is hardly a mainstream opinion in most circles imo.
I'd say a more common explanation these days is lack of interest, rather than aptitude.
So while I can see why a toxic community would stop women from contributing to an existing project, I think the 'starting' a project would be more interesting to look at. Most open source projects start by accident not intent. One would expect women to not have the same probability as men of a project hitting it big presuming they make the code public.
> The first algorithm intended to be executed by a computer was designed by Ada Lovelace who was a pioneer in the field. Grace Hopper was the first person to design a compiler for a programming language. [0] Throughout the 19th and early 20th century, and up to World War II, programming was predominantly done by women; significant examples include the Harvard Computers, codebreaking at Bletchley Park and engineering at NASA. After the 1960s, the computing work that had been dominated by women evolved into modern software, and the importance of women decreased. [1]
What a crazy history, and wiki entry. My supposition is that as soon as it started to become even slightly obvious that software was going to eat the world, the dominant social paradigm prevailed.
[0] WTF, how is she not way more celebrated? She invented the effing compiler!
Babbage designed the first programs for the analytical engine.
> Grace Hopper was the first person to design a compiler for a programming language.
Hopper was the first person to use the word "compiler" for a program, but it was not a compiler by the modern meaning; it was a linker, and not the first.
> Throughout the 19th and early 20th century, and up to World War II, programming was predominantly done by women;
Not by the modern meaning; they manually entered programs designed by men.
> After the 1960s, the computing work that had been dominated by women evolved into modern software, and the importance of women decreased.
The design of programs has always been done by men mostly.
>> The first algorithm intended to be executed by a computer was designed by Ada Lovelace who was a pioneer in the field.
> Babbage designed the first programs for the Analytical engine.
I am trying to understand the wiki entry in my gp vs. this post [0]
Is the wiki entry entirely wrong?
[0] https://blogs.bodleian.ox.ac.uk/adalovelace/2018/07/26/ada-l...
Yes. The reference it cites contradicts it: "The first algorithm intended to be executed by a computer was designed by Ada Lovelace" with a link to https://www.cs.virginia.edu/~robins/Ada_and_the_First_Comput... which says "Many people, for instance, incorrectly claim that Ada was the first computer programmer. (Babbage, not Ada, wrote the first programs for his Analytical Engine, although most were never published.)"
But, it stands to reason that Babbage had some problems he sent through his machine. I would wager anyone that had access to it tried their hand some.
It is a Wikipedia article on a topic with clear "culture war" significance. The topic is not an objective individual person, event, place etc., but a narrative about events unfolding over a considerable time span, which concerns itself with a subset of the population and their interaction with technology and thus society.
Yes, of course it's entirely wrong. Wikipedia cannot be trusted for this sort of thing. It is a tertiary source by design (https://en.wikipedia.org/wiki/Wikipedia:Wikipedia_is_a_terti...), which entails that the most obsessed editors can cherry-pick what they like from already biased secondary sources (the ones considered "reliable"; the reputations of which may not be tarnished by "original research" or by conclusive proof coming from outside the list).
Methodology seems questionable.
By their own definitions earlier in the article, it seems like Pylint would provide biased results.
In any case, I would generally agree with your take but express it a bit differently— the group of programmers who really blow me away with their ingenuity and commitment to excellence has significantly more women in it than the larger population ratio suggests should be expected.
- 380 men
- 19 women producing good code
- 1 woman producing bad code
I think the error bars on that one woman would be quite large, too large to draw any conclusions.
Now if I guessed wrong and you've actually had not 400, but tens of thousands of colleagues who you've been secretly tracking and ranking in your spare time, maybe you'd have enough data to draw a conclusion.
I'll be just as lazy...4x the number of male duds across that number of observations? It's going to be statistically significant.