Testing for gender differences in Python programming style and quality on GitHub

https://academic.oup.com/jcmc/article/29/1/zmad049/7596747

by bomewish • 2 years ago

35 91 2 years ago

academic.oup.com

gnabgib• 2 years ago

Article title: Programmed differently? Testing for gender differences in Python programming style and quality on GitHub

bissps• 2 years ago

First, I infer the gender of users from their usernames and the information provided on their profiles, labeling users as feminine, masculine, ambiguous, and anonymous.

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.

TMWNN• 2 years ago

many

cjbgkagh4• 2 years ago

While transgenderism is by no means common, it is unusually frequent in programmers, and by my observation even more so in more productive programmers who are more likely to be selected if selecting people from code written. The ‘programmer socks’ meme does appear to have some merit.

shiroiushi4• 2 years ago

Similiarly, there are many males who present with a feminine username and profile for various reasons, such as a desire to be female.

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.

defrost• 2 years ago

The premise is that "Women code differently, but no worse .."

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.

Supermancho• 2 years ago

such as a desire to be female.

...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.

donatj• 2 years ago

TL;DR They gendered people based on name stereotypes and ran Pylint against their code.

Methodology seems questionable.

By their own definitions earlier in the article, it seems like Pylint would provide biased results.

desert_rue• 2 years ago

Yeah, style != quality

taeric6• 2 years ago

While I generally agree, I'd also hesitate to wager against them being correlated.

jaza• 2 years ago

Over the years my coder colleagues have been about 95% male and 5% female. Of them, I rate about 80% of the males as producers of quality code, and about 95% of the females as such. That's just my experience and my subjective scores.

mikepurvis• 2 years ago

Setting aside the questionable methodology, I expect a lot of gigs are far more accepting of mediocrity from a man than a woman, so women doing software development are much more likely to self-select out if they're not up for it— whether in actuality or due to imposter syndrome.

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.

fallingsquirrel• 2 years ago

The minimum number of colleagues you could have had to arrive at those ratios is 400. Assuming you've had 400 colleagues, we can break down your ratios into:

- 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.

danielmarkbruce• 2 years ago

If you are going to suggest someone has effectively said something meaningless, why not break it down properly and calculate the error bars?

I'll be just as lazy...4x the number of male duds across that number of observations? It's going to be statistically significant.

consumer4512• 2 years ago

For historical reference:

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!

[1] https://en.wikipedia.org/wiki/Women_in_computing

callingbull2• 2 years ago

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.

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.

consumer4512• 2 years ago

I am just trying to understand the history. Let me start with one point.

> 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...

TulliusCicero• 2 years ago

The underrepresentation of women in open-source software is frequently attributed to women’s lack of innate aptitude compared to men

???

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.

anon291• 2 years ago

I guess the thing for me is that my one open source project of note was an accident and from what I gather most male open source founders are similar. For example, Linux is notorious for being Linus's hobby project that hit it big.

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.

joshdavham• 2 years ago

The underrepresentation of women in open-source software is frequently attributed to women’s lack of innate aptitude compared to men: natural gender differences in technical ability

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!

stonethrowaway4• 2 years ago

One follows from the other.

MeetingsBrowser4• 2 years ago

Could also be that women are less interested in an industry dominated by people who assume women are inferior

benatkin• 2 years ago

Often it's the opposite. When you aren't very good at something, you have to work twice as hard. I've met plenty of men who thought that programming was cool so they became obsessed with learning how to code without having the innate skills. It mostly doesn't work out, but during that period of time, they were more interested in learning to code than a lot of programmers. And many do try open source. That's why there are a lot of junk pull requests on Hacktoberfest.

Dan42• 2 years ago

I'm 47 and I also have never heard this in my life. So I don't think it's just a matter of younger generations.

latexr4• 2 years ago

It wasn’t too long ago that very topic was all around tech news.

https://en.wikipedia.org/wiki/Google%27s_Ideological_Echo_Ch...

latexr4• 2 years ago

That assumption is also wrong, and while less insulting on a surface level, I’m not sure it’s any less harmful. Women are interested in all STEM fields but have historically not been treated as capable because they’re women. We have tons of data on that, including experiments where using a feminine or masculine name changes how contributions are received.

https://www.independent.co.uk/news/business/news/gender-ineq...

tzs• 2 years ago

If we change that from "women just generally aren't interested in software development at the same rate that men are" to "women aren't as interested in software development as a career at the same rate as men" then it might not be so wrong.

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.

zahlman8• 2 years ago

That assumption is also wrong... Women are interested in all STEM fields

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).

azaras• 2 years ago

You are right; it is about interest; women like social sciences, and men like STEMs.

Is this a product of nurture or culture? I don't know.

bitcharmer• 2 years ago

You are absolutely right. Most men don't want to be a nurse just like most women won't prefer to be a car mechanic.

HN is in complete denial of this simple fact. It's hilarious how seemingly intelligent people can be so badly influenced by identity politics.

rsynnott2• 2 years ago

This would also require there have to been some enormous _shift_ in these 'innate aptitudes' in the 1980s or so; there were lots of female programmers in the 50s and 60s.

But yeah, I'm 39 and have never heard anyone serious claim that the gender imbalance is due to innate ability.

zahlman8• 2 years ago

Back then, the title of "programmer" tended not to describe the person who came up with the logic, but rather the person who punched the holes in the key cards.

legostormtroopr• 2 years ago

From the first line of the abstract: "The underrepresentation of women in open-source software is frequently attributed to women’s lack of innate aptitude compared to men".

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.

xvector• 2 years ago

TL;DR - guys on average seem to have more lint errors

taeric6• 2 years ago

I confess I find it surprising that they code differently. That is, my prior would be that there is no real difference based solely on sex. Their may be differences based on education. But why sex?

stonethrowaway4• 2 years ago

Because it makes for excellent headlines and advances the author’s career.

Brian_K_White3• 2 years ago

Why not sex? You only have to meet a few men and a few women to observe different preferrences for different strategies to the same problems. Even different perception of problems at all in the first place.

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.

taeric6• 2 years ago

You only have to meet a few people to know that you will find different preferences period. And I know plenty of men that prefer the same strategies as plenty of women that I know. Such that, no, I don't see any real differences in thought work between the two. Same reason I don't think they do math differently. Reading is probably more similar in technique than not. Even if choice of reading topics is wildly different.

I don't think it is impossible that there are differences. I just don't have it as my prior.

neilv• 2 years ago

These metrics are mostly superficial.

stonethrowaway4• 2 years ago

Because the author isn’t a bona fide dev and couldn’t tell shit from shinola wrt code quality. Like a lot of other research it’s nonsensical.

stonethrowaway4• 2 years ago

Technical ability and programming are orthogonal.

dzink• 2 years ago

Territoriality in code has been a very male pattern that seems bizarre, given the type of work we do. In purely anecdotal experience: A small team hires its first female developer in a team of several. The senior developer on the team recommends against hiring her, but management prevails. In their separate projects everyone in the team is a top performer - they finish the work for a fraction of the estimated time, the female developer knows the languages they work with so well she can identify bugs and future browser quirks even when reading code on paper. Everyone is detail oriented and keep their code meticulous. The team clocks extra hours regularly without complaint.

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.

bomewish• 2 years ago

I see two issues with the paper.

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.

sincerecook• 2 years ago

In my experience the average female coder is significantly worse than the average male coder, but the ones that are good, are usually very good. It just seems to me that males have a greater inclination towards logically structured thinking.

Testing for gender differences in Python programming style and quality on GitHub

Related Stories