> To what extent does gender bias exist when pull requests are judged on GitHub?
As long as we ask gender-related questions, we'll find gender-related answers. Especially noteworthy:
> Perhaps only a few highly successful and prolific women, responsible for a substantial part of overall success, are skewing the results. To test this, ...
In other words, the researcher started with the hypothesis that women were discriminated against (as they say in the first part), then was surprised by the result and tried some more. Maybe, from time to time, it would be great to stop spurring the hate against the other gender: 1. Make sure everyone has access, 2. Analyze by some other criteria – It's not because gender is a readily available variable that studies should use it as something significant. Maybe shyness and marital status is more important, who knows.
Apart from not liking the idea of trying to find all little differences based on gender, the rest of the paper is interesting and in series of "Maybe women are X and Y? Here's how we'll check, here is the conclusion."
Last point, an equalitarist researcher would have studied the hypothesis "Is it possible that maintainers are simply kinder to women?". A lot of societal behavior could be explained this way, yet it's never a variable of those studies.
That's because the mantra in the media, and outside over and over is that women are discriminated against, and that is obviously the reason for X. I think in particular to Github or other FLOSS software development that it'd be less likely for women to be descriminated against. First, handles online don't necessarily correlate to a male/female persona. I know plenty of grown men who profess their affection for my little ponies. And women who like transformers.
Beyond this, I can't speak for anyone else, but short of having an obviously sexual image in one's user icon/graphic, I'm unlikely to even notice gender, and more likely to look at the code. I would think that not interacting via voice/video tends to do this, and numbs anything that might be gender bias in a more interactive setting.
I've said this for years... (anecdotal experience ahead) I once had a series of IRC discussions with someone about the online art scene (mid 90's) for several days... only ever found out the person in question was not only a woman, but featured in Playboy at one point (after it came out, she subsequently changed handles and reduced her time online). No idea. Being able to interact behind a computer screen is a great equalizer. You can't know for certain if Numb3rsFan12358 is a man or a woman.
>Being able to interact behind a computer screen is a great equalizer. You can't know for certain if Numb3rsFan12358 is a man or a woman.
That's one of the things that most vexes me about that push to removing online anonymity. Ostensibly, making everyone interact on the internet under their real name will "prevent harassment and bullying". But I have no idea how anyone with even passing familiarity with internet communities could believe that (cough Wil Wheaton cough).
>Ostensibly, making everyone interact on the internet under their real name will "prevent harassment and bullying". But I have no idea how anyone with even passing familiarity with internet communities could believe that
The idea is that anonymous people 'harass' the over-privileged so-called celebrities on Twitter.
Of course if everyone was anonymous then there would be nobody to harass, and if everyone was pseudonymous then at least anyone would be a name-change away from harassment.
Personally I think that one interesting thought experiment would be that everyone sees everyone else as a different name. So we all see totally randomly generated strings as each others' names. If you block someone then they can't just make a new account and go to your profile, because the new account can't see BlisteringOakPlatypus (assuming some sort of gfycat-like scheme), as that name is only my randomly generated identity from the perspective of their old account.
You also can't bandwagon on people: "hey, go harass DribblingPharmacologicalHippopotamus" wouldn't have any effect.
How would you connect with people you know? You could generate a one-use code they could put into their account to find your profile and communicate with you. Or you could generate a code that lets them just view your profile, without communicating with you. Or for either of these, a many-use code. This lets you be a 'broadcaster' like a celebrity on Twitter, but just because you broadcast doesn't mean you have to receive messages back.
How would you find new people? Well on a Facebook-like platform, through friends-of-friends. What happens if you block someone and they reconnect to one of your friends and can then see you again? The same way you deal with encountering people you don't like IRL through friends of friends. You just have to deal with it like a human.
On a Twitter-like platform, you just follow particular tags or whatever, hashtags, etc.
On a bulletin board, you add people that you see make points you like/appreciate/find to be helpful.
Indeed - to take their conclusions to an extreme, they seem to be saying women's rejected pull requests are due to bias, whereas men's rejected pull requests are indicative of competence (or lack of it), and that women's accepted pull requests are due to competence, whereas men's accepted pull requests are due to bias. Definitely seems to be attempting to make the facts fit the argument rather than the other way round.
tl;dr is that Github is fair to men and women submitting pull requests. (Women have a very slightly higher chance of getting their pull requests merged for who knows what reason.)
I've witnessed plenty of gender bias in IT (and my comment history goes into quite a bit of detail in past discussions), but I can't say this article seems to be indicative of it to me.
It appears that when you account for the majority of situations, women actually have a higher acceptance rate then men for pull requests, but in a rather subjective measurement ("identifiably gendered profiles" for "outsiders"), the trend reverses. I don't know that there's necessarily a lot of evidence that maintainers are going around looking for things that identify genders in a profile before they accept a pull request, and if that single measurement is indicative of bias against women in one specific situation, it seems they have even stronger evidence of bias against men in other situations.
There seems to be questionable correlation that requires belief that a unfalsifiable and subjective judgment is occurring. There's no hard proof that this correlation exists to begin with, much less proof of it being causal in nature.
There's plenty of real gender bias and discrimination going on. Let's try to focus on that.
As long as we ask gender-related questions, we'll find gender-related answers. Especially noteworthy:
> Perhaps only a few highly successful and prolific women, responsible for a substantial part of overall success, are skewing the results. To test this, ...
In other words, the researcher started with the hypothesis that women were discriminated against (as they say in the first part), then was surprised by the result and tried some more. Maybe, from time to time, it would be great to stop spurring the hate against the other gender: 1. Make sure everyone has access, 2. Analyze by some other criteria – It's not because gender is a readily available variable that studies should use it as something significant. Maybe shyness and marital status is more important, who knows.
Apart from not liking the idea of trying to find all little differences based on gender, the rest of the paper is interesting and in series of "Maybe women are X and Y? Here's how we'll check, here is the conclusion."
Last point, an equalitarist researcher would have studied the hypothesis "Is it possible that maintainers are simply kinder to women?". A lot of societal behavior could be explained this way, yet it's never a variable of those studies.