The Unit of Caring

This article is an instance of a way of talking about women in computer science which annoys me a lot.

Some people did an analysis of internal code submissions at Facebook and found that women’s contributions were rejected more than men’s contributions. Some other people promptly did a re-analysis and found that the effect vanished entirely when controlling for experience - so, what is actually going on is that more of Facebook’s women are junior engineers, and their contributions are rejected at the same rate of all the other junior engineers, and a smaller share of Facebook’s women are senior engineers, and their contributions are rejected at the same rate as other senior engineers.

But, says the article, we would be wrong to interpret this as we might be tempted to interpret it, ‘there does not seem to be bias in the code review’, because we should take the opportunity to be reminded that female perspectives are important and also presumably the reason women are likelier to be junior employees is discrimination in promotions.

I think this completely misses the point.

If I am a woman considering working at Facebook, I want to know whether my code will be reviewed more harshly than my coworkers. I want to know whether I have to be twice as good to get half as far, or for that matter 103% as good to get just as far. The original study was interesting because if it reflected that other things equal women’s code got rejected more, then it would be informative for people who might want to work at Facebook (about the kind of work environment they’d get) and for people doing code reviews everywhere (about maybe whether those should be blinded somehow because gender was affecting reviewing). 

But since there’s no evidence that other things equal women’s code gets rejected more, blinding code review won’t help and this isn’t evidence that women at Facebook are being held to higher standards (they still might be, but these statistics have nothing to say about it.) That matters. A lot. Instead, all this seems to be is evidence that, yet again, there are way less women in the field to start with. At my brother’s middle school, where every student who wanted programming got to take it, there were thirty boys and two girls interested. We know we’ve got a problem there; people are trying from all different angles to fix it. 

And when someone equivocates between these two classes of problems as ‘both related to sexism probably’, I feel like they are seriously failing to think about or empathize with the perspective of actual women in CS, to whom ‘at Facebook you have to be better at coding to pass review’ and ‘at Facebook your code will be reviewed purely on its merits, but because of complex societal and historical forces women are likelier to be in junior roles’ is a huge difference and a hugely significant one. It’s insulting and frustrating to see these wildly different claims treated as pretty much the same thing.

• 19 May 2017
• #gender in CS
• 178