BBC admits its viral “women write better code” story was fake news | Hacker News Reader

BBC admits its viral “women write better code” story was fake news

hequal.wordpress.com

32 pointssakabaro15 comments

Open articleView on HN

Original article [0].

Here's the BBC's analysis of the complaint [1].

[0] http://www.bbc.com/news/technology-35559439

[1] http://www.bbc.co.uk/complaints/comp-reports/ecu/womenwriteb...

Here's the original HN discussion: https://news.ycombinator.com/item?id=11074587

The hequal article is pretty bad, and at least as biased as the BBC one, but they are at least right in their conclusion.

I haven't ready the study itself, so it might be the BBC that misunderstands things:

1. Using "pull request acceptance rate" as a proxy for "code quality" is highly dubious. It might equally well indicate that their PRs are more relevant, smaller and hence easier to review, more basic or any number of things.

Since we are looking at a ratio, the gap could also be in the numerator. Maybe men are just more likely to submit PRs, for a given skill level. Many people claim that female developers are being put down by their peers, so it would be natural if they are more hesitant to file a PR, only doing so when they are certain it's good. Etc.

2. The findings are based on a subset of total pull requests, namely those with users whose gender can be identified one way or another. Clearly the slight gender gap could just as well lie here, for example "skilled female developers are more likely than unskilled ones to be show their gender on Google+, compared to the ratio for men".

Also, since there is such an enormous difference between the absolute number of male and female developers (probably at least an order of magnitude), it's quite surprising that the gap here is only a couple of percentage points. Given that difference, it's easy to imagine lots of potential reasons for the small bias.

This site appears to be a mens rights site, pretending it's "egalitarian" when it's anything but. Just skimming both this article and the About page, it's pretty strongly insulting feminism and dismissing their complaints, while at the same time complaining about misandry and '"genuine" discrimination'.

Soooo this article challenges the original article's validity with stuff like:

> There are far more male users on GitHub then one could argue that men are in fact far better at coding because far more are actually doing it and many women are missing in action. Perhaps the small number of women who get involved are marginally better on average, but they would still be vastly outnumbered by men who are equally or more capable

How is this more valid? They complain that the study is not scientifically peer-reviewed (which is fair enough, I don't challenge that) and then come up with this dodgy analysis? Gee they sound stupid.

This could be valid with any number of probability distributions.

Comparing two populations (in the statistical sense) is tricky, especially if you have a huge difference between them. Then you can make some really weird claims, especially by omission.

What does "women write better code" mean? It could mean any number of things. Some I thought of:

* On average, a woman will write better code than a man. This is saying the mean of the "woman code quality" distribution is higher than that of the "man code quality" distribution. But again this metric is problematic without standard deviation. Because if there are simply more men than women who code, you still have a better chance at hiring a man who codes better than average if you can exclude the lower part through some bound (e.g. basic aptitude tests or interview), especially for long-tail distributions.

* Woman as a whole group write better code. I don't even know what this might mean, but maybe they aren't so obsessed about writing shitty javascript frameworks. Could be interesting for a psychologist. I just made a few things up for the sake of an example explanation: E.g. that the "women" population prefers to code on an existing project vs alone (i.e. new project), that that population prefers projects with a more social structure and therefore pull requests so through smoother because everybody know everyone, etc.

So you really can't say anything for sure without the data, but you can make up click-bait-y articles. Note how there are conveniently no graphs in the article, just a picture of a redheaded women looking at a laptop.

There's no such thing as an objective source. Advertisement, journalism, propaganda... they all blend together. It really shouldn't be shocking that so many are retreating into their own echo chambers. What's the point of trying to find a reputable source of news, when none of them are?

Because some of them are? And some of them retract their mistakes?

The post linked goes into depth as to why this retraction is too little too late. Having no filter process that prevents obvious click bait nonsense, with a retraction a year later is not worth praising. Best to internalize some skepticism about all "news" and question studies "especially of the gray science variety." Being bothered because it's easy to label outlets with biases you don't like, while forgiving those you do, is just human nature, and I'd wager at the root of why my comment keeps fluctuating wildly as to its rating.

Aside from the story, I really do think that women write better code.

I think it stems from society's tendency to be overly critical of women and the things that they do. If an external force is hypercritical of you, then most people would internalize it. I think this logically leads to better outcomes in things that require critique, like code.

Now this idea has much farther reaching (and more negative) consequences for women imo, but I think it applies here.

The issue is 'I think women write better code' to 'Women write better code, study suggests'. The latter suggests the author's own opinion without any proper research as they claim to be.

"Women write better code" seems like an ill-defined statement to me, even assuming you agree on "better" code. Does that mean that the average woman writes code better than the average man? Does it mean that the average "programmer" woman writes code better than the average "programmer" man? Is there anything in there about the the variance in female or male programmer ability? There are a huge number of metrics that would fit the statement, some of which may be true and some may not be true, and many of which might have different conclusions about how "actionable" that information is.

> Aside from the story, I really do think that women write better code.

It's your opinion, not backed up by any scientific facts, just like this fake article. It doesn't matter what you think, it doesn't make it a true statement.

> I really do think that women write better code.

Is such a view helpful though? I've found I really can't judge this is advance at all, regardless of gender. Or anything else really. So any generalisation like "x write better code" is going to disadvantage somebody. The best you can do is give everybody the same chance (and obviously you should be doing that already).

Fun anecdote: We used to look at degrees, e.g. was it CS or from a "good" university/college. But we crunched the numbers, and it turned out degree-specifics had almost 0 correlation with how good a candidate was. There'll always be one English major who writes kick-ass code. So now we use an online code tests instead to filter out applicants.

I don't think women are more or less capable than men when it comes to coding. There's no particular aspect of the process which could make it strongly gendered. However, I regularly observe a selection bias, which sometimes makes it seem like women _average_ high.

Women who are mediocre at coding have no particular motivation to go into the profession. A lot of the messaging about the career is how hostile it is to women. If you didn't have a burning passion and a strong aptitude for coding, why would you opt yourself into that? You'd have to have a serious skill and/or passion for the field to want to even start.

For men, though, that hurdle doesn't exist to the same degree. Passionate, skilled developers still follow their professional dreams, but there are also numerous men-folk who ended up in programming simply because it paid well, or because it was a nice indoorsy type job. Without that first cultural hurdle pushing them away from the career, they slouched their way through university and got a mediocre job as a mediocre developer.

There aren't as many professional, mediocre female developers, because only because only those who are really serious and passionate bother to push their way into industry. As we approach 50/50 balance, I'm confident we'll see a much more balanced mediocrity as well. :)

Reply on news.ycombinator.com