December 1, 2011

...Learn TDD with Codemanship

Women In Computing - The Potential Negative Effects Of Positive Discrimination

This is a potentially touchy subject, and I may well be ill-advised to even go there, so please bear with me as I try to navigate through it.

I'm one of many men working in computing who'd be very happy to see more women working alongside us.

Nobody seems quite sure what we can do to encourage more young women into the field, and there's much evidence to suggest that the damage is done - certainly in terms of girls seeing programming as "boy's stuff" - by a very early age. Which means that, whatever we do, the target of our efforts needs to be girls under the age of 7 so we can maybe change their minds before Polly Pocket changes their minds for us.

One thing I'm pretty convinced won't help is positive discrimination for women in computing higher education and the work place. And I know many women who have achieved great success in software who feel the same way - indeed, many feel somewhat insulted, as such talk belittles their very real (and very un-positively discriminated in favour of) achievements. Indeed, all of us who've learned our SOLID design principles know that the "L" is female.

The statistics of it makes it unworkable. It's also highly unethical and very unfair both to the men who get discriminated against - recruitment and promotion is a zero-sum game, remember - and the women who succeed because they're good at it.

The current ratio of men to women in software development, for example, is roughly 10:1. That ratio is baked in right from the start. It's not that women find it 10x harder to get into computing. It's simply that 10x less even want to.

if I'm hiring a team of developers, I may have to interview ten people for every person I would actually hire. (And I'm being generous here.) So a team of 4 might require me to see 40 candidates, regardless of gender. It can take weeks - months sometimes - to build even a small team.

What happens if I resolve that 2 of those team members must be female? Finding 20 females candidates will be as hard as finding 200 males candidates, because there are 10x less. My options are to either spend a lot more time looking - and the clock's ticking, folks, because this is business - or to lower the bar so that it's 10x as easy for a female candidate to get the job as her male counterpart.

And here's where it matters, for me. Not only is this grossly unfair on male candidates, and on the business that will now suffer the consequences of hiring substandard developers after massively lowering the bar, it's also very unfair on the good female software developers out there.

I've seen with my own two eyes that women and men are roughly as good at writing code. That is to say, the proportion of good programmers to not-so-great programmers is gender-independent. Let's continue with our model of 10% good and 90% not-so-good. If we lower the bar for women by a factor of 10 to recruit the team without taking significantly longer, we end up interview an average of 1 female candidate for 1 female role. In other words, we hire the first woman we see. That's a bit too much of a lottery for my liking.

What that means is that, as a woman, you'll be surrounded by other women who come from the not-so-good pile. Who does this help?

The same goes for promotion. If you make it actively easier for women to progress at work, you'll end up with a lot of very disgruntled men and an upper tier of tech management that comes from the not-so-good pile. Now who does that help?

It seems to me, from basic statistics and personal experience, that positive discrimination would do nobody any favours - least of all women. Widen the goalposts and what you end up with is a bad reputation for women in computing, and a whole heap of pain for men who've done nothing wrong, but who now have to work 10x as hard to make the same progress in their careers.

Now, I've taken a pretty extreme ratio for the purposes of illustration. But whatever ratio you apply, if you require a greater proportion of women being hired and promoted than actually enter computing, you'll get to some degree the same effect.

The right model is to encourage more women into computing in the first place. And the time to address that is very early in their educational and personal development.

ADDENDUM: female friend has just pointed out that if she was from the not-so-good pile, positive discrimination would be very mch in her favour. Worth thinking about.

Posted 9 years, 2 months ago on December 1, 2011