Hacker News Reader

Comment by haberman | Hacker News Reader

In general it's very hard to control for all factors. The most compelling analysis with robust controls that I have seen is: http://www.nber.org/papers/w5903

The paper shows that adding a screen to orchestra auditions significantly increased the number of women who were chosen for symphony positions. The study is very compelling because the cause/effect relationship of gender-blinding the auditions can be observed very directly.

A musical audition is somewhat unique in that you can conduct the "interview" in a way that the "interviewer" remains blind to gender/race/etc.

View on HN

One problem with this is that orchestral music is a very traditional, opinionated industry. There could be biases specific to that setting that don't show up in others.

Are you suggesting that the software tech industry isn't in its own way at least as traditional and opinionated?

The software industry is at most, what? 100 years old? 55 years old if you want to start from the beginning of the rise of business computing? It's an industry created in the midst of massive social changes that has been chasing moving targets from the beginning.

Orchestral music (as we know it, an argument could be made for orchestra-like organizations going back another few millennia) has been around since the 16th century or so. That is a much longer time for exclusive traditions to develop.

I'm sure they're both traditional and opinionated, but I would expect that the scale at which they manifest them will be very different.

Well it wouldn't be that outlandish to do truly blind programming interviews.

Just use a collaborative editor and possibly wacom stylus (can't be any more cumbersome than the traditional whiteboard), plus a text chat interface to conduct the interview. The interviewer interacts with the candidate in real time but doesn't see him or her.

Where cheating is a concern you can have a different employee (who does not provide interview feedback) monitoring the candidate over a camera to ensure they're only using the approved terminal, not phoning a friend.

Any medium to large-sized software company should be able to pull this off, and it would eliminate the possibility of gender bias in the interview results.

I agree with you that this would be a great experiment. However, that won't stop any potential bias once they get the job.

simple. require everyone to work in cubicles that are closed off where your coworkers cant see you. all voice chats must be done without video and with voice modulation technology so that gender can not play a role. Also, the persons name must always be changed to J Smith to not hint at gender or race.

Or change the name 'J. Doe' so you can't even tell if they are alive or not.

Agreed, of course. But I still think it's a good idea to eliminate one opportunity for bias if we can.

>wacom stylus

That might not work as intended. According to Wikipedia (http://en.wikipedia.org/wiki/Graphology),

>Uniformly the research indicates that gender can be determined [from handwriting] at a significant level.

Considering the subject is graphology, of course, one may question the quality of the studies cited but at least the journals like Journal of Applied Psychology appear legitimate.

In my experience, at the hiring level gender bias works towards normalizing the current skew of modalities. It is after hiring, when growing leaders and assigning responsibilities that bias is more often clear.

Reply on news.ycombinator.com