Comment by hedora | Hacker News Reader

Comment by hedora | Hacker News Reader

I found the results to be unsurprising:

Masking a woman into a male voice created a slight disadvantage, the other way a slight advantage.

Also, women underperformed men by about 0.5 on the scoring metric before the modulation.

After interviewing 100's of candidates, I realized women are more easily intimidated on average and also need to be much more careful in how they present themselves in interviews. (Some non-trivial percent of interviewers think aggressive males are confident, but aggressive women are combative. The candidate doesn't know if I am such a person when they come into the interview.)

I noticed that, on average, women with scores that match those of men (on my own scoring) significantly outperformed the men when hired.

I settled on adding 0.5 points to the score of every woman I interview (then rounding to a whole number). Also, I champion them if they are borderline in the panel discussion.

Based on the gender ratio and performance of the team I recently helped hire, this works great (when I am the interviewer).

Edit: There was a wonderful article by a VC along these lines. Some other VC bragged that their female founders were all way more competent than their average successful male founders. The first VC pointed out that this was as close as possible to incontrovertible proof that they were holding women to a higher bar before funding them. I can't find a link...

> I settled on adding 0.5 points to the score of every woman I interview (then rounding to a whole number). Also, I champion them if they are borderline in the panel discussion.

So you deliberately engage in gender discrimination? You just straight up give women a bonus, across the board, explicitly on account of their gender?

Imagine someone saying the inverse, "I find that men outperform women on average, so I settled on adding 0.5 points to the score of every man I interview."

I'd be fine handicapping in the other direction if the raw scores empirically favored women.

There is nothing sacred about my personal opinion on how people do in interviews. I judge as objectively as I can, and that leads me to unfairly score women 0.5 points too low (based on hindsight and hired candidates' job performance).

There is plenty of scientific evidence saying that coding interviews are way more stressful for women, causing them to underperform on average. Not correcting for that would be sexist.

You are publicly bragging about using sexism as part of your hiring criteria. You are adding points to women, or inversely taking away points from men, by pure nature of them being men or women. You are then hiring based on that.

You understand that if your information got out, its likely to result in you being unemployable right?

"I don't get it. Why are they confessing?" "They arn't confessing, they are bragging."

Setting a hard rule like this is quite strange…

Did I understand that right? Women you interview get an automatic 0.5-1.0 handicap?

0.0-0.5 depending on how they do. A solid "meh" is still a no hire.

A "they almost got it, and are obviously intelligent" turns into "they froze up, but would have gotten it otherwise".

(To be clear: I've recommended hiring men that obviously froze up. I don't care how programmers perform in high stress environments. We're not trying to reboot the computer in Jurassic Park, FFS.)

We're not trying to reboot the computer in Jurassic Park

That would be stressful, but not only that. Stress may come from pretty “calm” sides, like the lack of task explanation or illogical authoritative statements, indecision due to vague constraints, unreasonable time pressure, monotonous grind, working hard weeks right into the trash without any explanation (to name a few just from the top of my personal experience). Stress is something that bothers you so much for months or years but you can do nothing about it.

Reply on news.ycombinator.com