Hacker News Reader

Comment by program_whiz | Hacker News Reader

I have a question to further this debate.

Would people consider it sexist to administer a completely automated test of technical and personality questions which was used by an unbiased program to hire only the best qualified candidates?

What would you say if the results were essentially the same as the status quo?

By the Article's Author's account, she believes that we would probably maintain the status quo with such a test, because she thinks people are self-selecting out of STEM. Others seem to think that there is some other barrier to entering -- would a test like this fix the issue, or is there something else going on?

View on HN

Well, first of all, such an unbiased test does not and probably cannot exist. People have to write the test, and despite their best efforts, they will likely still be biased. And how do you determine what 'unbiased' is in the first place? Such a test would likely give us the status quo or close to it because most hiring managers are acting in good faith already (at least, I chose to believe that people don't mean to perpetuate sexism/racism etc, they just aren't necessarily trying NOT to do it.)

> What would you say if the results were essentially the same as the status quo?

OK, I'll bite. I'd say that's both surprising and interesting. Then I'd ask why the status quo is even relevant. "Best qualified" at the time of taking the test might not mean "most productive a year hence" when a candidate's circumstances might have kept them from reaching their full potential. Indeed, circumstances might have kept them from making it as far as the test at all. When there are statistically valid predictors of these things likely having happened, why couldn't - or shouldn't - Google take advantage of that knowledge? Also, no individual test can account for the common phenomenon of diverse teams outperforming monoculture teams. Hiring the best individuals is not the same as building the best teams.

Thus, even if such a "unicorn test" could exist, and even from the most hard-hearted "Google shouldn't consider social justice" perspective, the test would only be one input for selecting candidates. There would still be sound business focused reasons for overriding its results some of the time.

Not being sarcastic, is there evidence that more diverse teams are better? There seems to be some anecdotal evidence going both ways (i.e. more diverse viewpoints vs team cohesion). I can imagine that there's a balance point as well. You could have one extreme where people are so similar that its basically a one-person team with 8 arms. You could also have a team so diverse they can never agree, have no common language and no shared cultural aspects, or even shared goals. An interesting balancing act really.

The most often cited paper w.r.t. gender diversity (the kind most at issue here) seems to be this one:

https://papers.tinbergen.nl/11074.pdf

A more readable summary is here:

http://gap.hks.harvard.edu/impact-gender-diversity-performan...

The authors did indeed find a balance point, at about 50:50 (a far cry from the 81:19 among engineers at Google). OTOH, this was for a very different kind of task than programming. Another starting point is here:

http://www.chabris.com/Woolley2010a.pdf

It's particularly interesting that many summaries of this work will use vague terms like "composition of the group" to avoid mentioning anything in the findings about number of women in the group.

There's already a number of companies that do blind reviews of resumes, or applicants that intentionally hide or change their names - not sure if it makes a big difference, also because there's just less female applicants in general.

Actually, how many female software developers can't get a job nowadays?

The problem isn't that women aren't getting jobs (the market is currently in too serious of a demand for hands on keyboards for that) but that they aren't getting as good a job as you would expect from their qualifications and/or that they aren't being paid as much.

I'll have to go see if I can find it, but recently there was a study in Britain, France and ? (Germany I think) on compensation. Ah, here it is:

https://www.economist.com/blogs/graphicdetail/2017/08/daily-...

While there was a definite gap for the population as a whole, when you zoomed into the same job at the same company, the pay gap disappears.

No idea if that applies to women in tech but it was new info to me. Does anyone have a source for a pay gap in tech?

Reply on news.ycombinator.com