Comment by _justinfunk | Hacker News Reader

Comment by _justinfunk | Hacker News Reader

Parent Full thread

A hiring process is by nature a discriminatory event. Right?

You have a pile of resumes and you discriminate between the candidates during the selection process. This happens on a number of different criteria.

The criteria that we should not discriminate on are sex and race.

Anyone that doesn't agree with this is out of the mainstream.

The issue is that if you cannot discriminate on sex and race, and 90% (made up number) of your qualified candidates are men that are asian or white, then 90% of your workforce will be men that are asian or white.

So, then you have to make hard choices, guided by the current laws, and a desire to create a corporate culture.

If more than 50% of your users are women but only 10% (again, made up number) of your employees are women. AND you can legally choose a qualified woman over a qualified man. Then you have a good reason to hire the qualified woman over the qualified man.

TLDR; discrimination is part-and-parcel of the hiring process. discrimination on the basis of sex or race could be a legal and accepted thing for a company to do. discriminating on the basis of sex or race for bad reasons is bad.

what happens if among 100 applicants, 80% are male, 20% female?

Within each bucket, top 25% are qualified; hence 20 male, and 5 female.

But you only have room for 10 positions, 5 for male, 5 for female to reflect customer population

For male, you end up hiring top 5% whereas for female, it's top 25%.

What's an equitable way to address this?

Expand the pool of applicants until selecting the top 5% produces a male/female balance closer to that of the general population. Smaller companies don't necessarily have the ability to do this, but Google has the resources to do it.

Expanding the applicant pool will never reduce fairness, and it gives you more opportunities to hire the best of the best, assuming your hiring pipeline can actually select them.

Yes, Google has the resources to do that, and so can other big companies. But you'd have to pay top 5% female engineer more than top 5% male engineer because there's a lot fewer of them to go around. Simple supply and demand.

Do you then raise the male salary? That will only contribute to larger male candidate pool.

Paying them equal to what the men get might be enough, since the last time people did a comparison, women at Google earned less. (It's possible this is fixed now, but since they aren't open with salary numbers, people assume it's not.)

You can address an unbalanced hiring pool of qualified candidates (i.e. your 90%) by expanding the hiring pool (through various methods) until its balance comes closer to the balance of the overall population. Doing this doesn't require discriminating against anyone. See https://www.theguardian.com/technology/2017/aug/07/silicon-v... for example, there are other tech companies with race balance much closer to national averages. It's certainly possible to pull more women and people of color into your hiring pool, and you don't ever need to turn down a single person based on their white skin color to do it. The universities you hire from and your degree requirements for applicants can both have a significant impact on things and it's easy to adjust those.

The downside is that some qualified candidates (white man and woman of color alike) will potentially miss out on jobs once the candidate pool is grown, because you've pulled a bunch of extremely talented people into the pool who set the bar higher. That could certainly make people feel discriminated against if they don't have enough insight into the process.

Or Google can reduce the overrepresented candidate pool to match that of underrepresented by randomly select the appropriate number of candidates from the overrepresented (i.e. Randomly reject, lol). This should have no effect on te bar each candidate is expected to pass, underrepresented or not.

Interestingly this hypothetical scenario would probably increase the false negative rate for overrepresented applicants due to random rejection, which effective lowers false negative rate for underrepresented applicants relative to the overrepresented.

I think we are circling the drain of the right approach to this problem.

If Google's pool of qualified candidates were a "proper" mix of diversity, then I don't think they would need to have the practices outlined in the manifesto.

So encouraging women to get into STEM and to facilitate pulling talent from around the globe to get them into the pool of candidates is 100% the right move to address the diversity issue.

BUT, if women don't come in the numbers you need them to and your international outreach yields mostly asians... what do you do?

Reply on news.ycombinator.com