In an interview, Damore said part of the "lowered bar" is extra interview rounds for minority candidates. You could call that "more energy", but it also reduces the chances that a given candidate suffers from bad luck.

Luck doesn't reduce skill, it alters opportunity.

Consider a 90% free throw shooter vs. a 50% shooter. To get the job, you have to make 3 shots. But the 90% shooter gets 3 shots and the 60% shooter gets 5 shots. All of a sudden the 60% shooter is more likely to get the job.

It highly depends on the details of the interview process, which I don't know, but just to discount it with that logic is impossible.

This analogy doesn't work at all. 3/3 shots is 100% success rate whereas 3/5 shots is 60% - you're literally describing lowering the bar for the 60% shooter.

Additional rounds of interviews are more like trying to accurately diagnose a condition using multiple different tests because the initial test is known to have poor sensitivity and will produce false negatives. Doing multiple rounds may, of course, increase the chance of false positives (reduce specificity); but the assumption in this case is that when hiring minorities the sensitivity of the interview process is much worse than the specificity.

Right, that's why I said it depends on how the interview process is done. Some companies do thumbs up / down by round.

On a reread, Damore actually implies that Google's policies as applied to their interview process are "decreasing the false negative rate" for minorities. Whether this is harmfully discriminatory or not is open to opinion, but what I think is clear there is that the statement is favorable and understanding of his minority peers - the policies did not let anyone through who should not have been. I certainly don't think he should be fired for having given that statement.

I don't see how you don't recognize this as effectively lowering the bar (if not intentionally).

>Additional rounds of interviews are more like trying to accurately diagnose a condition using multiple different tests because the initial test is known to have poor sensitivity and will produce false negatives.

The difference is that what they're testing for isn't a binary proposition (do you have the disease or not), but a spectrum (what is your skill level). Viewing this in terms of false-positives or false-negatives is insufficient. If we think of programming skill as a spectrum, we can ask what is the average top-% of candidates who pass the interview (we might guess its top 5% of all developers). If everyone has the same test then the average top-% is unchanged regardless of any efforts to get more minorities to take the test. But once you start giving more tries to minorities your average top-% necessarily reduces.

Whatever your test is designed to admit (say you're interested in hiring only the top-10% of developers), the average of those who pass will be higher precisely because of the chance factor. Being significantly better than the intended cutoff gives you a better chance at passing and so those who pass skews towards better than the intended cutoff.

I'm not saying whether this is a good or bad thing, but the average skill of those who pass must reduce. It is very straightforward to see this as effectively lowering the bar.

Your hypothetical is self-fulfilling.

Imagine that if you're a minority, you have nearly a 0% chance of getting hired if you're the only minority in the hiring pool [1]. As an employer, wouldn't you want to counteract that by making sure that the decision to hire/not-hire isn't affected by status-quo bias, so you don't overlook qualified candidates?

https://hbr.org/2016/04/if-theres-only-one-woman-in-your-can...

> Google's stance, as stated in the article, is that they divert more energy in to finding minority candidates. That's not the same as a lower bar. Abilities are expected to be distributed equally still.

That has a very pernicious effect across the industry though. Think about what that policy does to other companies.

The other companies won't have as many high quality women because Google scouted and hired them already, but will have just as many low and medium quality women who aren't good enough for Google, and more high quality men who were displaced from Google. Which skews the gender ratio even more and creates the impression that women at those companies are lower quality than the men there or else they would have been hired away by Google -- because it causes that impression to be the truth.

And you can't fix it by having all companies adopt that policy, because it would still transfer high quality women from lower tier companies to higher tier companies, causing problems for all the women who don't get to work for the companies in the highest tier. Even the high quality women who are still in second tier companies.

The lower tier companies are where almost everybody actually works -- small and medium companies employ more people than huge companies because there are so many more of them.

Google is being quite selfish with a policy like that.

Only if you assume that finding qualified minority candidates is just as easy as finding other qualified candidates. Or that hiring chances are equally distributed.

HBR found there's an innate bias against any minorities in hiring pools [1], and considering that women make up a much lower percentage of potential CS positions, the deck is probably stacked against them. This means that for other companies, they have already passed on hiring the qualified minority candidates. Speculatively, Google could be trying to counteract this by diverting more energy into finding minority candidates.

Also, is it Google's responsibility to make sure other companies have the best candidates, minority or not?

[1] https://hbr.org/2016/04/if-theres-only-one-woman-in-your-can...

> Only if you assume that finding qualified minority candidates is just as easy as finding other qualified candidates. Or that hiring chances are equally distributed.

Nope, it's independent of any of that, because the effect is relative to what other companies do rather than any of those things.

And when you do that experiment in the real world rather than a lab, you get the opposite result anyway:

https://www.washingtonpost.com/news/innovations/wp/2016/07/1...

http://www.abc.net.au/news/2017-06-30/bilnd-recruitment-tria...

> Also, is it Google's responsibility to make sure other companies have the best candidates, minority or not?

It's not about who gets the best candidates -- presumably the men who are displaced are of equal quality and then go to work for the same other companies. The problem is that it creates an unfair black mark against every woman who doesn't get hired at Google despite Google having a special preference for them, and then leaves them in an environment with an even worse gender ratio than it was already.

And these other companies feed into Google. Plenty of women get their first jobs there and go work for Google later. If Google makes it harder for the women there and increases the number who drop out as a result, that's bad for everyone including them.

>The problem is that it creates an unfair black mark against every woman who doesn't get hired at Google

Can you elaborate on this more? Or give examples?

Suppose there are 5,000 women and 20,000 men with CS degrees who are seeking new employment right now. 500 women and 2000 men are above the 90th percentile, 500 women and 2000 men are between the 80th and 90th percentiles, etc.

Google has 2000 job openings. If they hired without gender preference they would end up with 1600 men and 400 women, but they make an effort to seek out women specifically and instead they hire 1200 men and 800 women. They've now hired all of the women above the 90th percentile and 300/500 between the 80th and 90th.

The gender ratio below the original 80th percentile is still 4:1, but above the 80th percentile it's 14:1 and above the 90th percentile there are no remaining female job seekers. People notice things like this -- that none of the available top engineers are women, even though there are still less talented or experienced female applicants. It creates stereotypes. It deprives the women below the 80th percentile of their role models and mentors. People start expecting women to be worse on average, because of those available to hire, Google has actually caused that to be the case.

And things go downhill from there very quickly if more large companies do the same as Google.

But those are just hypotheticals. Is there actually any evidence showing this happening?

It's not a hypothetical, it's arithmetic. The only hypothetical parts are the number of job seekers and the number of hires, but change them to whatever you like, as long as the proportion of women hired is different than it is in the applicant pool the effect still persists.

But isn't that true of any hiring practice? Should people start lowering their standards just so other companies can get better candidates too?

It's only true of hiring practices that cause the gender ratio in your company to be different than it is in the industry as a whole.

And then watch out if your company bases its diversity targets on numbers from the rest of the industry. All the good ones may be taken (in order to get to the mythical 50%), but now we want to follow suit (to get to 50%), but what's left? What a mess.

This is another of the "facts" that Damore's memo mentions but there's no empirical proof for. Do you have any?

Proof for what? Smaller number of available women? That's obvious. That if larger companies hire the best of them, the remaining supply/quality will shrink? That's obvious too.

No, that there's a hard number that Google recruiters are supposed to achieve (50% seems to what I've seen repeated around here.)

Does anyone have any actual proof that Google is aiming for 50% or to represent the general population 1-1?

It sounds like they kept meetings deliberately off-the-record, but then again here on https://www.google.com/diversity/index.html there's stuff like "we’re committed to making our workforce more reflective of the world we live in". Number goals are unlikely to be published for obvious liability reasons.

This argument reminds me of the "waterboarding is not torture" argument, both suffer from slothful induction fallacy.

If waterboarding is not torture then why would you apply it to detainees to confess information they otherwise wouldn't?

Equally, if diverting more energy in to finding minority candidates is not lowering the bar for them then why would you need to divert more energy to find them?

"If waterboarding is not torture then why would you apply it ..."

This applies to any sort of interrogation tactic along the spectrum between The Comfy Chair and Execution. Each step would not be applied if the detainees confessed at the next-below step.

>Equally, if diverting more energy in to finding minority candidates is not lowering the bar for them then why would you need to divert more energy to find them? reply

The point is to find the ones that can pass at a higher rate than you normally would. No bar is lowered, and your representation increases.

If you search harder for them, the bar is lowered by virtue of letting them be less outstanding / noticeable.

That doesn't follow if they're passing the same hiring process. It's not hard at all to get a contact for a phone screen with Google: I've been contacted by Google recruiters on linkedin with a barely filled out profile with one or two unremarkable positions listed. If you're counting the initial discovery phase as part of the hiring process, you're mistaken.

I believe the memo + followups included claims that the hiring process is not the same either.

If that's the case, then yeah definitely a problem

> If waterboarding is not torture then why would you apply it to detainees to confess information they otherwise wouldn't?

Actually effective techniques like building rapport aren't torture, and they get applied to get information out.

You are right, my argument is incorrect.