When the Australian public service experimented with blinding demographic data on resumes, fewer women and minorities were hired:

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

A blind recruiting process only increases diversity if there exists significant bias against diverse candidates. I wouldn't be surprised if Silicon Valley has a bias in favor of diverse candidates, like Australia's public service.

And I think that this is the bind that Google and others are finding themselves in: the assumption is prejudice so they make honest and effective efforts to combat prejudice and don't get the desired result in terms of outcomes.

See the example of the GitHub-sponsored conference whose trial of blind reviewing resulted in zero women speakers.

So Google is getting hammered from two sides, on the one hand they're getting sued because the outcomes are not what they're supposed to be, on the other hand they are already doing diversity programs that at least skirt or already dip into illegal practices. (In his interviews, James Damore has mentioned secretive diversity session that were not recorded, unlike everything else at Google, and that told of practices such as quotas that to him seemed extremely sketchy or downright illegal)

Of course this goes back to Pinker's comment in his debate with Elizabeth Spelke that we must distinguish the moral proposition that people should not be discriminated against on account of their sex from the empirical claim that males and females are (biologically) indistinguishable, because distinguishing them is actually essential to protecting the core of feminism. Because if you don't distinguish and it turns out that your empirical proposition is false, you must then either (a) discriminate or (b) distort/suppress the truth, or both. And that appears to be exactly what is happening.

https://www.edge.org/3rd_culture/debate05/debate05_index.htm...

UPDATE: Google's Diversity Efforts Fall Flat -- https://www.axios.com/googles-diversity-efforts-are-making-l...

You have an excellent point, but your use of the phrase "diverse candidates" misses the concern that the most important diversity is in thought and culture, not just predetermined demographics. A "diverse candidate" in reality may not look like an official "diversity candidate".

Another question is what makes you think that all whites think the same and therefore it will narrow the spectrum of ideas?

This assumption is never discussed and always accepted as true.

Why is that? Why are all the whites (or white males) identical in their thinking?

It's clearly not true. Hiring a Slavic man would, in abstract, be an increase in diversity in a lot of shops. They're likely Caucasian, though. One response is that it's important to officially value historically disadvantaged groups, but many Slavic groups are historically disadvantaged.

So, in a weird way, diversity initiatives come from a very ethnocentric (American) perspective.

Note that men and women typically present themselves differently on CVs and in selection criteria, with men typically giving themselves credit for work done in teams while women typically talk about the teams they were part of.

This would easily explain why blind resumes resulted in a larger proportion of men being selected.

edit: to contrast with the orchestra scenario, a resume is not a skill test, while an audition for an orchestra is a skill test for exactly the type of skill that the candidate is being chosen for.

Interesting hypothesis, but apparently not what happened:

"The trial found assigning a male name to a candidate made them 3.2 per cent less likely to get a job interview.

Adding a woman's name to a CV made the candidate 2.9 per cent more likely to get a foot in the door."

http://abc.net.au/news/2017-06-30/bilnd-recruitment-trial-to...

It's an interesting result, but is it accurate to call this "blinding"?

I feel like hiding demographics and actively deceiving people about demographics are different processes and should have different names?

Calling it "blinding" might not be accurate but, if anything, using deception, the study showed that men were discriminated against, not women -- contrary to what the initial hypothesis suggested.

Given "men typically giving themselves credit for work done in teams while women typically talk about the teams they were part of", then if you see a female CV it's more likely to be understated than if you see a male one. Which predicts two things:

* People aware of this will prefer otherwise identical CVs with female names (expectation is the underlying candidate is stronger, just more modest)

* Blind hiring will get you more men (because hiring managers can't correct for the male/female difference in self promotion anymore)

See, that is what is known as "sexism".

In no company in the world is this a "diversity policy." Diversity policies are the exact opposite of this. If they were as you describe, I'm sure the the author of the memo, and everyone who agrees with it, would be completely in support of them.

There are companies that strip personally identifying information from the resume in the first few phases and iirc diversitiy did go up. Been a while since I read about it so I don't recall thr complete details but there is a radio lab episode about it.

This year, GitHub did that for a conference, letting people submit talks and compete for selection completely stripped of identifying info.

The only people selected were men.

That conference was canceled because it didn’t stand with Github’s stance on diversity, etc.

[1]: https://news.ycombinator.com/item?id=14480868

—

Companies actively do the exact opposite because they need/want more diversity. It’s absolutely not like the music world where a large amount of women play instruments... the number of women in CS professionally is low. The number of female CS graduates is and has been low.

Simply stripping info doesn’t mean anything if at the end of the day a diversity agenda has to be met.

That also depends on how many submitted talks.

It would be a farce and ridiculous if 100 talks were submitted, and 99 of them were <gasp> males, and then they shut down and complain because no females were chosen.

It's called statistics.

This is why in the face of that kind of odds, a diverse conference has to look very hard for female speakers.

Even if there are factors that mean there will never be 50% female submitters, a proper diversity policy means the organizer sees this as a symptom of a problem they can help solve. Perhaps if a conference can keep at least 1/3 female speakers every year then 25% of submissions will be from women in 10 years? And perhaps if 1/3 speakers are women, more women will visit the conference, widening the audience?

The whole point of active diversity policy is that the status quo is to be changed by active means.

One of the big issues here is whether "active means" includes firing people who earnestly question the "active means".

I have no idea if that's true of tech companies. Regardless, you are confusing the part-- tech companies-- with the whole-- the set of all companies (of which orchestras are included).

I was responding to a non-expert's cost-benefit analysis that confused the part-- a specific type of diversity policy-- with the whole-- the set of possible diversity policies. She generalized a set of trade-offs and claimed unasked/unanswered questions around the field based off that part. I referred to a well-known historical diversity policy for which her proposed trade-off does not exist.

That ought to arouse skepticism that a non-expert who isn't genre-literate in the field is able to find the best solution to the problem.

The curtain has been tried. It didn't get the desired results so was retired.

http://mobile.abc.net.au/news/2017-06-30/bilnd-recruitment-t...

What if you do that and the number of women in the orchestra shrinks?

Or even if the number of women just stays the same. Is the effort to level the playing field then deemed a 'failure' if the results don't change in the way they're expected to?

That's the important question here, especially since there have now been preliminary experiments in tech interviews suggesting such an outcome is not implausible:

http://blog.interviewing.io/we-built-voice-modulation-to-mas...

If you've removed possible bias from your hiring process, then you have to accept an imbalance is either a result of uneven retention inclusive-or something operating earlier in the process of training your candidates.

The point is to remove as many obstacles as possible. We want the best to get the jobs, but if the are barriers based on gender or race or religion that's potentially denying us a brilliant woman, or black man, or Muslim.

It doesn't necessarily mean you end up at 50/50, I'm quite willing to accept that the sexes might be biologically different in brain structure and that this might mean one sex might have different average aptitudes, but I won't accept a societal construct on top of that.

You should never hear that someone shouldn't try to do something because of their sex. Never.

Well unless it involves the more obvious anatomical differences. The average man is going to have some difficulty becoming an effective wet nurse.

I wonder, though, how companies' hiring processes alone are supposed to fix the societal biases. Isn't it a false choice when given only the two explanations: A) We fix biases in the workplace/hiring process and get a 50/50 distribution in the workplace or B) There must be biological differences.

I'm personally rather convinced of a third alternative: There is some bias and even outright discrimination in some workplaces, but gender norms in society as a whole create such a big difference in the distribution of applicants/talent pool that it becomes impossible to correct as late in life as the hiring process. I think the gender norms are very deeply rooted and entwined through most aspects of life and that if we truly want to fix the problem, as opposed to get good PR and pay lip service, we need to change gender roles drastically. And that starts pretty much from pregnancy and goes way beyond recruiting.

I had to scroll down too far to find this comment. Diversity hiring is treating the symptom, not the disease of segregation at birth into distinct gender roles.

I think it's bad to write "woman, or black man, or muslim'. Being a woman or a black man is something that is determined before birth. Your religious views are adopted after birth. That's why I think it is bad to speak of muslim people, as if that defined them the most. That leads to seperation and distrust, we-against-them. Your sex and skin colour is given, your religion can change.

> That's why I think it is bad to speak of muslim people, as if that defined them the most.

Lots of people primarily identify by their religious affiliations. I think that's fair.

> Your sex and skin colour is given, your religion can change.

Ask a Muslim which will change first: their gender, their skin color, or whether Mohammed was Allah's prophet.

All that being said, your point about religion and race being separate categories is fair and well taken.

> Ask a Muslim which will change first: their gender, their skin color, or whether Mohammed was Allah's prophet.

You are generalizing in a way that will create animosity. There are liberal people which identify as muslims.

Even though most religious people identify themselves by their religious affiliations, not all do, and not everbody is religious.

I think religion is a cancerous meme, that we shouldn't pay respect to. I don't even want to be asked whether I am religious, nor what religion.

My name allows a funny pun in that regard: I'm Christian, but I'm not.

> You are generalizing in a way that will create animosity

> I think religion is a cancerous meme

My underlying point was that people with faith don't see their faith as a choice about the existence and character of their god (or other non-deist beliefs). It's not objective or accurate to say religion is different because it's a choice.

If anyone is offended by my point, I would challenge them to rethink their assumptions about how tolerant and accepting they are.

I wouldn't say that your point necessarily offends anybody. My point is that it might be not clever to present it like that. One way to get rid of enemies is to befriend them.

If 50/50 isn't the target, how will we know when we get there?

Do you have to know the target? Correct for bias you have, then question what it is upstream that is creating that bias.

How do we correct for the bias if we don't know what the unbiased state would look like?

> How do we correct for the bias if we don't know what the unbiased state would look like?

The goal isn't to choose some specific gender balance and then take whatever steps necessary to produce it, it's to eliminate unfair bias. The way to do that is to find it, not indirectly by looking at ratios, but directly by actually finding it.

High school teachers who say things like "women are no good at computers" to young women should be reprimanded etc. Poll women who didn't go into tech and ask them why not, and if any of the reasons are unjust then change them.

If no one can find anything like that then the gender balance at that point is what it's supposed to be. We're obviously not there yet, but the way we know is because we keep finding things like that, not because the gender balance is uneven.

I agree. But I notice that most discussions on sexism in technology companies focus on the outcomes; gender ratios and pay gaps. It would be interesting to try to measure the bias directly. Hide a bunch of microphones in offices and see if the number of sexist comments is larger at Google than it is at a law firm or a hospital.

Perhaps somebody has already done this. What would be the correct search terms?

> But I notice that most discussions on sexism in technology companies focus on the outcomes; gender ratios and pay gaps.

It's a specific instance of the more general manage-by-metrics disease. The thing you actually want is hard to measure, but it correlates with something that is easy to measure. So instead of doing the hard work to understand what is actually happening in detail, they measure the easy thing and optimize for that instead. Even though doing that frequently breaks the original correlation.

The result is the bureaucracy edition of a paperclip maximizer. You get what you measure instead of what you really want. Or bang your head against the wall, if the thing you measured is actually stickier than the real problem.

> t would be interesting to try to measure the bias directly. Hide a bunch of microphones in offices and see if the number of sexist comments is larger at Google than it is at a law firm or a hospital.

To some extent this is just the same disease. Is sexism supposed to be alright if it turns out there are equally large amounts in both places? Should we be satisfied that it's the root of the problem if there is very little at Google but even less somewhere else?

Stop trying to measure things against other things and just consider them in their own right. Sexism is bad regardless of how common it is. You don't fight it because there is more of it over here than over there, you fight it everywhere because it exists when it shouldn't exist.

Sure, but knowing where it was most prevalent might give us information about how to promote a better culture.

Frankly I do not see how 50/50, or any fixed number, can be a representation of the unbiased state. The unbiased state is unknown.

If you look at what the orchestra did and remove factors you know are irrelevant for the applicants then you should move towards a more unbiased state independent of what the number actually is. The problem is to find out what bias you have. One way is to compare the people who apply with the people you hire. You can see what traits you select on, then you can decide if you think those traits are good or bad.

You ensure a fair process (without attempting to overcompensate/correct for other processes out of your control), and whatever comes out is the unbiased state.

You can't escape the fact that (assuming any innate differences whatsoever) equality of opportunity will mean unequal outcomes, and equality of outcomes requires unequal opportunities.

> We want the best to get the jobs

I think it's important to remember that this might not be the case. We want the best team (or orchestra, or whatever).

These are perhaps best seen as two different things. Anonymizing applications will prevent discrimination/prejudice. This is objectively a good thing. It's not controversial in any way.

An active diversity goal in hiring must go one step further and mean you see diversity of the group as a goal in itself - which, as he says, often means you some times hire a less strong individual.

This is the controversial part. But let's remember that the goal of recruiting is not to hire the strongest candidate but to make the company successful. If you buy into the argument that diversity has a value, such as that a diverse company is better at making products for a diverse audience, or that diversity creates a better work environment and therefore higher productivity - then hiring a female candidate over a better make one can make sense.

Google are countering an existing bias by hiring people contrary to the bias. The orchestra is removing the bias.

Is it possible for Google to remove bias? Isn't the real problem further upstream? I bet that if you look at all applicants applying for jobs at Google you will see the same skew as Google have in their offices already.

But that is the exact opposite of what is happening with positive discrimination and probably something the initial author would prefer.

There was an argument made about it some time ago, that in 19th century playing music was an excellent education for a lady, so thousands if not hundreds of thousands of women had to take piano/violin/flute lessons and become skilled at playing music. Yet that time period has yielded close to zero great women musicians/composers, even though the society has pushed them at a disproportional rate to pursue music playing. Obviously there are other factors here, like the expectation that a woman is a housewife first even if she can play music really well. It's an interesting debate I think.

The equivalent of your example in tech industry leads to number of men in software growing - which is precisely what diversity policies try to counter.

There is also the study where they changed peoples voices in phone interviews, and the women with male voice were suddenly less likely to be hired (although the effect was small, iirc).