Hacker News Reader

Because "society" is made up of (among others) companies, and Google is a large company. The way things change in a society is when the members of society start to change them.

If a company with the weight and power of Google feel something is wrong and isn't doing their damnedest to try to fix it, then it won't change.

Now, from the linked piece, it's unclear exactly what has been proposed, so I'm not offering an opinion about whether it's a good idea or not -- merely answering the question in the general. For the record, I'm broadly sympathetic to the notion that fixing diversity starts rather a bit upstream from internal Google-policies.

The thing is, though, all companies are drawing from the same talent pool. Google hiring an above average number of diverse applicants doesn't raise the percentage of diverse tech workers. Those diverse tech workers came from somewhere. Google's act of hiring them in turn reduced another company's pool of diverse workers. Or if they came from a university, it reduces the number of diverse candidates that other companies can pick from.

Ultimately the total diversity of tech is dictated by the individual decisions of people when they choose their field.

> Ultimately the total diversity of tech is dictated by the individual decisions of people when they choose their field.

I might add that the total diversity of tech is dictated by the individual opportunities to make a decision, though I generally get your point.

The industry is not a closed system. People can leave or join it at any point in their careers, and they may (read: do) do so because they can't find jobs or the jobs they do find are hostile to them.

Unemployment in tech is exceedingly low, and there have been more openings than workers for a while.

And the latter point is essentially the other side of the same coin as compared to the point that I'm making: it's whether prior people join or leave the system as a whole that ultimately dictates diversity.

I agree that tech isn't a closed system. But moving diverse candidates from one part of the system to another does not actually increase diversity. Getting people into the system does that.

That there are more openings than workers does not mean that every potential worker experiences the same hireability. The entire issue here is that hiring biases towards a certain type of person and against another.

A woman may struggle to find a job while an equally qualified man does not. The woman may exit the industry. She does not become part of the 'pool' unless she stays.

This is how pushing for diversity in hiring can expand the pool of available workers, and how a company hiring marginalized people can be doing something other than cannibalizing other hirers (because those other employers are not bothering).

> A woman may struggle to find a job while an equally qualified man does not. The woman may exit the industry. She does not become part of the 'pool' unless she stays.

No, she joins the pool the moment she starts and she leaves the pool when she leaves.

If a company wants to keep people from leaving the pool, then the company should focus on reducing attrition. Hiring more diverse candidates doesn't necessarily reduce attrition. Remember Uber had 23% of it's workforce women at one point, it didn't help their environment.

When it comes to considering the working environment, I think it's informative to consider investment banking: a field infamous for pretty much all sorts of toxicity. Yet, it doesn't suffer from any particular gender diversity problems.

That's not to say that we should be OK with an abusive environment, or that we shouldn't always strive to make the environment better for everyone. It's just unlikely to be a key causal driver of low diversity in tech.

Not quite. The question is how do you choose the most qualified candidates? As an example, a lot of emphasis is put on CS degrees. But if you've been in hiring, they aren't an indicator of how good someone is going to be, and interviewing candidates is hard. I can teach people who can think logically how to program. What I can't teach is having the guts to do the right thing, diligence, curiousness, willingness to learn, etc.

I work for a big tech company; it is very hard to bring in people without CS degrees because of the interview process. Weirdly enough, the only place I've seen this happen is IBM, where they had a written exam based on logic questions, not programming questions (although the two are clearly related, and one was about scheduling/threading, it could be solved by somebody without knowledge of the terminology).

I have never seen numbers that indicate hiring CS majors is more cost effective, better, etc, and most job ads happily say "CS or related", but then the interview just doesn't reflect that. Maybe it is slightly easier to pick from a pool of people who have prior experience, but it's lazy. I think we can do better.

"Enter the field" doesn't just mean majoring in CS. It broadly refers to everyone working in tech. The point remains, if only 20% of people in tech are diverse, then Google hiring 25 or 30% doesn't change overall diversity. It just means some other companies will have less than 20%.

> "Enter the field" doesn't just mean majoring in CS. It broadly refers to everyone working in tech.

That's exactly what you get though, if you have interview processes that select for CS capabilities (which are not the only useful capabilities for a dev to have).

Hence, my sympathy for the idea that the real fix lies upstream: in order to get more women and minorities into tech, more members of these groups need to choose to study the relevant degrees in college. There are reasonable points to be made about the visibility of role models from these groups, but nothing much will happen to diversity hiring until the pool itself is more diverse.

> If a company with the weight and power of Google feel something is wrong and isn't doing their damnedest to try to fix it, then it won't change.

Great point. If any company is in position to spend resources and have an effect on society, Google is definitely one of them. And of course it is a great goal.

However they need to spell that out in plain and simple language and not hide it between the lines or say they are doing it because (as the article quotes) it would otherwise "make it difficult for the company to hire and retain workers, posing a long-term risk to its ability to innovate".

If innovation and ability to retain was directly tied to diversity each business unit down the lowest level would eventually figure it out in a few years. There would be no need for diversity chairs, advocates, executive positions. These groups would go out, seek, hire and accordingly compensate people just like they seek, hire and accordingly compensate experts in those fields.

> If you have a fever because of a bacterial pneumonia, antibiotics are much better than Tylenol.

You also don't go "but don't have tylenol to break the fever because there's a root cause too," because an unchecked fever is itself damaging to the brain. And you especially don't do that if the root cause is unknown, unclear, or will take a long time or is impossible to cure.

The root causes here are complex and will take decades to undo. Dismissing efforts in the meantime to address the symptoms is effectively dismissing the harm these systemic problems cause now.

I know this is an aside, but speaking as a physician the treatment of fever doesn't actually do much of anything [1]. The rates of certain injuries e.g. febrile seizures are the same with or without Tylenol [2].

[1] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1488736/, https://www.ncbi.nlm.nih.gov/pubmed/12076499/ [2] http://adc.bmj.com/content/88/7/641

That's definitely interesting! I didn't know that.

Still, I think the point is still valid that symptoms are still often worth treating, even if you are also working on the root cause, right?

You're right. Fever often feels bad, and like pain, is often worth treating. I just ask the patient if it's bothering them, and treat it if it is; the only exception is sometimes leaving a fever unmasked is useful to determine if one's treatment is working.

I had this misconception too & Google has not had 50/50 parity as a goal (unsure about specifically what this proposal is but I don't think it establishes quotas either). The discussion has been centered about reaching parity with the available talent pool as you describe. It's even in the article:

> Eileen Naughton, who leads Google’s HR operations, said the company remains committed to an internal goal to reach “market supply” representation of women and minorities by 2020, which could help bring hiring in line with the diversity of the candidate pool.

Thank you for this--I missed it in my read of the article! Thinking about it further, it's a careful balance between defining market supply correctly (e.g. high performing individuals who would normally be hired at Google) without the definition itself being discriminatory (e.g. high performers look like our current high performers).

It is interesting though that if are far from market supply parity now but they want to reach it by 2020 then between now and 2020 they need to apply really strong positive discrimination (over market supply parity) when hiring/firing.

Assumes equal distribution of skill across diversity lines.

This can be the case, but as gender bias for STEM starts at a very young age and we try to artificially correct it later, that may not be the case.

In spite of individual skills, however, there may be value in diversity of thought.

So the challenge becomes creating equality (job opportunities are purely derived from the individual's skills in spite of diversity characteristics) vs creating maximally performing organizations, which may benefit(or suffer) from those diversity characteristics.

The article quotes Google as shooting for “market supply” representation, instead of artificial targets of the proposal.

I can’t think of any legitimate reason they can’t achieve market representation so that makes sense to me.

Forcing 50/50 at Google would force many, many smaller companies out of balance. They should spend their efforts on the top of funnel imho, with more high school and university programs to support women and minorities.

Edit, here is the proposal:

https://www.sec.gov/Archives/edgar/data/1394096/000121465918...

That's overly simplistic. If women or minorities see women and minorities doing well at companies, they are more likely to apply to them.

If you want to break the pattern it makes sense for companies to take the initiative. Look at what Apple did at WWDC -- greatest percentage of women on stage ever. Having said that, the article has no details on what the proposals were, so it's unclear just how heavy a burden it placed on the company.

Building off your example, the company going out of their way to try and bridge this gap could hypothetically alleviate the 90/10% gap in the future. The gap is in part caused by women not having the same role models higher up in the STEM world to look up to and follow. In this case, Google is part of the society that leads to your gap in applicants and can definitely make a difference through their hiring policies.

I completely understand the other side that this may not necessarily lead to the most qualified candidates being hired though. It is a difficult topic to discuss, and we just don't know yet what effect equal hiring policies have because they should more strongly affect future generations.

From the article:

  > the company remains committed to an internal goal to reach 
  > "market supply" representation of women and minorities by 2020.

The term "market supply" isn't defined in the article, but I assume it's basically what you're talking about (except overall job market, not just applicants). If that's true, then Google is already accounting for that factor but minorities are still underrepresented.

Watch out for variance too.

If variance is different in the two populations then a high hiring bar will overweight the high variance population.

> That sharp of a gender gap is not the case in Eastern Europe.

Because gender equality is not in such a good shape in EE. The equal representation of women in STEM is inversely correlated with gender equality, it seems. Look at the data. In countries with generally high levels of freedom women don't choose STEM careers (look at Scandinavia), while in poorer and more backwards (in terms of g. equality) countries sometimes more women go to STEM careers than men. (That's partly because in rich and free countries women can choose not to earn STEM salaries and also STEM salaries are not that far from other professions', while in other countries women are often under pressure to go for STEM careers despite they don't fancy that at all.)

You're right, equal representation simply requires that Google's new hires reflect the current underlying labor market demographics. Attempting to skew that distribution in either direction is discrimination.

However, I would like to see tech companies provide more merit-based scholarship and mentorship opportunities for underrepresented demographics... the price of college and weak minority support networks contribute to the systemic biases in our industry.

You might also want to look into "Disparate Impact" [0] and "Unconscious Bias" [1].

> For example, if XYZ Company hired 50 percent of the men applying for work in a predominantly male occupation while hiring only 20 percent of the female applicants, one could look at the ratio of those two hiring rates to judge whether there might be a discrimination problem. The ratio of 20:50 means that the rate of hiring for female applicants is only 40 percent of the rate of hiring for male applicants. That is, 20 divided by 50 equals 0.40, which is equivalent to 40 percent. Clearly, 40 percent is well below the 80 percent that was arbitrarily set as an acceptable difference in hiring rates. Therefore, in this example, XYZ Company could have been called upon to prove that there was a legitimate reason for hiring men at a rate so much higher than the rate of hiring women.

[0] https://en.wikipedia.org/wiki/Disparate_impact

[1] https://www.youtube.com/watch?v=nLjFTHTgEVU

Here's a little thought experiment: Let's say the very most talented potential hires almost never apply for a job, because they get headhunted. That means the pool of applicants for an open position will mostly be a mix of competent-but-not-brilliant (or at least not-yet-proven-brilliant) and incompetent people. The hiring process, if it (mostly) works will lead to hiring the competent people while leaving the incompetents behind, who are then free to apply for job after job until the hiring process somewhere fails to filter them out. The point is, the incompetent people will spend much more time applying for jobs than the competent people will. This also means that at any given time, the incompetents will make up the vast majority of applicants to any given job - in other words they are vastly overrepresented in the 'market supply' that a company gets to see. This leads to some interesting effects. One is that many companies claim to hire only the top 5%, when in reality it just means that they hire only the few competent people in the pool that they see (and these competent people are more likely to be e.g. the top 50%-5%, since the actual top 5% got headhunted). Another is that if a company used a naive metric of the 'market supply' to determine things like the demographics of the available workforce, that metric would mostly be comprised of the demographics of said incompetent people. Which might lead to some interesting biases (IIRC it's fairly well documented that men tend to overestimate their abilities quite a bit more than women do - the saying is, if a position requires 5 skills, a man will apply if he has 1 of them, and a woman will not apply if she has only 4 of them). The point is, if you wanted to do something like benchmark a company's internal diversity against the market supply, you might need a better way to estimate that market than just looking at applicants, since that is mostly noise and very little signal (and the point of most hiring processes is to filter out all that noise)

Companies have a huge influence over the distribution of who applies by the wording and placement of their job adverts.

Interesting, so this sounds like we could start to apply principles of prospect theory and the system-level discovery utility of lawsuits. I've seen medical literature that argues malpractice lawsuits actually serve a valuable discovery function and may not actually add cost to the system since they make other parts of the system cheaper (preventing excess care due to underdiagnosis, for instance).

So, which sex generates more discrimination law-suits per capita? What is the payout on the average lawsuit? Does changing the sex mix decrease the lawsuits per capita from each sex? Is there a baseline "noise" of lawsuits?

I can't think of a way to fix society that doesn't involve companies consciously hiring diverse (and qualified!) candidates.

Companies are part of society too!

Perhaps people want a short term solution when the real solution needs to be in education, outreach, mediatization, etc.

I think this is known as "Pipeline problem". Has been discussed to death all over the internet.