Hacker News Reader

Why do people jump to 50/50? Because that seems to be the goal of people pushing diversity: desirable fields like software should reflect the population at large [1]. I think the assumption that 50/50 is desirable is wrong, too. So far I haven't heard of any compelling reason that genders are equal in all jobs - only the opposite (in both directions - some jobs have way more women). In fact, how do these folks even know 50/50 is fair to women? What if it turns out that women are actually better at software engineering, and the ratio should be 2:1 in favour of women? How are they determining these numbers? Or are they just making them up?

He's also challenging the point that identity=diversity. Getting a man and a woman that were born in the same area, had similar wealth growing up, went to the same schools, have same "political" views, etc. is not nearly as diverse as getting two men from different parts of the country with different political views, differ in wealth, etc.

None of this should detract from trying to eliminate bias. Blind reviewing sounds great, for instance - perhaps they should do that for the first pass where it's feasible to hide the candidate's details. Or provide transcripts of phone interviews then have people rate them and see how biased things are. There's ways to do research on this issue, but it seems that even suggesting there might be differences is a way to provoke "outrage".

1: I see less outrage about low female representation in mining and hunting, for instance, or people demanding dentists stop employing so many females as dental assistants.

The point about ratios is that tech is where a lot of the money, and thus a lot of the power, is. If there's a perception that it's more difficult to access that power because of your presented gender, then there will be outrage.

Interestingly, there is also no outrage that men die on the job twelve times more frequently than women, nor that they're underrepresented in higher education. Note that men are overwhelmingly more likely to be homeless, commit suicide, be victims of almost every kind of violence (women are slightly more likely to be victims of sexism violence, but men also underreport, so who knows?), die a decade younger, receive far fewer social services, etc. But hey, as long as tech companies are 50/50, the gender problems are solved.

All good points. Too bad it's not trendy to fight for those things at the moment. Sorry.

Diversity seems to be just a slogan these days. There's few contexts where gender, ethnicity or race really matter as specific traits to bring to the table. And I wonder if human empathy is not sufficiently capable of overcoming those differences. I mean, me as a white male, should I expect that a black woman can not design a product for me because she doesn't somehow get what it is like to be white or male? Why would the reverse be true in any way?

It's not that you can't design a product, it's that you may not even think of the special considerations required. For example: think of facial recognition apps that fail for dark-skinned people, or apps that aren't usable by the visually impaired or colorblind. Cars, mirrors, and countertops that are inconvenient for people of below-average (male) height. Nobody is purposefully excluding these people in these cases, but nobody is going out of their way to be inclusive either.

So you're saying that people developing face recognition just forgot black people exist? That even with a few % of employees being black, there wasn't enough testing to figure this out?

Versus the obvious answer that there's less contrast on darker skinned people, making it harder?

Likewise for all these special considerations. That's the kind of stuff you can checklist. It's no different than keeping track of required behaviour needed to write, say, portable C. The idea you need a member of every group in order to write software for them is quite an odd one. I'm aware of ARM's endianness even though I don't use non-LE platforms for any development.

And even then, suppose you can't just go by a checklist, isn't that what testing is for? Kids make poor software engineers, but we figure out how to make software for them.

> So you're saying that people developing face recognition just forgot black people exist?

Yes. This happens all the time.

In addition, simply being aware of what problems exist to be solved is also a huge thing. Engineering isn't just writing good code. It is solving real problems. Having people from a diverse set of backgrounds and experiences will lead to more opportunity to identify problems that can be solved.

Although that's not a good example, I agree that there are cases where you need some "inside knowledge" from the group you're targeting. By products I don't mean only software. It could be a social project, or anything else where culture is significantly important and not being intimately familiar with it could endanger the success of the project.

These seem like edge cases. How many CRUD web apps have gaping holes that would be corrected had the developers only been less white? And are the consequences of failed facial recognition so severe? The racial issue was fixed years before I got a camera that could leverage it, and how many people critically depended on it in that window? In fact, I can't think of a single other issue of tech that works better for white people, much less one that is actually important. And even if you were concerned about this, it's not justification to hire in proportion to broader demographics. None of this is to attack diversity, just these particularly weak arguments. We should absolutely maximize opportunity for everyone, because it's the right thing to do; not because it might help to avoid the once-in-a-blue-moon racial faux pas.

The less barriers to entry, the more women chose to stay away from blue-collar jobs. That's diversity. Forcing half and half irregardless of merit nor interests is madness.

I'm all for equal opporrunities and encouraging education path that go above and beyond the male/female stereotypes, but without forcing the end result to be equality of outcome.

That never works.

> the idea that diversity is just have 50:50 of characteristic X and characteristic Y.

There's more nuance to it than that. It's about having a profession be representative of the overall proportions of each characteristic within the population at large. It's only 50/50 because half of the population is men and women. For other characteristics, like say African-American, you'd be looking for 12/88, not 50/50. Instead, tech is more like 1/99, so clearly there's a lot more work to do there. And I'd like to point out that none of the so-called biological arguments that were advanced on sex apply to race, yet racial diversity is even worse than gender diversity. Clearly there's more at play here.

The problem is that if you force the exact ratio for every characteristic (sex, race, orientation, age, etc) for every job some people would end up being forced to do jobs they do not want to do.

From what I have seen and read, the biggest barrier to a 50:50 split is that we all have a tendency to define groups that we belong to and others that we do not.

Imagine a event that you enjoy doing (concert, conference, what have you). Now say that you find out that 95% of the participants is going to be part of a group that you do very clearly not self-identify as being part of. Would you feel equally confident going there if 95% of people were part of a group you did identify with? Many will ask "Why is my members of the group that I self-identifying with choosing not to go there?", while having the opposite effect when ones own group is a strong majority. Even if you had two equal choices to go to a place with 50:50 and one with 90:10, the reduce doubt will softly bias people towards the 90:10 if one self-identify with the 90% (exception naturally exist as with any soft bias).

I have seen some data on this from reports. A university student that fail an exam will be several times more likely to switch program if they belong to the minority gender in their class. The same effect can be observed for women in IT program as with men in education programs.

Basically, it is extremely hard (if not impossible) to get a 50:50 equilibrium without some form of homogenization where the unique identifies of the groups get merged into a single group. The one good thing is that as a matter of ability, people can belong to multiple groups and it can be enough to belong to one common group and have a perceived 50:50 splits based on other groups.

I share your confusion about the 50/50 split.

> Recognising some groups lack one or more of those things is a good start. Recognising some groups (deliberately or accidentally) monopolize those things to the detriment of others is a better start.

If like to propose an even better recognition: that no one is monopolizing anything and both groups are making the choices that best match their priorities.