You can’t fix diversity in tech without fixing the technical interview

> When we throw up our hands and say "cultural differences", it implies that we don't need to ask why such differences might exist and whether we should actively seek to change them. That's an important conversation.

Except the article (and conversation) isn't about cultural differences, but instead about "diversity problem[s in tech]" and in this case, how the technical interview unfairly penalizes underrepresented groups.

1: I believe that in general, Caucasian Russian males can be safely lumped into the "white male" category that dominates the tech industry.

Turns out, people who are bad at technical interviews, are underrepresented at companies who use technical interviews to hire employees.

Perhaps my take on things is different since my degrees are in math, so I don't have a bunch of 'canned' algorithms stored in my head, and so in a whiteboard would come up with my own technique, even for possibly basic things like trees / sorting.

This in particular is the concept of a "model minority".

Culture difference cannot explain why women are not well represented in tech.

Culture difference can be a part of systemic bias, anyway. Let's say you have some programming job. The interviews focus on questions about frobbing linked lists (which is often the case). Now, it turns out that people from demographic A are more likely to focus on these questions when practicing for interviews, whereas demographic B focuses on honing the skill set they need for the actual job (algorithmy linked list stuff may not be necessary at all!).

You'd say that it's not possible for /culture/ to change how you study. There's hard evidence against that -- take India for example, CS students in India focus a lot on solving typical interview questions and less so on other skills. This stems from the general culture of finding and using a formulaic method of succeeding at academics. I don't like this culture (it ultimately leads to cases like kids who can solve whatever they need in quiz papers but don't actually understand the topic), and always have spoken out against it, but it's there.

But, you say, that's a different country, what about a minority demographic within the same country? Firstly, minorities more often than not are segregated into their own communities (socially, if not physically) due to various voluntary and involuntary reasons, and the effect can be the same as being in a different country (albeit milder).

Now realize that an interviewer will probably design questions around the kind that they are used to and that they (and their peers) would do well with. This is an unconscious bias -- interview questions are more often than not tests of how similar you are to the (often non-diverse) peer group of the interviewer, and not actually a test of how good you are as a candidate. This is an association fallacy, "all people like my peer group are good programmers, hence all good programmers are like my peer group". Of course, interviewing is often about avoiding bad programmers much more than it is about hiring good ones -- generally folks are okay with accidentally not hiring a good programmer, but want to avoid hiring a bad one at all costs. Thus tweaking your interview process to target a subset of the set of all good programmers isn't that bad, is it? Except that this targeting carries an unconscious bias inside it, and unfairly ends up disproportionately excluding minorities. If you're going to use a process prone to false negatives on purpose, be damn sure that the likelihood of someone being a false negative is in no way correlated with the demographic they are from. You may not have designed the process with this correlation in mind (hence, "unconscious bias"), but this bias is still there.

-----

And it doesn't have to be culture differences. You have a lot of other behavioral differences due to economic status or demographic which are unconsciously excluded in interviews. E.g. women are less likely to apply for jobs where they do not have all the listed qualifications, but men are very likely to apply, and get hired, because not all the qualifications were necessary anyway. The fix here is to be clear on what you absolutely need in a candidate, and what would be "nice to have".

And it doesn't have to be behavioral differences either. Sometimes it has to do with defining your inputs. Many companies focus on hiring from a set of people which itself is biased. It can be by hiring only from certain colleges (and then inexplicably whining about the "pipeline" -- you chose the pipeline!). It can be from having unconscious or conscious discrimination against autodidacts. Or code bootcamp folks. If you restrict your inputs to an already biased source, you will get biased results.

There are tons and tons of articles and papers out there about this. These aren't unknown effects. They're well established.

None of this is on purpose, of course (at least, I hope it isn't!). But it's still a bias, even if it's unconscious. And the onus is on the creators of the bias to fix it.

There are a million reasons out there why various demographics aren't well-represented in tech. It's just lazy to blame it on the demographics and say "it's cultural differences" without realizing the mistakes made by tech cos in making culture a factor in the first place.

I worked at Google. Turnover was really low. 10 year average stay doesn't sound unreasonable for me if you select only the employees that have left.