The Kolmogorov option

>seems very hard to understand if most people believe that there are differences between the genders that could affect the "outcome"/gender gap in tech/whatever.

You're changing your original summary of the argument by adding the qualification "that could affect the outcome". The whole question at issue is whether the differences that there are between men and women would be expected to have a significant effect on outcomes in tech.

> Internet anonymity

With policies similar to Facebook's real name policy and Google's long memory, that veil of anonymity is getting pretty thin.

I'm all but certain that, despite my lack of personally identifying information on my profile here on YC, that my company could trace back anything I wrote here to me as a person. I'm also 100% certain that a state actor could do the same (even if they are less likely to do so).

Is my company regularly Googling my name and my aliases? Beats me. Though if they were concerned about smears against their reputations or leaks of sensitive information, I'm sure they would start.

Most of all, consistently apply the maximum level opsec right from the beginning. Even if the first hundred posts on your new blog are inoccuous, you can't let them be connected to you when five years from now you find out that you actually want to write about something controversial for once. The hardest part is planning ahead and investing a lot of time and effort before the point where you actually publish the dangerous thing you don't want to be doxxed for.

A blog (or any other publishing mechanism) needs an audience, assuming your goal is to affect public discourse. You can post anonymously or using a short-lived pseudonym on Hacker News and other forums, or on any blogging/publishing platform like Medium, and this is very secure if you never use the same identity before or after. But you're very unlikely to affect many people with your writing that way.

You can't use your regular identity's reputation to drive attention to a one-off post ("Steven Pinker responds to the Google Memo controversy!"). And you can't (reliably) produce one or a few posts so well written and timed that they spread virally and affect many people even though you've never published under that identity before.

So you need to spend a lot of time and effort publishing other things under that identity to acquire a following. You need to be able to consistently write things that attract the kind of people you want to reach and give you the right reputation.

Obviously regular good IT security is needed: don't let your server / publishing account be compromised, vet your hosting service's security, etc. Access it via Tor, setup posts to go up at unpredictable times not tied to your timezone or what you're doing at the time (work/leisure), cultivate a style of writing that's clearly similar across that identity's posts and dissimilar from other things you write. Don't refer to the pseudonym when in any other identity; even if others mention it, don't say you've read it or discuss it. And so on.

Occupational preferences are not the subject of the document, though. The subject is occupational qualification. The specific things he criticizes Google for doing are actions taken once minority candidates have already applied for a job, that is, once they have already expressed interest in the occupation they're applying for: they're about reducing unconscious bias in hiring, reducing the false negative rate in Google's notoriously noisy interview process, ensuring that internal groups meet standards of representation, etc.

I just reread the document to confirm I read it right the first time. The subtle implication, and I'd argue one of the most harmful effects of the document, is the claim that biases (whether from nature or nurture) in interest also translate to biases in ability. There's a huge difference between saying "More men than women are interested in the job but we've got to hire the most qualified people regardless of either gender or interest" (and he almost gets there in his discussion of men seeking high-status jobs!), which I don't think I could really complain about, and "More men than women are interested in the job so our efforts to hire women are 'effectively lowering the bar' and probably illegal."

If you are at the point where you have 10 resumes on your desk from 10 qualified women, and 10 resumes on your desk from 10 qualified men, and you put together a team of 9 men and one woman, no matter how true it is that women as a whole are less interested in the job, it's not relevant.

(Incidentally, that's a quote from the portion of the abstract where they're mentioning existing research instead of their own, and it cites three sources, the first of which is a book by a lawyer that appears to be targeted at the general non-fiction market. I'm rather surprised that this is a permissible citation in a scientific paper, and I'm wondering if my standards are wrong or this isn't a super reputable journal.)

I don't think anyone is disputing that there are biological differences between men and women (on the average) that result in unequal distributions of some outcomes. An easy example is "fastest marathon" or "number of babies".

The dispute is that there are biological differences between men and women that are relevant to qualification to work at Google in software engineering.