I don't think this is an okay implication to make based on such little information (we're supposed to trust that the people he spoke to were "wishy-washy" about "diversity"), and I'm asking if it's just as okay to question his own intentions.
I am making no such implication. I am simply saying that I don't find it "highly dubious" that Google has extreme levels of gender bias in teams correlated with area, based on some amount of data (looking at basically all the open reqs in backend in Google NYC over a couple months).
I do find it worrisome to see "sexist" thrown around so quickly: this is a question of objective fact, and we either need to be comfortable with applying the label "sexist" to lots of scenarios and not interpreting it as equivalent to "morally inferior and execrable", or we need to avoid the label, because otherwise there are chilling effects towards discussions of objective fact. The assertion was made, in the lawsuit, that backend teams at Google do in fact fail to hire women. In order to discuss that, we have to be willing to talk about teams that fail to hire women.
That said, if I'm going to support free speech, I should be consistent, and I welcome your attempts to question my own intentions. But to that I would reply that if I was looking for someone to marry, I'd specifically want to avoid them being on my team (even if I'm okay with the ethical breach, conflict-of-interest rules would prevent us from marrying), so joining a company that hires women but not for my field would actually be just what I'd be looking for in this case. Perhaps more factually-tenable aspersions to cast would be that I'm gay and get hopelessly distracted if I'm working with men? That's not true either, but it'd make more sense.
But I also want to know that the company is taking the problem seriously. If your ratio of applicants is biased at 50:1, either you're doing something that I'm not super happy with, or more likely, you're trying your best and your recruiting funnel is doing something that I'm not super happy with and your ability to change that is limited -- which, to be clear, is not really your fault, and more the fault of how company politics works and what power you have to fix the recruiting process, but if your company is set up so that you as a hiring manager don't have the power to make your recruiting pipeline better, that says things about your company's management culture / power structure that are a negative indicator to me. (In particular, since I tend to work more in support functions like building platforms than direct profit centers, my management chain having limited political sway in the company is a pretty big concern for my long-term happiness, regardless of the thing they're trying to accomplish with that sway.)
If you tell me your recruiting funnel is biased at even 5:1, I'm not going to insist on 1:1. I know it's a hard problem. If you tell me it's biased at 5:1 and you're working to change it, fantastic, that's more than I've heard from most people.
And I'm not a hiring manager, I'm a senior engineer who at the moment is wearing a different hat half the time. I'm not sure I'd trust any point of the funnel to non-technical people when it comes to hiring technical people. There are too many potentially interesting things that could be mentioned that someone who doesn't have a deep understanding of the field would discard good candidates out of ignorance.
This is how I feel about people who tell me how stupendously low the pass rates of their interviews are.
Also, we already had everyone on the team go through the test. I went through it myself before seeing the questions, and then tweaked it afterwards removing things I thought weren't going to give a good enough signal to noise ratio. It's not like we're pulling up the ladder behind us or something. Maybe we have high standards, but I've worked on teams before where hiring was less strict and there are people capable of getting negative work done. I'm not going to risk having these people on my team, even if it artificially bumps up my hiring numbers.
https://danluu.com/hiring-lemons/
https://danluu.com/programmer-moneyball/
Which you should read, along with the multiple other things linked from them about interviewing and hiring.
I can make a very similar argument from long experience, but Dan is more HN-trendy and thus more likely to be believed.
Unless you've been exposed to the raw resume data that gets sent, you have no idea what I'm dealing with, and how unqualified the majority of applicants are. I'm an engineer myself, from an unknown university, in a foreign country. My degree isn't even in CS, and I had better problem solving / coding skills straight out of high school than the majority of the applicants.
We on the team all went through it retroactively, and all got decent scores.
And actually, we aren't having any issue hiring - it is ticking along nicely and I couldn't do more interviews while maintaining my sanity than I already am. I brought it up because I'm still surprised how much gender bias there is among the applicants.
Select all of the following which manage the lifetime of an object:
shared_ptr
unique_ptr
raw pointer
Assuming foo() prints "foo" and bar() prints "bar" and both return false, what would be printed from the following expression?
if(foo() && bar()) {std::cout << "baz";}
Plus 2 small algorithm problems that can each be solved in 5 lines of simple code, which they have 15 minutes each to do (top candidates take under 5 mins). The environment gives them a place to write tests (with some basic tests already written) and lets them compile their code and run tests as much as they want before they submit, with normal compiler messages if it fails to compile and also any debugging messages they put in their code printed. There is also a tutorial that they can repeat as many times as they want before starting the main test.
In later interviews I have mentioned to candidates that others couldn't solve the pre-screen problems, and they are always surprised that people so unqualified are applying for the job.
I've seen such a large difference in ability on the interviewer side. In a three tier interview process per 100 people:
60 fail the 15 minute phone screen. Just voice, no programming, just describe some OO concepts like encapsulation, interfaces and inheritance.
30 fail the first in house test. This is writing a for loop to reverse a string and describing at a high level how to sort a 2gb file with 1gb of ram. Most fail at the string reversal.
8 fail the second in house test. You draw up an architecture of a basic app. It's open ended.
For every 100 developers about 2 know their stuff well enough to write decent software.