Tuesday, December 2, 2014

The End of Google?

No, It's not that European law-suit. This post from "Jim's Blog" (tag line "Liberty in an unfree world") holds that political correctness, partially driven by Google wanting to get cozy with the government, has led to the hiring of unqualified women and the firing of males who told the truth about it:
Google had an elaborate system of metrics to try to measure how people and teams performed, and these metrics showed that the women were no good. Therefore the metrics were sexist, since they had disparate impact. They were replaced by non sexist metrics, which metrics showed that women are wonderful. Metrics that show advantage men are disparate impact. Metrics that show advantage women are just a reflection of the fact that women are wonderful.
Sexist males were laid off, where sexism was largely manifested by doing work that females could not understand or participate in. Which is to say, they laid off their smartest males, in part because smartness tends to exclude women, in part because the new metrics that showed that women are wonderful also showed that their smartest people were doing a really bad job.
The blog is pretty dead-center of the NRx (Neo-Reactionary) school of thought with the idea that genetics ('human bio-diversity') dictate intelligence and that traditional gender roles (where women are certainly not doing math and science) are objectively preferred.

The interesting part comes in the comments where someone purporting to be an actual employee of Google declares the whole thing bullshit. Blog-author Jim, explains to the alleged Googler that Google has two disparate hiring processes--one that hires men and another, secret one, that hires unqualified women and minorities.

No, says the apocryphal Google-Employee, that's not true--or, if it is, he sees no sign of it. He's hired women through the normal process and no unqualified woman was hired on his watch. He wants to know how Blogger-Jim knows ("knows") all this stuff? Jim says he has friends who work at Google who tell him this. Who he can't reveal anything about.


So, Is Political Correctness Destroying Google?
No. The Omnivore actually knows people who work at The Googles and reached out to them. While one couldn't comment (He Did Nothing), even off the record (because of their position this was not unexpected) the story is bullshit (The Omnivore has also been used as a reference for a female Google Employee and is pretty sure the interview process The Omnivore participated in was not part of a secret hiring cabal). Even the employee who couldn't comment was scornful.

These are people The Omnivore has known for decades (in one case, uh, many, many decades) and would have been able to provide nuance if these allegations were true. Google is not more dedicated to social justice than it is to being incredibly smart (one of the Google-friends contrasted Google to Akamai --another "smart people company" where she had also worked--and said the best Akamai programmers were average at Google. The Omnivore believes her).

That said, of course, there's no way to know, is there? The Omnivore admits that the article--with its prescriptions of DOOM and focus on the injustice of Social Justice has a certain ... Truthiness ... would you say?

After all, if you go read the article there's:
  • No sourcing. Everything is presented as commonly-held fact.
  • Zero nuance. Everything has one cause (political correctness) and one effect (firing the smart people).
  • Evidence to the contrary (Google's valuation has reached an all time high in the past 12 months. If we assume markets are good for setting prices, what of that?)
This is a very strongly present argument for something that the author (a) admits they only know second hand (if even that--the sources never come up in the body of the article--only after Jim is challenged) and (b) are incredibly impactful (Jim is drawing bright-line conclusions from what would be statistically limited policies: Google does employ women--but not nearly as many as white men).
We Don't Know What Those Jobs Are, Either

While it's fair to say that Jim's blog, much less the entire NRx movement, doesn't depend on getting some facts about Google right, The Omnivore thinks this post--the response from a Googloid--and the author's refusal to either back down or back up his assertions is telling. Why?

Well, if there's one thing that NRx absolutely needs at its foundation it's this: it has to be smart and right. If it it's wrong, say, about human-bio-diversity (black people are genetically less intelligent) then instead of being a teller of hard truths in a self-blinded world, it's just recycling racist tropes at the speed of the Internet. That's not the way to bring the NRx Utopia into existence (read it!).

Warning: Objects Under Consideration May Appear Smarter Than They Are
What The Omnivore was struck by in the blog post was the utter surety with which the author was making extreme, questionable, and unsourced claims. Yes, this is kind of "What everybody does"--but the NRx isn't supposed to be everybody. In fact, that's kinda the point-of-pride. Secondly, NRx readers are supposed to be Less Wrong, overcoming cognitive bias types who, as the song goes, Won't Get Fooled Again

So what's going on here? What if the NRx was formulated (unintentionally) to play on cognitive biases the same way optical illusions prey on visual 'software errors'? For example: We believe mean statements to be smarter than nice ones. We are hardwired to badly evaluate risk (and favor more 'glamorous' risks over more mundane ones). Our culture is rife with end-time scenarios and they are popular. We are susceptible to flattery--even if we know we're being flattered.

The NRx memes play on all of these (if you are in the NRx target market, anyway) and the Decline and Fall of the Google Empire does too. It predicts disaster in the meanest of terms (if not for the world, for the Internet--and if not just for Google, all smart companies who will fall to the disease of Social Justice). It overestimates risk-factors badly (even if those 2% black Google employees were senior engineers, how much damage would they do to a company of more than 55k people?). 

And, finally, if you are a white male who feels even a microscopic decline in their privilege? It flatters you--you are special. You will be elevated when the End Times come (GNON! GNON! LET IT BURN!!).

What if that's all just an illusion?


  1. [veronica here]

    Just to drop a note, I think Less Wrong is a pretty great place, at least the sequences, which are indeed very smart. I'm less keen on the forums as they do include some fairly noxious people. On the other hand, Ozy Franz has seen the latest survey data and there are more left-leaning people on LW than neo-reactionaries. In fact, there are more transgender women than NRx's -- and look, those woman cannot all be Justine Tunney types.

    (Speaking of which, I wonder what that blogger thinks of *her*? The downfall of Google? Its savior? Inquiring minds and all...)

  2. I think this is pretty spot-on. I'm no NRx expert, but this fits my knowledge of them perfectly.

    As for Less Wrong, I don't think it's actually neo-Reactionary. But it's the kind of thing that neo-Reactionaries like. After all, it is very smart-looking.

  3. According to reports which I just made up, Sergey Brin and Larry Page agree that Jim's claims are 100% weapons-grade bolonium.

    See what I did there?

    Isn't Ms. Tunney the one who petitioned the government to make Eric Schmidt the CEO of America? If so, I (and others, I'm sure) often have difficulty telling when she's trolling / winding up conservatives from when she's expressing her actual opinion(s). Maybe that's the whole point?

    Ah, these youngsters and their odd ways of expressing themselves.

    @Ommie: "many, many decades"? C'mon dude, I know how many growth rings you've got and that seems kind of a stretch, unless perhaps you interpret "many, many" differently than I do (I maintain that you can't count to "many, many" on one hand, but what do I know). Or maybe I don't know your other persona, the one who goes about decapitating people whilst shouting "there can be only one!"?

    I interviewed with Google a few years back, as it happens. They asked me a couple of their (in)famous (and just as famously useless) brain-teaser questions, of which the main one I remember was to estimate the weight of the ice in an NHL hockey rink. As I recall, I got it right to within a factor of two or three, which would of course have been impossible for any mere woman who'd never heard of hockey and/or didn't know what ice was.

    So I've got that going for me, which is nice. But I didn't get the job, because it appeared that they were interested in other stuff too besides my ability to deal with meaningless interview-bullying tactics. Oh well.

    -- Ω

    1. Well, approximating 4 decades. That's a lot to me!

      -The Omnivore

    2. Sure. The way I see it, "many many" ought to make you bust out some exponents or at least take off your shoes. And I'd imagine that anyone you'd known for approx. 40 years is fairly likely to look like you - because folks who don't remember the '60s all look alike...

      Yeah, that's it.

      -- Ω

    3. [veronica]

      For the record, they don't ask the brain-teaser questions anymore. They found they're a pretty useless predictor of actual talent. Which kinda seems like a "duh" to me, but there is no end to the bogus folklore in tech interviewing.

      Anyway, these days it's all coding-on-whiteboards and stuff, along with a crapton of analytics on the backside.

    4. [veronica again]

      And on Justine Tunney, yeah, it's obvious she's turning the knob all the way up to "so crazy it's funny," but she seems the type to quickly jump from one extreme idea to the next, half for shock value and the other half cuz it tickles her weirdbrain. (I say as a girl with a weird brain.) But the whole thing is pretty toxic.

      That said, she's adorable, which has to count as a plus if for nothing more than the degree it must confuse loveshy NRx-ers.

    5. I all kinds of agree about the prevalence of "bogus folklore in tech interviewing", up to and including the distinct possibility that the great majority of them are completely counterproductive. I recall a recent thread on in which a frustrated tech manager complained about having lost out on getting to hire quite a few promising IT professionals due to their getting weeded out by HR types with no domain knowledge (or respect for it). That pretty much squared with my already low opinion of their "profession". I know that I couldn't do what they do - they have to deal with people at their worst with dismaying frequency, and I'm happy to leave them to it. But I greatly dislike their all-too-common lack of transparency and accountability, especially during the hiring process. The manager's point was basically that HR folks were fine for screening candidates for jobs whose personae and skill sets were broadly similar to theirs - soft-skill-intensive positions in sales, marketing, HR itself, etc - but no good at assessing STEM types, who tended to do relatively poorly by those same criteria.

      Anyway, I'm glad Google isn't asking those questions any more, even though I tend to do well with such things. If I were hiring an engineer, I'd be more interested in "what can this person do?" as well as the necessary "could this person fit within my existing team?"

      Best technical interview I've ever had was for a software QA position last summer: after getting the social-fit stuff out of the way, I was given a laptop with PuTTY and Firefox on it, and told the hostname of a KVM containing a broken Apache web server. I had to stand up the server in 15 minutes. There were a number of problems as I recall: there was a typo in httpd.conf, a faulty IP address in the host table, the route to it was missing, stuff like that. So it wasn't all about rote memorization, which is all too typical in tech interviews; it implicitly acknowledged that we all use Internet resources to look up things like error codes, but what mattered was whether we had the fundamental tools (stack backtraces, grepping logfiles, etc) and could put it all together in a realistic situation under time pressure.

      Sorry for going kind of off-topic there, but interviews are kind of a sore spot for me lately :-(.

      -- Ω

    6. [veronica]

      Reminds me of some of the Cisco tests I took ages ago: Log into this router. Fix it.

      I liked those.

  4. I'm totes stealing the section on cognitive biases...