Wednesday, May 3, 2017

Facebook, Heal Thyself.

The tech titans of Silicon Valley like to preach the gospel of political correctness. The great majority of them lean left and they firmly support diversity. That is, they support it everywhere but in their own shops.

As the proverb says: Physician, heal thyself. Apparently, the message has not made its way to Facebook, the tech behemoth whose Chief Operating Officer, Sheryl Sandberg has become a crusader for gender equality in the workforce.

Apparently, she had too much time on her hands. So she set out to teach women to get what was rightfully theirs, by leaning in. She never considered that many women did not want to become like Sheryl Sandberg. And she did not imagine that her advice, which prescribes macho posturing, was as likely to cause trouble as to garner raises. Being assertive, aggressive and in everyone’s face incurs risks, especially for beings who are constitutionally weaker than their male counterparts. Besides, team players do not lean in. Being overly concerned about yourself does not often go down very well in a corporate environment. As articulated, it was bad advice.

One adds, as a point of interest, that the tech company that has made the greatest strides toward gender equity—Marissa Mayer’s Yahoo!—is being sued for discriminating against men in promotion practices. We do not know why Mayer failed to revive the struggling company she took over, but one suspects that other tech companies will not be rushing out to emulate her example, thus to hire and promote more women. And corporate boards in the future are likely to think twice about hiring a pregnant CEO.

Now, the Wall Street Journal reports that Facebook employees have discovered insidious forms of gender discrimination in their own ranks. The horror of it all.

Apparently, female coders do not receive the same level of respect as male coders.

The Journal explains:

Last year, a longtime engineer at Facebook Inc.FB 0.21% gathered data that revealed a controversial finding: Code written by women was rejected much more frequently than code written by their male colleagues, according to people familiar with the matter and screenshots of internal discussions viewed by The Wall Street Journal.

For many female engineers at Facebook, the finding confirmed long-held suspicions that their coding faced more scrutiny than men’s.

The results touched off a debate within Facebook over alleged gender bias among some of its most-valued employees: the engineers who build the features used by nearly two billion people every month.

What was causing this disgraceful sexist discrimination, which was happening under the nose of the ever-watchful Sandberg? How did she, great feminist icon that she has fashioned herself, not notice? Was she too busy promoting her books and crusading for the feminist cause?

Facebook was on the case. The company launched an investigation. It discovered:

In an internal post published a month later, Jay Parikh, Facebook’s head of infrastructure, attributed any gap in rejection rates to an engineer’s rank, not gender. Many employees interpreted this new analysis as a sign that female engineers weren’t rising at the same rate as men who joined the company around the same time.

The higher ranking engineers are mostly male. It was an Ah-ha! moment. The discrimination was even worse than they had imagined. Male engineers were getting promoted more than female engineers. The only rational conclusion, to the politically correct mind, was that this situation had been produced by sexist bigotry. Of course, one might also suggest that the best coders were getting promoted more than those whose work was of inferior quality. Or else, women might be spending less time writing code because they have pressing family obligations. Or else, some people might be better at coding than others. After all, the situation as Facebook is replicated throughout Silicon Valley. Are they all a bunch of sexist bigots? And how does it happen that their companies, rife with discrimination, are doing so well in the marketplace?

What if women do not want to code? What if women are less interested in engineering, as a course of academic study or a career? One recalls that women in college are more likely to major in subjects like art history and psychology while men are drawn toward engineering programs. Ought we not to respect the choices that women make, freely, one hopes, about their careers?

So, we are supposed to be astonished to read that code written by women at Facebook received 35% more rejections than code written by men. Did anyone ask whether the code written by women was as good or better or worse than the code written by men? Is gender the only relevant factor? If both men and women were doing the same coding for the same application, then we might have a fair comparison, but what company has two people doing the same job? Or else, the company can ask two executives to review the same code, telling them that one was written by a man and the other by a woman. Doubtless that would do wonders for company morale.

Also, the Facebook analysis also did not take into account an engineer’s relevant pre-Facebook experience. An engineer who had worked in the field for five years before coming to Facebook would probably be better than someone who was beginning a career at Facebook.

Facebook and the rest of Silicon Valley show what happens when deeply felt ideals run into reality. Either you sacrifice the company to your ideals or you paper over the problem with noble sentiments and calls for people to overcome their bigotry. We certainly would not want to think that all of these committed diversity-mongers are really just ordinary hypocrites? Would we?

James Freeman summed it up in a Wall Street Journal column on the topic:

From everything we’ve been told, there is absolutely no shortage of highly sensitive progressives at the social-media giant, and yet the organization doesn’t seem to be leaning in that far toward a more diverse workplace. There is of course an old-fashioned argument that an organization should simply seek to hire and promote employees based on the content of their character and their talent and forget about what people look like on the outside. 

11 comments:

  1. Well, there are a lot of reasons why a block of code could get rejected, ranging from 'the fundamental logic here is unworkable' to 'didn't you mean variable xusername rather than yusername on line 7100?' to 'the comments on this code are hard to follow.'

    The article makes it sound like having code "rejected" is some kind of soul-shattering experience, but I would think in most cases it's just a matter of fix something and submit it again.

    It would be dangerous to have reviewers being nervous about rejecting code based on sensitivities (about gender, about organizational level, or anything else) for the same reason it would be *really* dangerous to have airline copilots being nervous about pointing out something like "Shouldn't we have be using 10 degrees flaps for this takeoff length?" to the captain. (Indeed, the latter phenomenon had been occurring with sufficient frequency and devastating results that the NTSB and FAA encouraged the development of Crew Resource Management programs, which among other things encouraged crewmembers concerned about a safety-related issue to *speak up* regardless of the relative ranks involved.)

    ReplyDelete
  2. Thank you... very good observations!!

    ReplyDelete
  3. Feminists, as I understand it, believe that biology is everything. Therefore, ANY rejection of a woman or her work is only and obviously due to misogyny. (No other explanations need apply. Go away. Be gone!)

    ReplyDelete
  4. Stuart: So, we are supposed to be astonished to read that code written by women at Facebook received 35% more rejections than code written by men. Did anyone ask whether the code written by women was as good or better or worse than the code written by men? Is gender the only relevant factor? If both men and women were doing the same coding for the same application, then we might have a fair comparison, but what company has two people doing the same job?

    Indeed. I've never heard of such statistics. As a programmer, I've never even heard of "code rejection". Usually code is only rejected because it doesn't work, doesn't do what its supposed to do, or causes unacceptable side-effects, and a programmer can figure most of that out for his or herself.

    I suppose bigger companies have testers, but I don't know why a tester would necessarily know the gender of the programmer.

    Of course there are two issues in programming - visible functionality for users and aesthetics for the code itself, and I've rewritten other people's code when it fails to meet my standards of sanity.

    You know, things like putting calculations inside of a loop that are independent of anything that is going on inside of the loop. I've also redesigned user interfaces components that were written in a certain way because it was easy to program, but made more steps for users.

    Rewriting other people's code might be a higher insult than rejecting it and making the author do it again, which might actually teach them to be a better programmer.

    I recall one lecture a guy suggested programmers should work in pairs, and he imagined they'd help challenge each other become better programmers. I do think this is true for testing each other's code, since it is fun to try to break someone else's work by acting dumb, like entering a negative number or illegal one, like April 33. I could see value of reading someone's code and offering feedback, but all of that takes more time, so it only really makes sense when you discover original programming can take 20% of the time and debugging can end up 80% down the line.

    Or I remember in my data structures class my prof said the average programmer writes 50 correct lines of code per day, which actually made me feel better at taking my time. But perhaps he was referring to assembly language.

    ReplyDelete
  5. Stuart: What if women do not want to code? What if women are less interested in engineering, as a course of academic study or a career?

    These are rather sexist questions, presuming all men are a like, and all women are alike, and their bell curves of interests have no overlap at all.

    What if people without nerdy glasses and bad skin do not want to code? What if people who are less nerdy are less interested in engineering?

    Indeed what if?! I have no idea what to say about that except that we shouldn't act all communist and assign people jobs they don't want to do.

    Even Donald Trump seems more progressive than Stuart. Trump said a woman was worth 4 men because he knew they had to work harder to prove themselves.

    ReplyDelete
  6. "These are rather sexist questions..."

    Ares, why don't you check the FACTS you're so rather fond of, and see what they say. There's no doubt that men have a greater preponderance of being in a profession like fixing lawn mowers. A sentient being might also extrapolate that men also are more likely to engineer physical realities on a macro scale.

    Open your eyes, man! You're so easily offended by words and suppositions. You might want to check to see if reality comports with those suppositions. Women aren't as interested in porn as men are, either. Is that sexist???

    Answer, please.

    ReplyDelete
  7. One thing in the article that's a little confusing: "The code shapes the user’s experience, and experts say the absence of diverse viewpoints on a coding team can lead to a product ignoring the needs of certain categories of users."

    Normally the way things work with software is that there is a detailed functional specification that defines how the product or feature should interact with the user *before* the code gets written. That specification can be written by someone other than the person who will write the code...a marketing person or a systems engineer, say....or can be written by the same person that will write the code, but before starting the coding. The implication of the quoted statement from the article is that at FB the user interfaces are defined on the fly as code is written...if this is really true, then the Diversity argument makes more sense. (Only I would argue that it's Diversity of personality types, educational backgrounds, etc, not just the holy trinity of Race, Gender, and Sexual Preference)

    ReplyDelete
  8. Ignatius Acton Chesterton OCD said... Ares, why don't you check the FACTS you're so rather fond of, and see what they say. There's no doubt that men have a greater preponderance of being in a profession like fixing lawn mowers. A sentient being might also extrapolate that men also are more likely to engineer physical realities on a macro scale. Open your eyes, man! You're so easily offended by words and suppositions.

    I'm not offended by gender differences. I'm offended by meaningless generalizations that have nothing to do with the situation at hand, i.e. women who are programmers.

    Stuart asked "What if women do not want to code?" as if that sentence could mean anything. If I asked "What if women did not want to cook?", what could that question ever mean? It's an abstraction, grouping all women together, and imagining a single "want" among all of them.

    Some women like to cook, some don't. Some men like to cook, some don't. Some cooks like to cook, some do it more for the money, although actually most cooks really aren't paid well. Perhaps one gender has a higher proportion who like cook. Perhaps one gender has greater skills on average by whatever standards you might imagine, who knows! Even if one gender likes to cook more than another, it says absolutely nothing about an individual man or an individual woman, neither their interest nor skill. Absolutely nothing.

    So a blanket statement without qualifiers is either saying nothing at all or is implying rhetorical nonsense.

    The only clear reason to rhetorical nonsense is because Stuart has a bias - in this case demeaning individual women who have chosen to enter a male-field and become programmers for their own reasons we don't know and shouldn't need to know to explain questionable statistical outcomes.

    Why speculate at all on that "wants" of women who are there doing a job?

    Of course feminists can be sexist as well, when they assume statistically gender-divergent outcomes implies gender-divergent treatment. But at least that's a meaningful question like any bias.

    Again, if I were to guess, I'd follow Trump's opposite speculation and suggest that a woman in a man's field will tend to do higher than average quality of work, because she needs a higher passion to stick with it and she had to jump through more hoops to prove herself.

    Once these questionable statistics are better squared, they may find once they account for age and experience, women might find themselves having their code is less often rejected than men, and fools can speculate in the opposite direction - that we're afraid of hurting women's feelings or whatever.

    That's the thing about bias, you start with a conclusion, and whatever the "facts" seem to be in the moment, X or the opposite of X, you can speculate in the same direction, X implies Y, not X implies Y.

    ReplyDelete
  9. Ares: "I'm not offended by gender differences. I'm offended by meaningless generalizations that have nothing to do with the situation at hand, i.e. women who are programmers."

    It was a post about Facebook and women's opportunity there. Facebook is a technology company based on computer programming. That is the situation at hand.

    Unequal results do not always mean bias, unless one believes men and women are the same by nature, and are nurtured into biased results.

    ReplyDelete
  10. IAC, There is nothing you just said that contradicts what I said.

    ReplyDelete
  11. Mr. Schneidermann,

    Check out the work of Canadian academic Jordan Peterson. he has done a lot of work on the workplace and the roles of men and women. The piece i am thinking of is a study he did of Scandinavian countries, where every conceivable barrier to womens participation in non-traditional occupations has been removed...but, his findings indicate that women still prefer, for the most part, jobs in teaching, nursing etc.

    ReplyDelete