#biodeterminism

So let's talk about this Google employee's "anti-diversity" memo. And I mean talk about it, not just freak out over it.

(If you haven't read the thing itself, do. If the author is right about one thing, it's that different points of view deserve to be discussed on their own merits, rather than dismissed up front for seeming outrageous.)

And let's not kid ourselves, whatever that memo is, it is definitely not hate speech. The author makes his argument in a polite and nuanced ways, clearly states his intentions (which are far more benign than the "women don't have a place in tech" reporting makes it seem), and makes some legitimately good points. He still gets a couple of crucial things wrong, and that's what I want to address in this post.

First, here's a basic breakdown of the author's argument as I understand it:

PREMISE 1: Statistically, men and women as a group are interested in (/good at) different things on average. For example, men are more likely to be "thing-oriented" and women more "people-oriented". (strong, with caveats)

PREMISE 2: This difference is rooted in biology and unlikely to change because of social measures. (has a point, but is not the whole story)

PREMISE 3: The purpose of diversity measures in companies such as Google is to make sure that there are as many women working in every field as there are men. (very weak)

CONCLUSION: Trying to enforce equal representation while drawing from an unequal population is tilting against windmills at best, and counterproductive at worst. (shaky as hell)

(You might notice that I'm not going into the whole "nobody's allowed to say that" and "diversity of opinions" self-victimization thing here. That's a rhetorical device, not an argument, and also it is not what most people seem to be reacting to.)

Let's start with premise 1. We don't need to debate the factual point; of course men, taken as a group, have different preferences on average than women, taken as a group. If you pick a random woman out of the general population, the probability that she is interested in [STEM stuff] is lower than it would be for a randomly selected man. Duh. But there are two caveats here:

Caveat 1: What is true for the general population is not necessarily true for those women who apply for jobs at Google. If you pick a woman who applied for a technical position at Google at random, your priors that she's interested in / good at it should be the same as for a man who applied for the same job. Why? Because selection effects, obviously. Women who apply for jobs at Google are not a representative sample of the female population. Also duh.

Caveat 2: Be careful about your constructs. I haven't looked into the literature, but "people-oriented" and "thing-oriented" strike me as both very general and very loaded ways to frame the difference. I'm not saying it's not a valid distinction to make, but we need to remember that there is a difference between the actual data (which can be made sense of in multiple ways and usually presents a complex picture) and the labels we use to carve it up.

Which brings us to the second premise. Yes, men and women today have different preferences; but where do these preferences come from, and are they necessary and unchanging in the way the author suggests?

Now, I will not discuss evolutionary psychology here. EvoPsych has a lot of interesting things to say, but I will not get bogged down in just-so stories about how this or that might have been adaptive at some point or not. For the most part, these have about the same predictive power as psychoanalysis, meaning none; they serve to "explain" a status quo, but you could tell a completely story and "explain" the same data just as well. Let's not do that here; let's please leave EvoPsych to the professionals.

But I have to discuss reductionism. There is this tendency among people who've read a little EvoPsych, nicely exemplified by the manifest's author here, to look at some feature of the world, find an evolutionary explanation for it, and then say "that's it, feature explained". That's not only a cheap way to shave off complexity, it also reifies the status quo in a way that does not allow for human agency. Once you accept that you can explain the same facts in different ways, choosing between competing explanations becomes an engineering problem: Which of these explanations can I actually do something with? If an explanation leaves you helpless and cynical, saying that "things just are like that, there's nothing you can do", you are probably operating at the wrong level of detail.

So let's have some (still very low) complexity. Sure, if a broad distinction can be observed for men and women all across the globe, it is likely to have some sort of biological basis. (Though you could also say that patriarchal societies are dominant pretty much across the globe... but let's concede the point here, because it doesn't make much of a difference.) But that's not all there is to it. Society (and, on a finer level, human psychology) takes these biological tendencies and builds on them. It weaves them into narratives that evolve over time and tell you how to behave, above and beyond what biology alone "requires", and changing much quicker than biology itself. We build identities and scripts and rituals, based on and influenced by our biological tendencies, sometimes exaggerating them, at other times sublimating them, warping them, even resisting them. (Just look at how many religions are built on a foundation of resisting some of our "natural impulses".)

What does that mean for diversity programs, then? It means that we do not, as the author suggests, simply have to resign ourselves that a lack of diversity in a given institution is biologically determined and there's nothing we can do about it. Some of the root causes may be biological, but that does not mean we cannot or should not examine our institutions to see where they don't conform to our values. Because that is what should be the driving force behind our aspirations: realize our values, based on a good image of reality; not looking at "reality" with a reductionist lense, shrugging and saying, "well, bad luck".

Of course we can, and should, debate those values. You can say that "having equal numbers of men and women in all professions and positions" is a stupid terminal goal to have, and I would probably agree. However, the purpose of all diversity programs that I know (I don't know much about Google's, but I'd suppose it's similar) is not to achieve a perfectly balanced male-female ratio, but simply to make it easier for those who tend to have it harder (because they're in the minority, because the work environment was not designed with them in mind, or whatever) to do the job that they are good at. In other words: (good) gender diversity measures are not about getting more women into certain positions at all costs, but about making it possible for qualified women to get there, without having to expend too much energy on surviving in a hostile environment. So much for premise 3, which is nothing more than a straw man. It's easy to attack your enemy if you misrepresent their position as something extremist and stupid as "we need more women here, no matter how qualified they are".

Which brings us to the conclusion. If enforcing equal representation is not the goal, and if corporate culture and institutional design have an effect beyond the biological base rate, and if the pool of women who apply for a job at Google is not representative of that base rate anyway... then yes, we can do at least some good by making it easier for those women who are interested in and good at [STEM stuff] to get into positions where they can use their talents without constantly fighting an uphill battle. Or more generally: If humans have any agency beyond the behavior "dictated" to them by their genes, then we can call them out on what they do and don't do, and they don't get to hide behind biology to say "can't help it".