So I’m going to move on to the topic of affirmative action. But before I do, I need to make one more point of clarification. Before I can explain why we need to be much more careful about affirmative action policies than we currently are, I want to differentiate between three different things we call discrimination. The differences are important because I’ve found that most arguments in support of affirmative action treat all discrimination as if it were essentially the same kind of thing.
So what are these three types? The first, of course, is hard discrimination. This is what most people think of when they hear the word “discrimination.” I guess I’ll attempt to give it a rigorous definition. I’ll define it as the judging of a class of persons’ capabilities in a manner contrary to or without regards to a test of capability, despite using such a test as the basis for judging another class of persons’ capabilities. If the mathematics department won’t even let you take the entrance exam because you’re female, that’s hard discrimination. If the coach cuts you from the track team because you have asthma, even though you outran everyone else at tryouts, that’s hard discrimination. If someone with a legacy is admitted to the university despite having a weak academic record, that’s also hard discrimination (this time a case of positive discrimination). This is the kind of discrimination women and racial minorities routinely faced in and before the first half of the twentieth century. It’s the kind of discrimination the first wave of feminism and the civil rights movement fought against.
(In a “happened to me” example, when an internship program won’t even let you apply because you’re a white male, that’s hard discrimination. We just call it “affirmative action” because it favors women and minorities).
The second type of discrimination is what I’ll call nondiscrimination, because discrimination is actually a really bad word to use. Nondiscrimination is the things people call discrimination but are actually just expressions of either biological differences or genuine hard discrimination that took place elsewhere or elsewhen. For example, a pharmaceutical company could have very few black employees because there are very few black people obtaining pharmaceutical degrees, rather than any racist hiring policies from the pharmaceutical company. And if you look at schools, it could be that very few black people get pharmaceutical degrees because very few apply, and this could be because lots of them drop out of high school, and so on down the line. In the end, it turns out that there’s a strong correlation between ‘black’ and ‘economically challenged.’ A disproportionate number of black people come from economically challenged backgrounds, which makes it harder for them to get a pharmaceutical degree. So while it looks like there might be some racism going on, a closer inspection might show that the race disparity of the company’s employees isn’t actually caused by any current racism. Yes, the economic disparity might have been caused in part by the oppression of the African Americans’ ancestors, but that doesn’t mean we should accuse today’s pharmaceutical companies of being racist.
Another example of nondiscrimination is where a socially-created disparity causes what looks like discrimination in some process that's not actually involved in creating the disparity. For instance, it may be that the apparent differences in mathematical performance between men and women are caused by social conditioning. However, if this conditioning does in fact lead to fewer women meeting the criteria for your mathematics program, then the fact that you have fewer women in your mathematics program does not mean that you are engaged in discrimination. There is discrimination somewhere, but it’s not in your program. Your program is instead engaged in nondiscrimination. It looks like you’re discriminating against women, but your gender ratio is really just a reflection of a discrimination which takes place elsewhere.
The third type of discrimination is called soft discrimination. Soft discrimination is where you use a nondiscrimination in an initial judgment before any performance data is available. If I see that you’re female, I have absolutely no performance data available, and I decide that you probably won’t perform well in the mathematics department, that’s soft discrimination. If I refuse to let you take the entrance exam or if I continue to believe that you won’t perform well despite seeing your high score on the exam, then I’ve crossed the line into hard discrimination. Soft discrimination is when you get more scared passing a big black guy in a dark alley at night than when passing a little white women in the same dark alley. Hard discrimination is when you still think the black guy is a criminal even after you see his police badge.
Sometimes soft discrimination can be perfectly justified. If only four percent of women who take the entrance exam pass, then I’m making an honest assessment when I say you probably won’t pass the exam. And if crime statistics show that black guys are significantly more likely to attack you in a dark alley than white women, then I’m justified in being more cautions. The key though is that the judgment needs to be temporary. It needs to be an “I don’t have enough data so here’s my best wild guess” sort of thing. Once you get the data, once the black guy shows you his badge or the woman passes the entrance exam, you need to update your assessment. If you don’t, or if you refuse to even look at the data, you’ve crossed the line into hard discrimination.
So now all of that is out of the way, it’s time to address the concept of affirmative action. It seems to me that much of the push for affirmative action comes from two desires. The first is a desire to address hard discrimination, while the second is a desire to address the underlying causes of nondiscrimination. These are both fine causes, but the problem is that affirmative action policies are often enacted where the “discrimination” taking place is actually nondiscrimination.
Let’s take our hypothetical mathematics program as an example. This program has an entrance exam, which anyone is allowed to take. If you don’t pass the exam, you’re not in the program. If you pass, you’re in the program. I know this is probably a simpler procedure than most programs have, but the simplicity is here to make the point clearer. Let’s say that due to biology or social conditioning or both, women rarely pass the entrance exam. The affirmative action response tells the people running the program to admit more women in the interest of gender equality. There are four problems with this approach.
The first problem is that the affirmative action application is way too late. Let’s assume that all of the mathematical ability differences between men and women come from social conditioning. Even in this case, by the time women get to the point of taking the entrance exam, the damage of social conditioning has already run its course. The women are already less qualified. Letting them in isn’t going to suddenly make them more qualified. In fact, it’s questionable whether or not such a policy will help alleviate the social conditioning in the long run. The idea is that future generations can say, “Look, female mathematicians! Maybe girls can do math after all!” The hope is that this kind of attitude will, over time, address the social conditioning that causes the nondiscrimination in the first place. But I seriously doubt this is actually going to work.
What happens when you take a mathematics program and set a lower standard for female applicants than for male applicants? You get female mathematicians, sure, but are they good mathematicians? If you let in a bunch of underqualified women, will they actually succeed in the program? Do you have to water down the program so they can keep up? Do you have to keep on giving them preferential treatment the whole way through? And if so, won’t, “Look, female mathematicians!” just be met with “Only ‘cause we made it easy for them!” If we let a bunch of underqualified women become mathematicians, I think people will start to notice that they’re all underqualified. You will be saying, “Women can do math,” and yet to try and showcase this point you will be highlighting women who aren’t as well-qualified as their male colleagues. Won’t this just make matters worse?
The second problem is that you’re placing the burden of fixing discrimination on the shoulders of the nondiscriminators. The mathematics program has exactly the same criteria for men and for women. In fact, it could very well apply its standards without even looking at whether the applicant is male or female. If you give everyone an identification number, then you just have to see which numbers passed and which failed and then you accept those who pass. You can make it literally impossible to perform hard discrimination by eliminating the information it’s based on.
(Incidentally, this is how science removes bias. As far as I know, it is the only surefire way to remove a bias. You can try to apply as much anti-bias as you want, but it’s really difficult to know how much is enough and how much is so much that it causes an overall bias in the other direction. It seems to me that the best way to remove sexism and racism in the workplace is to make sure that the people who make the hiring decisions don’t know the race or sex of the applicants).
In effect, you’re trying to combat discrimination by encouraging people who don’t genuinely discriminate to apply hard discrimination in the other direction. I’m not saying this can’t work, but it certainly sounds fishy to me. I think we need to brainstorm a few alternatives before rushing headlong into this strategy. There’s just something very counterintuitive about the idea of fighting discrimination with more discrimination.
The third problem is that you’re trying to use a blanket fix to a problem that comes in varying degrees. It just isn’t the case that every female faces the same degree of social conditioning when it comes to mathematics. Yet if we just apply a uniform tilt to all female applicants, then we’re trying to use a one size fits all patch for a problem that comes in a variety of sizes. Now when you get right down to it, most patches do this anyway. It’s only a real problem if differences in the cause of the nondiscrimination are large. For instance, if the root cause of the nondiscrimination of our pharmaceutical company is the fact that black people tend to come from economically challenged backgrounds, then giving all black people a leg up is going to unfairly advantage the few black people who come from economically comfortable backgrounds. A much better strategy would be to offer a leg up to anyone, black or white or Asian or Hispanic or what have you, who comes from an economically challenged background. At least that way you’re addressing the actual cause of the nondiscrimination, rather than some factor that happens to correlate with that cause.
And the fourth problem comes when you combine affirmative action with the unquestioned assumption of the “blank slate.” As I’ve mentioned earlier, when you assume that every discrepancy is social, you’ll set your “no discrimination” mark at a population-reflecting distribution. If 51% of Americans are female, then most people will say that a company in which 51% of the mathematicians are female is not being sexist in its hiring process. But if there is some biological factor which makes females less likely to have extraordinary mathematical ability than males, then there is actually sexism against men (though maybe not in the company. They could just be engaging in nondiscrimination). And if there is some biological factor which makes males less likely to have extraordinary mathematical ability than females (and which has previously been overwhelmed by social conditioning), then there is actually some sexism against women. This despite the apparent equality of the situation. In short, if we aren’t willing to consider how much of the discrepancy is biological, we won’t know where the point of zero discrimination actually lies.
In short, affirmative action is a questionable tool. It fights discrimination by using more discrimination, and it often applies itself in areas where there isn’t any genuine discrimination going on. Furthermore, people are trying to use affirmative action to fix a proble, despite the fact that they never bothered to figure out what it would mean to have the problem fixed! They say they want to end discrimination, but if you aren’t willing to consider the biological factors involved, you won’t be able to tell when the discrimination has ended!
(Post Script: There’s also the possibility of good, morally justified hard discrimination. For instance, maybe we really should have separate basketball leagues for men and women even though both would be engaging in hard discrimination. But that’s a topic for another day).