Categorizing People by Race (Spot the Reference to Trout Fishing in America!)

There is an ongoing three-way internet conversation between William Saletan, John McWhorter and Steve Sailer about whether people should be categorized by race in government and social science datasets. This recent contribution by Saletan has the linkage; as a bonus, here's John Althouse Cohen's take. They are talking about the black-white achievement gap in the current US, but as far as I can see, the arguments are applicable more widely. As you've all been dying to hear my opinion, here it comes.

Sailer writes:

The reason people all over the world and of all different ideologies can't help but be interested in race is a racial group is, fundamentally, an extended family. So, race is about who your relatives are, which is an inherently interesting topic.

I'm not buying that. I have no problem with Sailer's definition of a race as an extended family. I also have no problem with the statement that we're interested in who our kin are. And I'll even say that's hardwired. As the quip goes, however, I wouldn't die for my brother, but I would die for two of my brothers or four of my nephews. In other words: The degree to which we care about our kin decreases with the degree of relatedness (we're talking averages here), and the degree to which I'm related to a person plucked at random from a pool of people who share my race but are otherwise unrelated to me is trivial, so Sailer's argument doesn't work. Empirically, here is a paper by Robert Kurzban, John Tooby and Leda Cosmides presenting data which the authors argue shows that classifying people on the basis of their race when encountering them is not inevitable.

Having said that...

In my ideal world, everybody would be colourblind. In the real world, however, even if there would be no collection of systematic data on race and all old data would be destroyed, unsystematic data would still be collected by people perceiving the world around them on an everyday basis - and race happens to be a category people use. Not only that, but people, being the untrained social scientists they are, will develop theories about the patterns they observe. Black people earn less than white people? Black people are lazy! White people are racist! It's the genes! It's the environment! And so forth.

That's why, I think, you need systematic collection of data: 1. To give people a more accurate representation of reality than their everyday observations. (What?) 2. To better be able to test competing theories on the reasons for observed differences. (Why?)

I consider it inelegant to end a text with parentheses.