So those on the petitioners’ side of Grutter v. Bollinger, the University of Michican affirmative action case now before the Supremes, have argued (inter alia, as the lawyers say) that the school’s means of weighing race is really a disguised quota—and we all know from Bakke that quotas and set-asides are an equal-protection no-no. The enormous difference between a quota and a “weight” was one I never quite managed to apprehend, but there it is in the case law.

The case, in a nutshell, is this: Michigan assigns applicants “points” to for a variety of things: three points maximum for an excellent essay, five for a history of leadership or volunteer work, and twenty points for belonging to one of several otherwise underrepresented minority groups. The rationale for this is to promote diversity, and as we also know from Bakke, there’s a bit of dicta from Justice Powell indicating that diversity might indeed pass Equal Protection muster as a “compelling” motive for considering race in public school admissions, provided it was part of an individualized assessment that made race just one factor among many.

Now, petitioners argue that because Michigan’s program seeks to maintain a “critical mass” of minority students—somewhere in the 11–17 percent range—that in fact its point system is really just a disguised quota. Instead of explicitly setting aside (say) 14 percent of the seats in the incoming class, which would clearly be verboten, the school inserts in the admissions process a race “weight” (in the form of those points) calibrated to the target percentage. Whatever the proportions you would’ve used a quota to guarantee, you just jigger the strength of the weight placed on each racial group until you get that result.

This argument strikes me as correct, but if it is, then it presents a dilemma for any diversity-based affirmative action program, not just Michigan’s. Any such program would have to be “narrowly tailored” to achieve its goal. In other words, if the consideration of race is justified in order to achieve diversity, then the way in which it is considered as well as the extent to which it is considered must be designed to further that goal. Imagine, for example, that a school were already enrolling members of a certain racial group at a very high rate. Any further bonus attached to membership in that group could not be justified on diversity grounds: it would be mere discrimination. More concretely, if whites and Asians already constituted a majority at some university, an admissions procedure that gave some extra bonus to white or Asian candidates on “diversity” grounds would not last very long. So if a preference justified by diversity is going to pass muster, it has to be tailored so actually achieve that. It must, in other words, aim to achieve at least a fuzzy target distribution of ethnic representation. If you end up with quite a few members of minority group Y, but relatively few of distinct minority group Z, your weights are miscalibrated. Your program is open to rightful challenge (and suit!) by any member of group Z on the grounds that the system must assign fewer points to group Y or more to group Z if it is to actually be a diversity-promoting one, and not merely an arbitrary system of discrimination. It must, in short, calibrate those weights pretty carefully.

But if that’s a requirement any program must meet to avail itself to avail itself of the diversity justification, then it’s not clear how it can pass the other test: that is, being an individualized assessment and not being a quota. A school cannot know how many points to give me on the basis of my last name unless they know how many other Sanchezes they’re likely to admit. If the program is going to be narrowly tailored to achieve diversity, in short, it must be narrowly tailored to function as (at least) a pseudo-quota.

The boundaries, again, are fuzzier, because you might end up with a percentage admitted anywhere in a range of acceptably diverse representation, but it’s not clear why that really removes the quota-like character of the system. It’s settled law that it would be unconstitutional for, say, 13 percent of the new seats to be set aside for members of group X, so that applicants of other races could only compete for the remaining 87 percent, say. That’s precisely the language the Bakke court used in striking the sysem at issue there. Well, any consideration of race justified by diversity will have some lower bound to the range of representation it’s trying to establish, and the weight will have to be jiggered to produce that range. And if that minimum range is 13 percent, it’s hard to see how it’s not just like the forbidden sort of system. The top 13 percent from group X are, just as in that system, competing only against each other, precisely because the system is diversity-justified only to the extent that it has been deliberately engineered to produce that outcome.

I don’t see any way around this dilemma for any diversity program. So I don’t see how the Court can avoid deciding either that quotas are OK after all (or that they’re OK as long as you do it in a way that’s merely cosmetically different, so that they’re not obviously quotas on face) or that diversity is no good after all as an justification for considering race in admissions.