I posted this in the comments, but I liked it enough to decide to make a post of it:

I don't have a blanket objection to affirmative action, but I do think it's important to (a) have a theory as to which people you are giving preferences to, and why, rather than just give a preference to anyone who meets rather arbitrary ancestry rules (e.g., why should a the child of white immigrant physicians from Bolivia be preferred (as a "Hispanic") over the child of dark-skinned poor religious refugees from Iran?); (b)have some transparency, so that students are more or less aware of the scope of the preferences they may be benefiting from, and can choose whether to attend with reasonably full information (this is especially important at law schools, given the extremely high rates at which black matriculants at lower-ranked law schools either fail out or never pass the bar); and (c) do serious analysis every once in a while to ensure that whatever programs are established are meeting their goals (which should be established as part of (a)), and are not just continuing to exist out of bureaucratic inertia (or dogma) despite being counterproductive.

UPDATE: I should note that some or all of these "preferred practices" may be inhibited or prevented by the Supreme Court's affirmative action jurisprudence, which allows preferences only for "diversity" purposes. Of course, no one really takes this seriously, least of all the Court itself; if this had been taken seriously, Grutter would have had to come out the other way, because the district court found as a factual matter that despite Michigan's denials, the law school gave preferences only to select Hispanics (Mexican-Americans and mainland Puerto Ricans). Taking the tenth Mexican-American over the first Cuban or Columbian-American may make sense from a redistributivist perspective, but it hardly contributes to "diversity." Nevertheless, universities must at least pretend to obey the law.