I dunno about this. It seems to me that if you have an African-American candidate whose admitted to past cocaine use, that attacking him for past cocaine use is less an appeal to ugly racial stereotypes than a straightforward attack on his past drug use. An appeal to ugly racial stereotypes would be implying that a black candidate must have used cocaine in the past because, hey, that's what those people do. I don't personally have any problem with the idea that of a president who used cocaine in the past (though, admittedly, the George W. Bush experience hasn't been very pleasant) but insofar as some voters do have a problem with it, they're entitled to have a problem with it irrespective of the candidate's race.
It's certainly true that voters can legitimately object to a candidate's personal history of drug use without reference to race. But does Yglesias really believe that the GOP won't try to capitalize on Obama's past history with drugs?
Let me expand on the point I was trying to make. If Obama wins the Democratic nomination, it seems likely that the conservative attack on him would link his personal drug use with his liberal stances on racially tinged issues like crime and the death penalty. The message will be framed as "Obama: Too liberal" but it will carry racial implications. For instance, as an Yglesias commenter points out, analogies will be made between Obama and Marion Barry, a politician who (sadly) reinforced the fears of many white people about black leadership.
Unfortunately, these tactics reinforce each other. Attacking Obama for drug use (and thereby priming negative stereotypes about blacks and crime) is likely to activate the associated but more acceptable stereotype that blacks tend to be liberal Democrats, as this American Journal of Political Science article suggests:
Can stereotypes of ethnic groups have an indirect impact on voters’ judgments even if voters reject them? We examine the case of Jewish leaders and hypothesize that acceptable political stereotypes (Jews are liberal) are linked in voters’ minds to unacceptable social stereotypes (Jews are shady); consequently, a cue to the candidate’s shadiness works indirectly by increasing the perception that the candidate is liberal, even as the shady cue is rejected. Using three national survey- experiments we randomly varied a candidate’s Jewish identity, ideology, and shadiness. The cue to the rejected social stereotype indeed activates the more legitimate political stereotype. Moreover, voters give more weight to the candidate’s perceived liberalism in their evaluation. Consequently, the candidate’s support suffers. However, when the candidate takes a more extreme ideological position on issues, the effects disappear. The indirect influence of discredited stereotypes and the limits of those stereotypes have implications for our understanding of voting and of the legacies of discrimination.
At this point, these issues are largely hypothetical, but there's serious reason for concern given that John McCain is already making cracks about Obama and drugs. It'll only get worse if he wins the nomination.