Mark Blumenthal has an interesting column today over at NationalJournal.com about the possibility that the Bradley/Wilder Effect will skew general-election polling and what pollsters can do about it. He adds some interesting thoughts in this follow-up post, including some skeptical comments from Republican pollsters.
For what it's worth, I tend to share their skepticism, albeit for slightly different reasons. My sense is that Obama will get so much media attention--both paid and earned--and that presidential races are so personality-focused in any case, that, by the end of the campaign, the overwhelming majority of voters will think of him as this sui generis character named Barack Obama, not as "that black guy who's running for president." I'd guess the reason we've seen a Bradley Effect in the past is that a lot of voters don't end up knowing much about candidates running for state and local office, so race ends up looming much larger as a proxy, and you get a certain number of people who won't vote for the "black guy" but don't feel comfortable saying so to pollsters.