Snowman on Election Day
To follow up on Karl Rove and the phantom realignment, in general I think everyone needs to be much more cautious in their theorizing about political trends and the deep causal origins of electoral outcomes. It's fairly easy to construct a little narrative about postwar electoral politics that fits the data in a plausible way. So easy, in fact, that you should begin to get suspicious -- several different narratives seem to fit the bill.
What should really trouble you, however, is that if you look at presidential elections from 1948-2004, Democrats have won the popular vote seven times and Republicans have won it eight times. This means that on the list of plausible narratives about post-war electoral politics is that the outcomes are completely random and that DRRDDRRDRRRDDDR is just a sequence like you might get from flipping a coin. Which isn't to commit myself to the view that the outcomes really are random (who knows?), but merely that one ought to be very cautious about embracing some Grand Narrative or another.