If you were planning on spending the next two days trying to predict the results of the congressional race in New York's 23rd district, don't bother. From political polling wonder boy Nate Silver:
Let me declaim that the utterly fascinating special election in NY-23 has become nearly impossible to forecast. Special elections, with their low turnout, are intrinsically pretty difficult to predict. So are multi-candidate races. And certainly, races where there are substantial late-breaking developments — such as the Republican candidate dropping out four days before the election and endorsing her Democratic rival — present especial difficulties for forecasters. Here, you have all three of those circumstances, producing a perfect storm of uncertainty.
Three Big Questions in NY-23 [FiveThirtyEight]