Enten ponders it:

[H]istorically, the third debate has had little impact. The average change in margin since 1988 just before the third debate and after is only about a point, compared with 2 points for the first debate. This small effect likely has to do with the fact that the third debate takes place very late in the campaign, and there is little the candidates can do now that would be all that surprising.

The only factor that does make me doubt this analysis is that this year has not been one that has followed the rules. We'll have to watch tonight's debate and the days following to see if it does, indeed, make a difference.

Matthew Dickinson is betting on "a bit less volatility in the polling in the remaining 17 days before the Election.":

[T]his is not to say that Monday’s debate “won’t matter.” It should help at least some of the remaining undecideds begin considering their vote choice in light of the fundamentals, and in this regard should continue the process of bringing the polling closer in line with the actual vote come November 6. But I would be very surprised if it led to a bump, or slide, in Obama’s polling anywhere near the 2.7% impact of the first debate.