Nate at 538 confesses that he is "having a hell of a time figuring out what to make of the post-convention polling."
One thing he does discount is the big bump portrayed in Gallup's numbers. Specifically, he references the following paper published in Public Opinion Quarterly:
Likely (and Unlikely) Voters and the Assessment of Campaign Dynamics - Erikson, Panagopoulos and WleizienAbstract
Only in recent years has the "likely voter" technology been extended to polls well in advance of an election. In the case of the 2000 U.S. presidential election, CNN/USA Today/Gallup tracking polls indicated considerable fluctuations in likely voter preferences, greater than among the larger pool of registered voters surveyed. This article explores how Gallup’s likely voter model exaggerates the reported volatility of voter preferences during the campaign. Much of the reported variation in candidate preference reported by Gallup in that election is not due to actual voter shifts in preference but rather to changes in the composition of Gallup’s likely voter pool. The findings highlight dangers of relying on samples of likely voters when polling well before Election Day.
The conclusion is that 'Likely Voter' polling methodology seems to have some problems this far out from the election. For all the bouncing around, Nate's treatment of the numbers is designed to take the sturm und drang out of polling and presents these results:
Definitely down a few points from the days immediately after the Democratic Convention, but not quite the same as a 10 point lead for McCain/Palin.