So the question with this one datum sample, is was the more accurate Obama team turnout model (the polls with this highly polarized electorate were all about turnout models), due to luck or superior insight?
I think it’s difficult to say. If you look at the internals you can see exactly why the Obama campaign was more spooked with the Ryan pick than it was with the First Debate. It was on a slow decline from mid-July and accelerated by the RNC. It was the DNC that was the big boost that stopped the decline and the 47% comments added to that. If you note the First Debate, Obama took a sharp turn downwards, but some of the polls had this continuing for a while after the debate. In fact it would seem he simply flat lined and then picked up again.
Curiously the battleground state report from early September (pre 47% comments) nailed Obama’s lead in the states that he ended up winning two months later.