Not sure I follow your last paragraph. Are you guessing polls are skewed in 2012...if so how much?
I’m saying I would use a range of weighting to reflect likely best/worst cases for each side. I don’t have the numbers on hand at the moment, so I’m using fairly generic descriptors to say that I expect turnout to be somewhere between 2008 and 2010 numbers, but close to 2010.
On the whole, Rasmussen is the only pollster whose turnout model has been along those lines, though Gallup seems to be zoning in on those numbers now, as well — it’s harder to tell there, because they’re still using registered vs. likely.
But these polls showing D+7 or higher are patently ridiculous.