*****At any rate, polls based on likely or registered voters should be suspect. They very likely have a sampling that has too many Democrats in it. The poll results are therefore going to inevitably lean toward Obama. This could mean a nasty surprise for pollsters in November, but only if the polls themselves don’t discourage conservatives from voting because it looks hopeless for McCain and Palin.
This year of all years, DON’T BELIEVE THE POLLS. I’m betting they’re skewed because Operation Chaos gave us a false distribution of Democrats and Republicans among registered voters.*******
http://www.politicalbase.com/profile/twayneking/blog/&blogId=4881
That's a very interesting speculation, and one I've not heard before - the test would be to look at 2008 registration in predominately white ZIP codes in districts where there was not a strongly contested Republican primary, and see if you could find evidence of increased "cross-over" registrations after CHAOS was in full swing.
Another interesting aspect of this will be if a consensus forms as to whether CHAOS was a good idea; one effect of a longer contested Democrat primary season is that both Obama and Clinton were forced to to continue to build their organizations well after the point in the election cycle when the likely nominee is often clear. In particular, this may have been a big factor in Obama's very strong ground game even prior to the convention... this sort of GOTV and fundraising effort used to be a built-in Republican advantage in national elections.