I don't remember Democrats being much different in 2004. Is my memory failing me?
I don't think I made my point very clearly last night 8-(. It was to say that modern methods of aggregating polls such as Silver's or TPM's, which were in their infancy in 2004, had by 2008 become pretty reliable; Silver predicted 49 states correctly in 2008. By 2012 one would expect even the Republicans to have caught up with the existing technology.
Democrats in 2004 had a legitimate gripe about voting irregularities, but their candidate was flawed and unexciting, and clearly a member of the same establishment as Bush, both Skull & Bones members for example.
But my point was not about the characters of the candidates as much as the accuracy of the existing models. Anyone watching Silver's model could not have failed to imagine a huge Obama victory; all you had to do was look at the numbers. Instead, the Romney folks and Rove types chose to build models based on their own assumptions, mainly because they didn't like the outcomes of the existing models. As usual, their assumptions about the world turned out to be wrong, but unusually this is a case where the mistake can be measured.
The main point was that once again the Republicans have operated on models of the world that fit their views but not the facts. This time the failure is both obvious and quantifiable.