MR points to this very nice summary, complete with graphs.
I liked the technique of treating individual states as having two dimensions (Democrat share of vote 2004 vs. Democrat share of vote 2008, or polled vs. actual). It made it clear there wasn't much in the way of surprises, but that things shifted on a fairly wholesale level towards the Democrats.
GitHub Invites Contributions To 'Open Source Guides'
48 minutes ago