That's actually a re-weighting the the IBD/TIPP that came out today.
According to friend of the Blog Dave_in_Florida, that's what happens in 2 of the 4 models when you re-weight the internals to reflect reasonable expectations. Even the very generous 2008 turnout model predicts a 5 point Romney win. Using the Rassmussen party Id for weighting results in an astounding Romney + 12.
Be sure to look into Dave's archives, he explains (with Math!) exactly the formula he uses to re-weight the polls.
Yes, it's probably an outlier. Yes, the race is probably closer than that. Right now, I just don't care.
Excuse me while I attend this schadenboner.