Given his unwillingness to commit, it's actually pretty funny that everybody keeps attacking him for being too sure of himself.
Other than the chance of systematic error, 538 doesn't do much that any aggregate doesn't do, although it's nice to have a clear look at the correlation between moves in various swing states. I think he's mainly popular because he produces a lot of pretty reasonable writing explaining why what's obviously likely to happen is very likely to happen. But that's pretty good given his competition.
Thanks for the answer. As for the people attacking him, it's (probably) not him it's the people citing him that's causing the reaction.
We will see if he's right soon enough. I for one am extremely skeptical of his ability to account for something like poll bias, accuracy, how undecideds will break or one candidate outperforming the polls. Again, I am skeptical of all polls in general so I'm already biased against any of these prognosticators, that they then "massage the data" makes it imo worse. Be it Nate Silver or Michael Barone... which is the only guy I can think of right now who leans right and predicts these things.
Having said all that, for fun...
FEARLESS FORECAST:
275 - 263 ROMNEY
I give him Florida, NC, VA, Colorado and OHIO. Listed in rough order of confidence.
Come at me.
http://www.270towin.com/2012_election_predictions.php?mapid=blUh