It's always fun to make predictions, but our ability to get a handle on who will win and who will lose depends on brave souls putting their predictions down and standing by them when the results come in. Only then can we learn which commentators are responsible and which are not, or at the very least, which ones have some idea about what they're talking about. The Washington Post currently has a nice compilation of thinkers and pollsters who have been willing to go on the record.
In my mind, the people who follow the polls are on the right track. There are always freak occurrences that can change everything, but on the whole, freak occurrences are well ... freak. The better bet is to just go with the numbers, unless of course you're wise to the whole liberal conspiracy to rig the numbers.
In any case, a quick look through the predictions has almost everyone saying that Obama will get 303 electoral votes, which is much less than his 2008 total, which was a whopping 365. However, the predicted victory number for Obama is still more than Bush's tight 2004 win against Kerry ... he hopped over the 270 vote requirement with 286.
However, the election will be a test of many theories that have been taken as truth, one being the economy. If Obama wins despite the bad economy, then it will be one piece of evidence that near term data like tracking polls and state-level polls are more important than broad indicators about the country such as the GDP growth rate, etc. As polling gets cheaper and more accurate, it's likely that ordinary polling will become a dominant factor in election prediction models.
Here are some people to look at.
George Will and Michael Barone both predict a Romney victory by more than most people predict an Obama victory.
A lot of flak has been directed at Karl Rove for his shameless cheerleading. However, it may be less shameless than one might think. Of the conservatives on the WP's list, he's predicting a fairly close election. Or is that just part of the cheerleading?
Jim Cramer predicts Obama will win 440 electoral votes. I guess that's a joke.
One final thought, it's important to remember that there's more to being right than getting the winner right. If Obama wins barely, then there's no reason to think Silver's predictions are better than Rove's. What matters is how close you get.