John Sides isn't quite as frustrated yet about the idea that the economic "fundamentals" of the 2012 cycle make a Mitt Romney victory so inevitable that anything else needs to some fantastic explanation as he is at misreporting of independent voters, but I think he's getting close to breaking out the all-caps.
(Short version again: the prediction models are basically predicting a close race; as I read them, it's with Barack Obama as a slight favorite, but it really depends on how you look at it -- some models predict a Republican win, some a Democratic victory).
I think the level of frustration that John (and some of the rest of us) feel at this has a pretty simple cause: almost everyone who is open to persuasion on this point gets it, and so we don't get any more easy victories. That is, those who are open to reading what political scientists have to say -- and it's an impressive number of reporters, pundits, and practitioners -- are reading the Monkey Cage, reading the other political science blogs, reading evidence-based sites such as Nate Silver's, and taking what they read seriously. Indeed, I've been consistently struck since I started doing this by how many people really do want to learn what political scientists have to say.
The problem is that what's left are those who for whatever reason aren't very interested. There are lots of reason for it...their own biases, in some cases; in others, perhaps incentives for believing, for example, that election results are mainly driven by gaffes and other day-to-day campaigning events (not that campaigns are irrelevant! No one is saying that!).
All of which makes for a lot of frustration. But I do think that the flip side of that frustration is that a lot of progress is being made.