National Journal's Josh Kraushaar is the latest analyst to take on the "Obama must win Ohio" thing; he's focused on Virginia, instead. I'll keep this short, and just repeat what I said last time this came up: this is the wrong way to look at presidential elections twenty months, or even six months, out. Most presidential elections are won by at least three percentage points in the overall vote, and if that's the case, then state-by-state analysis doesn't matter.
What's more, for the most part, what's good for Barack Obama nationally is going to be good for him in Virginia, or Ohio, or New Mexico, or for that matter Rhode Island and Utah. To the extent that Obama's choices are going to affect voting in 2012, there's really no reason to believe he can carefully calibrate his policies so that they land where it's going to be electorally optimal, no reason to believe that in practical policy terms he's choosing between Virginia and Ohio. At this point, there's just no reason to think about the 2012 presidential race in terms of states and the electoral college.