He is a hill-climbing optimizer - great at incrementally homing in on local optima. But he doesn't seem to be able to get the whole of MS to change direction the way Bill did, in a way that crosses the chasm to a hopefully larger optimum elsewhere. Instead, he seems to take gambles that don't really pay off. A lot of people in tech believe it's because he's not deeply technical and as a result isn't able to connect the dots at a larger strategic level.
He is amazing at communicating to business people however - a boss at a company I used to work for once came back from a partner briefing almost like an evangelical convert. That boss came from sales, and I think that's really Ballmer's forte.
I think there's a perception with Ballmer that he been mainly interested in milking existing cash cows like Office and Windows rather than being especially forward thinking.
While we can all be happy about Apple and Google being more successful of late, give that man some credit for his role in more than 30 years of sustained growth.