I've already called out Howard Wolfson for his unsupported claim that secrecy over Edwards' affair cost Hillary the nomination. According to Wolfson, Clinton was the second choice of Edwards supporters, so had the news come out earlier, the combined might of the Edwards-Clinton coalition would've crushed Obama like a bug. Or something like that.
I've been looking at a lot of primary polling, and the evidence simply isn't there to support Wolfson's claims. In fact, the numbers suggest quite the opposite. Take Colorado's primary polling, for example. There wasn't much -- just one pre-primary poll -- but it's from the well-regarded Mason-Dixon folks:
Mason-Dixon for the Denver Post. 1/21-23. Likely voters. MoE 4%
So check it out, Clinton got exactly what the poll predicted she'd get. Obama? He got all of his supporters, all of Edwards' supporters, and most of the undecided. And this was a caucus, like Iowa, so the same general dynamics apply (only the most committed supporters participate, few low-information voters, etc).
What about a southern state, where Edwards' white support was supposed to go to Clinton? For example, Georgia:
Rasmussen. 1/22. Likely voters. MOE 4% (last non-partisan poll taken by anyone with Edwards in the mix)
Clinton kept her level of support steady, Obama vacuumed up Edwards supporters and the undecided. And while this may look like cherry-picking polls, it's actually quite a pronounced phenomenon in states where enough polling was conducted with Edwards in the mix (you'd be surprised how few states were heavily polled early on, with the assumption that all would be over after Iowa and NH).
One more, Missouri:
Rasmussen. 1/31. Likely voters. MOE 4% (last poll taken by anyone with Edwards in the mix)
Again, do the math. It's pretty stark, huh? I think the body of evidence continues to build that Wolfson, quite frankly, was full of shit.