Only three, PA, CO, and NV, IMHO, remain at issue. I continue to believe that Specter will beat Toomey, but until the data so track, I've to return Pennsylvania to the Republican column. Reid is, I think, absent major intervening external events or Ashjian's righting a desultory campaign, toast, and although I regret the loss of the seat, I can't say that I'll mind seeing Durbin or Schumer (even as my anarcho-libertarian views are more in sympathy with those of Reid on the whole, it is my sense) replacing feckless Harry. I've no idea what to make of Colorado, in which the trends are, AFAIK, mixed, although I'm inclined to think, contrary to (at least what was) the popular sentiment, Bennet will be a stronger candidate than Romanoff. My subjective estimates (PA 50-50, NV 20-80, CO 40-60), then, yield a result of slightly better than one seat more for the Democrats than my strict prediction suggests, and I regard as very high the likelihood that the loss will be six seats, even as one cannot conclude with great confidence what the composition thereof will be.

What I might have called the worst-case scenario two months ago is now the likeliest scenario.

Connecticut, to be sure, returns--decisively--to the Democratic fold, but Arkansas and Colorado are moved to the Republican side of the column. I still expect Beau Biden to top Mike Castle, but I imagine that I might have to move that race, too, in the next iteration.

Upon the (devastating) Dorgan announcement.

Giuliani out. :)

I don't imagine that I meant to give Toomey the win in my last prediction, and in any case I return Pennsylvania to the Democratic column here, in which--for now--I keep Delaware, in which I expect Castle's lead to erode upon the formal entry of Beau Biden; unquestionably a toss-up, though, and one that will remain so through election day.