I also caught Howard Dean on CNBC while driving — SiriusXM, relax — and he offered a strikingly unconvincing spin for the Democrats outlook for the midterms.

He predicted the Democrats were going to lose one Senate seat, and I presume he meant net. He mentioned New Hampshire, which means he has great confidence in Paul Hodes to win a race that he has never led in any poll.

He mentioned North Carolina, and I interpret that as a sign that Dean trusts Public Policy Polling, which may not be a wise decision.

GOP senator Richard Burr’s lead in Public Policy Polling surveys: 2, 5, 7, 1, 6.

Burr’s lead in surveys conducted by everyone else: 7, 10, 15, 10, 1, 14, 8, 18, 16.

You would think PPP worked for the Democratic candidate, Elaine Marshall, or something. Wait a minute, they did work for her this cycle!

He mentioned Kentucky, but it’s been fascinating to watch Rand Paul have one of the worst general-election debuts in modern history . . . and then, after a few weeks, perform okay. He probably won’t win in a blowout, but right now the evidence isn’t there to suggest the Democrats will win this.

You look at the map, you see a bunch of Senate races that Democrats could win with a bit of luck, a big gaffe by the GOP candidate, great turnout from unions, etc. Sure, they could win New Hampshire. They could win Pennsylvania, even though Toomey’s got the momentum right now. They’re probably going to be okay in Connecticut, although the trend is not their friend there. They could win Ohio, but they would need their guy, Lee Fisher, to be an exponentially better candidate than he has been so far. This week’s primary results in Missouri suggest it may be out of reach for Democrats. In Florida, they’re left hoping that Crist wins and caucuses with them. They have shots, though probably not enough to be favored, in Illinois and Colorado. Even with every Angle misstep, Harry Reid’s chances in Nevada are no better than 50-50, and he remains strongly disliked in much of the state. Finally, three longtime incumbents — Barbara Boxer in California, Russ Feingold in Wisconsin, and Patty Murray in Washington — all should win reelection, but with the polls tied or extremely modest leads, GOP wins wouldn’t be all that surprising.

How Howard Dean gets one net loss from all that is . . . well, it’s enough to make you scream, no?

He also said he couldn’t predict the races for House of Representatives because “it changes every week.” Er, really? What’s significantly different about this week’s political environment compared to the last week of July? How about mid-July? End of June?