OBAMA-ADJUSTED POLLS. Observing all due caveats about early horserace polls, it's possible to learn something if you remember that these polls are basically distorted by unequal name recognition. When I was a kid, the polls used to always say that Ted Kennedy was going to be the next president, later it was Mario Cuomo, etc.
But since name ID is itself a measurable thing, it's possible, if you're not bound by the strict code of professional ethics governing the polling industry, to extract some useful information by factoring it in. In the current polling, there are four candidates who are almost universally known among voters: Senator Clinton, Senator McCain, Giuliani and Gore. (And John Edwards, probably somewhat less so, but for some reason he isn't in the Newsweek or CNN polls.) And there are two who are not at all well known, Senator Obama and Governor Romney.
(By the way, that thing about the ethical code was a joke. Don't panic, Mr. Penn,.)
In the Newsweek poll, 81% percent say they know "A lot" or "some" about Clinton. Of Obama, only 41% say they know a lot or some, and only 14% say, "a lot," vs 45% who know a lot about Senator Clinton.
So this is a forty percentage point gap in basic ID and awareness. Yet it translates into only a four to seven-point drop-off in support, against the better-known Republicans. The top-line story here was that Clinton beats McCain 50-43 while Obama loses 43-45. But the more interesting thing is that the drop-off from Clinton to the unknown Obama is only seven percentage points, and only four against Giuliani. (That's the Newsweek poll; the CNN poll has the Clinton match-ups tighter, but the Clinton-to-Obama dropoff is about the same 4-6 points.) At 43%, Obama is actually outperforming his name recognition.
Whereas on the Republican side, when you drop from the big names (who I still believe will not get the nomination) to the lesser-known Romney, support disappears completely. Romney drops to 25% against Obama, 32% against Clinton.
That suggests to me some combination of (1) people who do know something about Obama are very favorably inclined toward him, and (2) people are strongly inclined to vote for a Democrat in 2008, which is also reflected in the generic polls. On the first point, it's worth noting that Obama is not significantly better known among Dems than among registered voters generally. It would be interesting to have some polls with matchups like Vilsack-McCain; if it's pretty much the same as Obama-McCain, that would suggest that the Obama support is mostly an artifact of the generic Democrat preference.
These are good polls for Clinton, but I think that given the current level of name recognition, they're at least as good, and maybe better, for Obama. It's awfully scary to think of going into the campaign with a relative unknown, who can be defined by attacks, but if you're going to have a fresh face (and Dems should), better that it be someone with obviously extraordinary political skills and a vision, and who starts off with a strong base of goodwill and admiration.