In presidential politics we always watch the swing votes, the "unsure" voters, those wacky undecided folks who, despite a billion ads and debates and stories, can't make up their minds.1
Well, it's too early to have a favorite, at least not for any rational person, but let's look at those "unsure" party nomination numbers just for fun.
On the Democratic side, the "unsure" vote was at 13% in March and has steadily dropped to 7% in September (CBS/NYT poll). On the Republican side, "unsure" was at 10% in June, jumped up to 18% in August, but by September settled back to 10%. There is a lot more churn on the GOP side, what with Fred Thompson (from my hometown of L'burg, Tenn.!!!) entering the race, the troubles with Bush's own numbers, and the war.
1 As a methodological aside, we often use the inability to generate an attitude about a political actor as a negative in computing political knowledge. You'll especially find this in anlysis of ANES data. No opinion counts a "0" and some opinion as a "1" in creating an index of knowledge (along with other, more obvious, items like who is chief justice of the supreme court or which party controls congress).