Iowa's caucus results have been "decided" several times already in both parties. On the GOP side, Romney had such a strong lead in organization and polling that early national front-runners Giuliani and McCain both declined even to compete in the kickoff event, the Ames Straw Poll. Their absence, and the fact that Thompson hadn't formally entered the race, allowed Huckabee to get some attention by finishing second to Romney at Ames. In the last several weeks, he rocketed to the top, then to a commanding lead, in the polls - to the point the pundit class then practically conceded the state to him (based on his appeal to evangelical Christians, who have made up roughly 40% of the GOP caucus-goers in recent cycles). Now, of course, Romney seems to be tightening the race again and if the trend continues, it's at least a toss-up.

The Democrats' side had been similarly dominated early by Edwards' strength, having begun organizing in the state by early 2003 at least, reflected in polls. Then came Hillary's move to the top, as the long shadow of her "Aura of Inevitability" lifted her to the poll lead and some began talking about who would be her running mate. Clinton stumbled over her own feet, allowing Obama to take the lead, as pundits were only to willing to assume he would then win the state.

Whew! All of this in a state notoriously difficult to poll because of the intricate nature of the caucus system. First, it's not a one-man, one-vote operation: delegates are apportioned by precincts based on the number of votes cast in previous general elections, not by how many show up on caucus night. Democrats heavily weight their delegates to precincts who voted most heavily Democratic, so smaller precincts can earn the same or even more delegates than larger ones. Figuring all this out takes the parties, who presumably understand their own systems, into the next day or longer, so the idea a pollster could get an accurate picture of what might happen is laughable - even if their sample were perfect, if they couldn't weight it by precinct.

Secondly, it's NOT a primary where you can stop by and vote anytime during the day and be done with it. Caucuses are held beginning at 7 p.m. on a week night - if you are late, tough cookies. You need to commit up to three hours or more, depending on how long it takes to distill the attendees into "viable" groups to award delegates. In between votes, there are periods for persuading others to switch. All this on a winter Iowa night, with many or most locations poorly heated . . . determining who is a "likely" caucus-goer is not at all the same as screening for a primary.

Third, because "non-viable" caucuses at a precinct must eventually join viable ones, people may easily end up supporting their second or third choice - or even someone they don't really like, if enough of the neighbors they do like are in that group.

David Yepsen reports for the Des Moines Register:

All that tends to overlook two other candidates: Democrat John Edwards and Republican Fred Thompson. They're both showing real potential to come up fast here at the end.

One reason is the indecision of a lot of caucus-goers in both parties. This cycle, polls have shown around half the likely caucus-goers in each party could still be persuaded to change their minds. In other words, the so-called front-runners haven't closed the sale.

Both Edwards and Thompson are pouring time and resources into Iowa these days. Edwards built a respectable organization in Iowa in his 2004 campaign. He was the front-runner here for a while, then gradually slipped as the attention focused on celebrity candidates Clinton and Obama.

Read the rest at the link above. The key points are the high percentage who say they could still change, indicating the strength of support the polls show may be illusory, and Iowans' history of confounding the pundits. It would be a mistake to count either Edwards or Thompson out already (both are now fully engaged in the state, the candidates who aren't can be written off). Edwards has the organization he built and nurtured for five years, people who are committed, motivated, and experienced in the caucus mechanics. Thompson has his star power, boosted by the surprise endorsement of Rep. Steve King, a popular conservative. He also must realize he needs a respectable showing in Iowa to help his chances in South Carolina and Florida in the coming weeks, or his campaign might follow its late start with an early finish.