Mickey Kaus has a good article in Slate about the various intricacies of the Iowa caucuses. There are essentially four measures of the outcome:


1. The entrace poll: A sample of people's preferences when they walk into the caucus sites.


2. The first round preferences: People's initial choices prior to some candidates being eliminated as not viable.


3. The second round preferences: The breakdown after the supporters of non-viable candidates have shifted to viable ones.


4. The delegate count: The pecentage breakdown of delegates awarded in each caucus.


The upshot is that in a close race between multiple candidates, we could get winners at each level. I'd add two points to Kaus's. First, since not every caucus has the same number of delegates, a candidate could have the most caucus-goers and still end up with fewer delegates. For example, in Caucus #1, all 500 attendees go for Dean, giving Dean all 10 of the delegates in that caucus. In Caucus #2, there are also 10 delegates, but only 250 people show up, all of whom go for Kerry. In terms of attendees, Dean wins 2:1 (500 to 250), but in delegates it's an even split (10-10).


Second, that's only tonight. The delegates selected will go on to county caucus in upcoming weeks where the process begins again when they select delegates to the state convention, where the process again works its way through. So tonight's outcome might be very different from the results in coming weeks and months.


Third, I'm not as critical of the process as is Kaus. Economists and rational choice theorists tell us that any method of aggregating preferences can lead to different outcomes. Look at Florida in 2000. Gore won the popular vote, but Bush won the election in the Electoral College. But did he? Depends on what method of recount you use in Florida? Or what judicial body you rely on to determine what type of recount or whether there will be a recount at all. Plus, just about everyone acknowledges that more people walked in to the booth intending to vote for Gore. Finally, what about the people who didn't vote, for whatever the reason? Shouldn't their preferences count for something? In short, no process is perfect, especially in tight races.


Also, caucuses are not general elections. The only substantive meaning to the caucus is to begin the process by which Iowa Democrats will determine the allocation of their national convention delegates. But the media wants to make more of this, so they demand to make it into a primary. But it's not.


Finally, I don't mind giving parties a bit more independence and autonomy. Parties should have some discretion over how the conduct their own affairs. Yes, they serve a public function but unless it's something as bad as a white primary, I say cut them some slack. Besides, if you are looking for undemocratic and non-transparent aspects of American politics, the Senate and the Supreme Court offer much bigger targets.