Josh put a post up in the wee hours of the morning examining the intricacies of delegate selection in caucus states. The main thrust of the piece was that since delegates have yet to be fully allocated in the caucus states, there's still significant room for on-the-ground maneuvering, which might ultimately lead to swings as large as "twenty to thirty" new delegates. I'll admit that there's still some uncertainty regarding a portion of these delegates, but much of the speculation in the post seems rather more breathless than the facts appear to justify.

Now one of the things I love about Josh is that he's always careful to hedge when he wanders beyond the bounds of his own expertise, and he flags those elements of his posts that are speculative or that rely on information obtained from others. It's probably a vestige of his academic training, but whatever the cause, it elevates him above a lot of other bloggers who are considerably less careful concerning facts. So it's not Josh with whom I'm taking issue here, but rather, the sources he cites in his post.

There's also been a fair amount of speculation along similar lines in recent days. The Associated Press put out a story back in February focusing on the caucus issue, particularly the process in Nevada; diarists at DailyKos have been waxing incandescent about the goings-on in Colorado; Iowa is still up in the air; and right here at TPM Cafe we've seen some conspiratorial thinking about Texas. So it would seem to be time for yet another long-winded, overly detailed post detailing the surpassingly strange Democratic nomination process.

Let's start by noting that not all caucus states are created equal. They fall, broadly speaking, into three general categories (plus Texas, which as usual, does things in its own special way). In the first category, nothing's definitively settled until the state convention (e.g., AK, CO, IA, ME, and NV). Most of the room for maneuver comes in this small handful of states, which will ultimately hand out 162 pledged delegates. In their systems, hundreds or thousands of delegates elected at the precinct level, many of whom have little prior experience and are all-but-unknown to the campaigns, have to show up at the caucuses at the next level and vote the way they pledged in order for the initial results to be translated into strength at the state conventions. We're already seeing some signs that in Colorado the Clinton campaign is enjoying somewhat more success at this task than Obama's folks. 
 
But that's just one category. The second uses the actual vote tallies at the precinct level to apportion its district-level delegates, some two-thirds of the pledged delegation. The caucuses also elect delegates to state conventions, sometimes via a multi-tiered process, who then choose the statewide delegates (or, in the case of Washington State, it's the district-level delegates to the DNC who make the selection). So it's possible that poor attendance or faithless delegates could affect the apportionment of the statewide delegates in these states, albeit unlikely. Overall, 95 of the pledged delegates in these states are already locked down by binding rules, and 51 are technically still in play (KS, NE, WA, WY). 

The AP story glosses over the third category - states in which the allocation of every pledged delegate is determined by the initial vote, and in which all that's left for the state conventions (where they are held) is to determine the identity of the delegates, not their allegiances (e.g, AS, HI, ID, MN, ND, VI). There are 129 pledged delegates from these states, all of whom will be allocated in accordance with the vote tallies.

Texas, of course, is sui generis - its rules mandate that the delegates selected at the county/senate district conventions reflect the presidential preferences of voters at the precinct level, but then allow those delegates to vote as they please at the state convention. And of course, there's an outside chance of some shenanigans in the Lone Star State because the vote tallies at the precinct level, though binding, weren't centrally recorded. But on the whole, the incomplete tallies we have are likely to be highly predictive of the ultimate allocation of the 67 pledged national delegates, both because we have no reason to presume them to be an unrepresentative sample, and because only the (much smaller number of) delegates to the state convention are allowed the freedom to vote as they please. That makes it easier for the campaign to vet the folks they'll back for those slots, and to keep tabs on them moving forward.

So when Hillary blithely uses the term caucus delegates, she's lumping together some radically different processes. The 72 pledged delegates from Minnesota, for example, are no more in doubt than the 72 pledged delegates from Missouri - all were fully allocated on February 5. (That's reason enough to conclude that's she's challenging their legitimacy, and not making a procedural distinction.) None of the 45 from Iowa, however, have yet been allocated. To be sure, it's always wise to take any of these tallies with a grain of salt before the totals are officially certified. But I can't see any valid reason to distinguish between Minnesota and Missouri at the moment.
 
If you add up all the caucus-state delegates, you find that 224 have been locked down, with 280 technically still in play. But it's in the first of our three categories, among the five states with 162 delegates, that we're most likely to see swings. Let's dismiss the notion that "twenty to thirty" delegates might be in play - the only way to get a tally anywhere close to that number is to count the 14 Iowa delegates that Edwards' share of the vote suggested he would claim. And it's not at all clear that his backers intend to relinquish those slots. The Colorado experience is instructive in this regard - one county showed a five-point swing toward Hillary, another a four-point gain, and a third ran true to the precinct results. A two- or three-point swing statewide for Hillary would enable her to take a delegate from Obama; a five- or six-point swing would allow her to take two. And that in the largest of the undetermined states. We're just not talking about huge numbers of delegates, and it's unclear that the dynamic will favor Hillary across the board. Of course, there's another dimension to these defections. If Hillary or Obama were to garner more than 50% of the state convention delegates in any caucus state, including those in which the pledged delegates are already locked down, they'd be able to win that state's UADs. But the margins in the caucus states make that even more unlikely than a significant swing in pledged delegates.

The bottom line is that the possibility of faithless delegates is the sort of process story that makes political reporters and insiders salivate, but is extremely unlikely to happen. (I'll qualify that by saying that if the race is effectively over by the time state conventions meet, as it was in 1984, we could see some substantial number of defections.) Both sides, to be sure, have to turn out their supporters and do their homework in selecting higher-tier delegates. But there's no indication that either campaign will fail egregiously in that task, and despite all the jockeying, we'll probably see less overall delegate movement here than we saw in the primary state of California.

If you've enjoyed this, please share it with other readers by clicking the 'recommend this' link. You can find more analysis on my blog. As always, I welcome comments and corrections.