It seems to me that the plans for an "Iowa entrance poll" -- like an exit poll, but tracking how Iowans intend to caucus, rather than how they actually did caucus -- has the possibility to totally destroy the caucuses. Imagine if the networks spend the night reporting that a plurality of Iowans decided to vote for Barack Obama. They report the win, there's much talk of what it means, everyone gets all excited. Then, Bill Richardson fails to make the 15% threshhold for viability and releases his caucusgoers to Clinton. Meanwhile, John Edwards, who's been amassing support in the disproportionately influential rural counties -- 25 caucusgoers in a small precinct have the same influence as 2,500 in a big one -- sees his strategy achieve terrific results. So Clinton comes in first, Edwards second, and Obama ends up in third -- even though a plurality meant to vote for him. That will, for one thing, blunt the impact of Clinton's win. But won't it also trigger a wholesale reassessment of whether this caucus system makes any sense at all? It would seem very hard for the major networks to commission and tout this poll, receive the results, watch them get totally invalidated by the caucus's procedures, and then pretend that nothing happened. There would be some precedent for that, of course, as exit polls in 2004 showed Kerry doing quite a bit better than he actually did. But the possibility for variance is much greater here, and the sanctity of the caucus much less.
