Yesterday I did a post questioning the value of the "Iowa ground game" stories in the MSM. So where does one turn for the real beef? The best thing I've read is a long round of diaries at MyDD by someone with the handle of desmoinesdem, an Edwards precinct captain who served Kerry in the same capacity in 2004. The latest (which usefully includes links to six earlier diaries which explain the mechanics of the Caucuses in great detail) addresses the question of the moment: who's winning in Iowa? And after patiently explaining why staring at polls is probably misleading, demoinesdem goes through the much more subjective measurements that a precinct-level operative utilizes to get a sense of which way the wind is blowing.
You should read the whole thing, but a few nuggets stand out. One is that campaigns tend to overlook potential caucus-goers who appear divided between other candidates; they're not on the campaign's radar screen, but could be very important to the actual outcome. Another is the often-forgotten fact that second-place preferences are not just an issue with lower-tier candidates, given the very high "viability" thresholds in smaller precincts. The 2004 winner, John Kerry, was "non-viable" in 222 Iowa precincts. And a third is that late momentum can be crucial, with factors like newspaper and elected official endorsements mattering a lot more in Iowa than in most places.
Speaking of endorsesments, in the comment thread to desmoinesdem's latest post, there's a discussion of the Des Moines Register endorsement, which is coming out this Sunday (much earlier than usual). The Register endorsement is universally thought to have helped John Edwards mount a last-minute surge in Iowa in 2004. But this time, the buzz seems to be that either Clinton or Biden will get the nod. And Biden, as desmoinesdem also notes, is the one low-tier candidate who seems to have some momentum at present--not enough to break into the Big Three, but enough potentially to change the dynamics.