Long answer, it’s long overdue, and this is at least a reasonable way to do it. I still think we’re better off just eliminating it completely, but that’s not going to happen.
Bare minimum, we need to see how each coach (or their designated peon) has voted.

I like this idea.

But it can’t be substantially worse than the present setup. (There’s still some conflict-of-interest issues here, although they’re somewhat mitigated; situations like the Big XII South Clusterfuck of 2008 are still problematic.)

While the faults of the coaches poll are many, the basic conflict of interest has the be the most glaring problem of the coaches’ poll, particularly in its current format. While I’m sure the vast majority of coaches in the poll take their responsibility seriously – i.e. hand it off to a sports information director to fill out – there are more than a few who subscribe to the Mack Brown school of ranking: #1 is my team, #s 2-13 are everyone I will play this season, and that will not change. At the end of the day, there is too inherent a conflict-of-interest in any coaches’ poll, and it should be done away with completely.

even under Adam’s novel approach, there will be games played within the poll unless every coach’s ranks are made public.

Have you gotten groupies yet as reward for your fantastic writing ability?

(1) Not every coach votes, and not every conference is counted equally: Currently, 59 of the 120 FBS coaches vote. There are six each from the ACC, Big Ten, Big 12, SEC, WAC, MAC, and CUSA; five from the Pac-10; four each from the Big East and Sun Belt; and two each from the Mountain West and Independents. On a percentage of membership basis, they’re representative (except for the MWC). On a pure count basis, some conferences are going to get screwed.

(2) A coach is going to have nominally more information on teams his team has yet to play than he will on teams he won’t play at all: In much the same way we may take a cursory glance at future opponents, a head coach has little time between gameplanning, recruiting, and public/media relations to scout a team he won’t see for 8 weeks. Asking him to rank those teams might take some of the “I’ve never seen them, fuck it, they’re #18” out of it, but not all of it.

(3) Similarly, a coach is going to place added emphasis on teams he’s already played: You want your standing as a program to go up? Vote that one team that beat you already #1. Imagine if Kirk Ferentz (who doesn’t vote; we’ll get to the ramifications of that later) voted Arizona #1, ahead of the rest of the Big Ten’s top teams. It wouldn’t be openly ridiculous, seeing as how the Cats were the only team to beat Iowa so far. And it would skew perceptions of his own team, because their only loss was to a team whose ranking is inflated by the fact that they won that game.

(4) If you’re not a voter, good luck getting a game: This is the big one: Telling voters they can only vote for teams they played means that wins against non-voting teams are worth considerably less. Sure, a win by Iowa might compel, say, Sweatervest to move the Hawks up slightly, but that would pale in comparison to a team that played his Buckeyes tough. And for non-con, forget about it; voting coaches would get all the games, if only because it would mean more chances for the opposition to get votes. Conversely, if it were an average as the article suggests, going independent or moving to the MWC, scheduling light on voting teams (or scheduling the voting cupcakes), and obliterating the voters could catapult a team to the top.

It might be a better system, but it’s not a fix. Sorry.

rigmarole.