I'm all for keeping the Coaches Trophy, but the Coaches' Poll had better not be used in the playoff team selection method.
There have been a lot of people throwing around ideas for the new college football playoff selection process. Most notable among them is Jim Delany floating an idea that any conference champions that finish in the top six get automatic bids. Top six of what? Possibly the BCS formula, possibly a new formula, or possibly according to a selection committee.
I'm not sure what will be in the final system, but I know some things that should and shouldn't be there.
SHOULDN'T: The Coaches' Poll
The Coaches' Poll is a relic from a time that has long passed. Coaches don't closely study much beyond their own team and their future opponents. It's an open secret that sports information directors do the actual heavy lifting on most (if not all) Coaches' Poll votes with the coaches only editing those ballots as they see fit.
Plus, the Coaches' Poll is a giant conflict of interest. Some coaches have incentives tied to poll finishes and most have some for BCS appearances too. Is a coach's vote definitely going to be valid if, by manipulating it, he can have a better chance at a couple hundred thousand extra dollars at the end of the year? It's an embarrassment that it's even in the current system. It doesn't belong in what comes next.
SHOULDN'T: The Harris Poll
The one and only requirement for being a Harris Poll voter is that the person be nominated by a school. Harris Interactive then draws from the pool of nominees to make the poll. There is no oversight to make sure the nominees are sufficiently knowledgeable to make a good vote or even that they watch all the important games (or any games at all). There have been other issues in the past, and the way voters are chosen reeks of cronyism anyway. It needs to go.
SHOULD: Something like the Legends Poll, if a poll is required
I'm not convinced that the new system should even have an opinion poll component. If the power brokers want one, and polls are almost as big a part of the sport's tradition as anything else, it should be something like the Legends Poll.
The Legends Poll currently consists of 19 former coaches from a variety of geographic backgrounds. They certainly know the game as much as anyone. Crucially, every single ballot on the Legends Poll is made public every week. That kind of transparency is vital, and the guys in it are qualified to make determinations between teams provided they watch enough games.
SHOULDN'T: Any of the current BCS computer polls
I of all people have no problem with computer rankings being used for the purposes of selecting playoff teams. It's just that none of the ones currently used by the BCS should be in there. Five of the six are proprietary, meaning that no one can check the work to make sure the data was correct and the formulas were correctly applied. Don't think that matters? A couple of years ago, someone caught an error in the one computer poll that actually is open. Correcting the mistake, which was merely the absence of one single final score between I-AA teams App State and Western Illinois, caused two pairs of teams to swap spots.
Furthermore, the computer polls aren't allowed by BCS rule to include margin of error in their calculations. The point is to discourage running up the score, but A) it hasn't discouraged running up the score, as the human pollsters certainly consider margin of victory, and B) it makes the polls vastly worse off. That rule is enough to make Bill James, one of the fathers of advanced sports statistics, call for serious statisticians to boycott the BCS a couple years ago.
If you're counting along at home, I've now described why all three elements of the current BCS formula shouldn't be retained in the future. Great system, huh?
SHOULD: A statistically valid computer poll
There are valid ways to come up with rankings based on statistics. Football Outsiders famously has a few, and there are plenty more to choose from. I actually don't care quite so much which systems get used, just as long as they are open and peer reviewed. Everyone should know precisely what the consequences are for various outcomes in the computer polls without having to wait for them to be published each week.
SHOULDN'T: Conference championships
Under no circumstances should the title of "conference champion" be included in the system. The playoff should match up the best/most deserving four teams, period. After all, Alabama was better than LSU last year, but LSU won the conference title because it played a slightly better game than Bama did in November.
The Tide wouldn't have been excluded from the bracket if Delany's proposal was in place last year, but nothing is there to stop a team like Alabama from being excluded in the future by that rule. Besides, the conferences are all different sizes. Some have championship games while some don't. The length of regular season conference schedules isn't even uniform either. Apply different systems to the leagues at the end of the regular season and you might get different champions. That's not a good determinant for playoff teams.
There is no way to make the complaining stop. Even if we wiped away the BCS and went back to the old bowl system, there still would be complaining that the top teams don't play each other. Expanding the field of potential champions from two to four is a good start, and it should stay at four for a good long while before anyone even thinks of reevaluating the size.
The selection process needs to be as simple, logical, and open as possible. Including an opinion poll by people who don't watch games and have a conflict of interest is not how you get there. Including an opinion poll by people who may or may not watch games and aren't selected rigorously is not how you get there. Including computer polls that are black boxes and can't include everything they need to is not how you get there.
If there must be a poll, it must be done by experts, and each and every ballot all year must be inspected. Any computer polls must be open and verifiable. The "conference champion" title must not be used because the conferences are all of different sizes and qualities with different methods of determining who the champion is. I personally lean towards a selection committee with them getting input from high quality computer formulas, but really, almost anything but the current BCS system will suffice.