Fixing the BCS (Formula)

© Copyright 2004, Paul Kislanko

The BCS system cannot be "fixed", for instance by replacing it with a Division 1A playoff, but the formula used by the BCS can be improved. In fact, it has been changed every year, and almost every time for the worse. And the 2004-05 offseason is the right time to change it, because the 2004 version was the worst of all.

Why the system can't be replaced is easy to explain. The reason that Divisian 1-A football is the only NCAA sport whose post-season is not entirely governed by the NCAA is that Division 1-A football sued the NCAA for the right to govern itself, and won. When the Division 1-A presidents want a playoff, we'll have one. Until then there's nothing the NCAA can do about it. The same thing would've happened in Division 1 basketball except that the NCAA learned from its mistakes, and shares the revenue from the final-64 in such a way that no subset of hoops teams can make more money by splitting off from Division 1.

But the BCS formula must be changed (again) because the 2004 version is counterproductive. It can be changed this off-season because there's no reasonable change that would affect the 2004 standings. Auburn would still have been third, and Texas would still displace California for the final BCS bowl slot. The problems with the 2004 formula are not with the final results, they are unintended (but entirely predictable) results from the knee-jerk (and not well-thout out) changes made after the 2003 season.

The Mess the BCS has Made

The committee must do something about how it uses the polls. The Associated Press was already quite correctly concerned about being the story instead of the story they were covering, but their carefully-designed process has been compromised because "regional" reporters have been harassed into changing their votes because of personal email attacks by folks outside of their coverage area trying to get them to change their votes. The most qualified of them to rank college football teams are already on record that they will refuse to participate in the future. The coaches voted to keep their ballots secret so they wouldn't be bothered by "why did you vote that way?" questions but that only got them bothered by "who did you vote for?" questions.

The BCS is killing the guy who feeds the goose that lays the golden egg. They've antagonized the fans, the voters in the polls, and the organizations that sponsor the polls, all because they made the polls the most important part of the formula.

On top of that, by making the human polls two thirds of the formula, they endorsed "running up the score" and gambling, since margin of victory compared to the Las Vegas spread is all most of the human voters have to go on when they're filling out their ballots. That may have been an "unintended consequence", but anybody with a working brain foresaw that when they made the announcement of the 2004 formula.

They really messed up the computer part. They said they were calculating the computer average "the same way" as the human polls are calculated, but neither the Associated Press nor USA Today/ESPN throw out the highest and lowest ranked team from every "voter's" ballot. Give me a break, if any computer's ranking should be thrown out for any team, then throw out that computer. Manifestly it is not the same counting method, as anyone with a working brain can tell. Of course, we're not talking about "working brains", we're talking about the BCS Committee, which only has the equivalent of an autonomic nervous system with a "stimulus-response" capability. "Thinking" is asking too much.

The Simple Fix

All of the problems that the BCS has created for itself can be solved very easily. All it takes is a little thought.
  1. Use the ballots from the human polls but don't count them the same way as the AP and UE do for entertainment purposes. See Part One - Polls for the reasons and one possible alternative.
  2. Expand the computer list to include enough computer rankings that the same technique used to summarize human voters' rankings applies to the computer rankings. See Part Two - Computers for the reasons, and use the same method to summarize the computers as is used to summarize the human ballots for the two human polls.
  3. Define the formula so that the polls and computer rankings have equal weight. Part and parcel of this is making visible the characteristics of the computers included. Just as the AP ensures its voters are representative of the entire community geographically, and the AFCA ensures its voters in the UE poll are representative of all conferences, the computers should be representative of those who use MOV or don't, are predictive in nature or reflect season-to-date, etc.

The computers (once there are enough of them) should get equal weight if we desire the whole season to matter. It is too much to expect human voters to do an in-depth analysis of enough teams to rank 25 of them taking into account more than the team's most recent game, especially since they have only a few hours to do the job. If the average of the human polls is half and the computer rankings is half, the effect is that the computers become a "tie-breaker".

That's all it would take. There could be no argument against the use of computer rankings because their characteristics would be known (even if the proprietary algorithms aren't). There could be no argument that subjective human judgement isn't included because that would be half the formula. There would be no conflict-of-interest issues for the coaches because their self-interest votes are only counted when a majority of the voting coaches agree, and similarly there's no journalistic integrity problems for the AP voters because they don't know how their sincere vote will affect the final ranking, or even their vote's contribution to the result.

Why Now is the time to fix it

The 2004 season may have exposed the flaws in the 2004 formula, but none of these recommendations change the ranking that resulted from it. Auburn would still be third, and Texas would still have knocked California out of the BCS. So for the first time in BCS history there is an opportunity to improve the formula to address its systemic flaws without appearing to be an ad-hoc adjustment to account for "last year's 'problem'".

Every other time the formula has been changed it has been because there was a perception that the "wrong" teams were selected first and second. This year the only real complaints are about the lack of a playoff and the correctness of some of the "automatic" bids, not the formula, and the loudest complaints about the formula are coming from those who participate in the voting. This has made the problem bigger than the BCS and the game.

Actually, I think the bowls and conferences are more likely to withdraw from the BCS than to fix its problems, unless they act to fix them now.