It appears I had too much faith in the BCS when I wrote last year's commentary. At the time, I argued that awarding six automatic bids to the major conference champions was merely an accurate reflection of the balance of power in Division I-A football. It is no different from I-AA, which gives automatic bids to the champions of the eight strongest conferences, or other NCAA sports that don't have enough bids to go around.

Furthermore, as a replacement for the old bowl tie-ins, the BCS made perfect sense, and was if anything a step up for the little guys. The odds of Utah playing in a major bowl game under the old system was zero. With the BCS system, they are able to play their way into a major bowl.

However, in preserving the Big East's automatic bid, the BCS looks less like a fair way to handle the bowls and more like a self-interested clique. Every competent analyst knew that, minus Virginia Tech and Miami, the Big East was no longer a major football conference. Averaged over the BCS years (1998-2003), Miami has been the #3 football program in the nation, and Virginia Tech has been #6. The best remaining Big East team, Syracuse, sits at #33. Looking at the whole conference, the rankings are #33 Syracuse, #40 Boston College, #41 West Virginia, #47 Pitt, #90 Temple, #102 Rutgers, and #108 Connecticut. And it gets worse, with Boston College also headed to the ACC, to be replaced by #50 Louisville, #72 Cincinnati, and #93 South Florida. Is this really the mark of a major conference?

For comparison, the best program outside the BCS conferences has been Southern Miss (#29). The best non-BCS conference, the Mountain West, has the following program rankings (again, 1998-2003): #38 Utah, #39 Colorado State, #46 Air Force, #60 BYU, #75 New Mexico, #76 San Diego State, #83 UNLV, and #91 Wyoming. I'm not trying to create an argument regarding whether the Mountain West should be given the Big East's BCS bid; the question is why the Big East champ is guaranteed a BCS bid while the Mountain West champ is not. Clearly the two conferences have comparable strength, which means that having a strong Mountain West champ (Utah) and weak Big East champ (Pitt) was not a huge surprise.

The obvious solution is for the BCS to acknowledge that the Big East is no longer a major football conference and remove their automatic bid. There would be no pressing need to give that bid to a different conference, as the Mountain West is no more likely to regularly churn out national title contenders as is the Big East. This means that either there would be three at-large bids, or the BCS could throw a bone to the non-BCS conferences by giving that bid to the highest-ranked champion from any of those conferences. Had such a system been in place this season, Cal would have been in the Rose Bowl and Pitt would have been left out of the BCS.

What's sad is that this problem was plainly obvious before the season, and nothing was done. As a result, we get treated to Pitt playing in a BCS game they don't belong in. Ironically their opponent will be Utah, which should be favored to beat Pitt by about 17. Perhaps seeing Pitt embarrass the Big East in a prime time New Year's bowl will create the same sort of dissatisfaction that caused the BCS to finally adopt a reasonable formula this year. Perhaps a huge Utah win might bring about the end of the BCS. Or perhaps Pitt will get lucky (I give them a 1-in-7 chance of winning), and preserve a system that is clearly biased against the mid-majors.

I'm happy to see that the new BCS formula worked out (for the most part), given that it was basically my idea. At the encouragement of past Big East commissioner Mike Tranghese, I made the following ideas in a detailed proposal last January:

• Eliminate the loss, schedule strength, and quality win components. These three components combined to create an extra computer ranking, and not a very good one at that. This had two effects. First is that it made the final BCS rankings effectively about 55% computer and 45% poll, rather than the 50-50 breakdown that was advertised. If you ever felt like whenever the polls and computers disagreed, the computers would always win, this is a large part of the reason.
• Reduce the weight of the computer component by half. The problem isn't that there is some inherent fault in the computer rankings that makes them inferior to the polls. The problem is that, from a statistician's point of view, the certainty that a team is superior to a team just behind it in a computer ranking is typically around 55%. In other words, the number of games played just isn't sufficient to get extremely accurate computer rankings (especially given the prohibition on using scores).
• Weed out the worst of the computer rankings. According to the reports I read, this wasn't a conscious decision by the BCS, but rather the New York Times realized that its computer ranking was disgraceful and voluntarily eliminated it from the running. The BCS could certainly go much further in this direction. Massey's ranking comparison lists as large number of computer rankings, many of which would qualify for BCS use (because of using win-loss data only). In addition, the ranking comparison provides a useful comparison with the "consensus", which is a reasonable ranking that uses some degree of margin of victory (and is thus more stable than the win-loss rankings alone). Of the seven win-loss ratings that are generally the most stable, only two (Anderson-Hester and Colley) are currently used by the BCS. This means that the BCS uses several ratings that tend to be unstable (Wolfe, Massey, and Sagarin) or flat-out bad (Billingsley), while failing to take advantage of several very good win-loss rating systems (Wobus, Bihl, Maas, Mease, or mine).
• Interestingly, I also suggested over the phone to Mike Tranghese that the BCS use the poll point totals early in 2003, and was told that the polls would refuse such a suggestion. Presumably the USC fiasco last season made people more open to the need for change. The effect of this is that the BCS formula is now able to discern whether two adjacent teams in a poll are close or separated. When there is a virtual tie, such as what there was between Oklahoma and Auburn late in the season, the computer average is the determining factor. When the teams are clearly separated by points, the computers are less likely to make the difference.

Note: if you use any of the facts, equations, or mathematical principles introduced here, you must give me credit.