In the last column I mentioned that the worst thing about the 2004 modification was the emphasis on the polls. The reasons are all related:
The reason the powers-that-be should be concerned about the emphasis on the polls is that an unintended but inevitable consequence is that margin of victory is now the most important factor for a team. The "shortcut" most voters take when evaluating a team is "did they do as well as expected?". What should be even more disturbing to the NCAA presidents is that the expectations are set in Las Vegas. They cannot be happy with the idea of coaches using the betting line to determine their fourth-quarter game-plans.
This is no criticism of the voters. The job is difficult and it is not one for which they are paid. When they adjust a team up or down based upon whether or not "they beat the spread" is the same as a stock analyst adjusting his "buy" rating based upon whether a company's quarterly statement met "analysts expectations." (Like I said, voters are human. I am not so sure the same is true of "market analysts", who as near as I can tell have never been very good at ranking stocks...)
Most fans thought they understood the polls and few fans trusted a program they probably couldn't understand unless it was a single statement "PRINT 'My team is best'". The computer programmers used by the BCS didn't help matters by keeping their criteria for judging teams secret, but that's for a different essay.
The BCS can't address the problems the polls have because the polls were not designed to meet the BCS' requirements. Their only purpose - dating back to the 1930s - is to entertain fans. Nobody ever appointed the Associated Press the commissioner of college football, or the annointer of champions. They had space to fill and papers to sell, that's all. That will continue to be the case, and there's no business reason for them to change their practices to make the BCS happy. The coaches probably wish they'd never let the AFCA talk them into the problem in the first place.
And the AP was correct to not care about the ramifications of the studies of de Borda's method with regard to elections - they were selecting the #1 college football team, not the president (or even dogcatcher). Things changed when the BCS started using the polls to do more than pick a number one, though, and from the BCS perspective it begins to matter how the number 2, 3, and so on are selected.
But once the BCS began using it, the purpose is no longer to pick one number one and entertain up to 40 or so other teams' fans, it is to distribute millions of dollars to eight teams by qualifying as many as 12 teams. With over 100 millon dollars at stake, it is worth looking at alternatives to the Borda count for counting ballots. There are a number of methods that have been designed to select a partial ordering of alternatives from a given set of ranked ballots over the last 50 years - the US is the only "Western Democracy" that is not familiar with them.
And therein lies the key to making the polls a part of the BCS while avoiding the coaches' conflict of interest and the media's ethical dilemna. The media people can continue to display their ballots to their audience and the coaches can keep theirs as secret as they want if the BCS uses a different vote-counting algortihm than the AP and USA Today/ESPN do.
The key is that the BCS would need to have access to the ballots (note, not to the voters' names, just the ballots). The BCS would use the expertise of the AP and coaches' poll voters to formulate its version of those polls in a manner more suitable to ranking the top 25 than picking the top one, and less subject to the "manipulations" of a single voter. The simplest way to do this would be the existing formula for the "entertainment value" the AP and USA Today/ESPN publish and a modification of instant-runoff-voting (IRV) for the BCS equivalents.
This modified IRV vote-counting method is less sensitive to manipulation by individual voters because an exceptionally high ranking for a team by one voter doesn't "count" until there's a ranking where a majority of the voters agree that the team should be ranked at least that high. It's a way to hold 25 separate elections for a "top 25", one for first place, another for second place, and so on. It's easy to see why this works: all teams who didn't win first place are compared based upon the number of first plus second place votes they received for the second-place contest. All teams that didn't win first or second are compared based upon how many first, second, or third place votes they received to decide the contest for third place, and so on.
The BCS vote-counting for the polls would not be made public until the BCS begins publishing its ratings. Since the impact of the polls as published by the media that conduct them is indeterminate with respect to the BCS standings (except for first place votes, the number of which have always been published), no matter how much a voter tries to manipulate the "AP" or "UE" poll, he can have no effect on the "BCS-AP" or "BCS-UE" poll without getting a majority of the voters to agree with his spurious position, and if he does, it isn't spurious - it's the opinion of the majority of voters.
In the grand scheme of things, there is no difference between the #1 as selected by the Borda count and IRV method when there's a clear distinction between #1 and #2, #2 and #3, etc. as is the case in 2004. However, when there is not a clear consensus for teams contending for #6 and the difference between #6 and #7 is 14 million dollars, the Borda count allows one voter to decide, and this method requires a majority of voters to decide. With this much money at stake, the BCS would be well-advised to choose a better method for counting the votes from the all-too-human side of the equation.
The "Score" column in this table is just used to order teams who appeared on the same number of ballots by the highest rank the team received. It is very similar to the Borda count scoring used by the polls.
Why this could work
This method protects the voters from any allegation of shenanigans: a too-high vote doesn't get counted until the majority of the voters feel the team should be ranked at that or a higher level, and a too-low vote becomes irrelevant as soon as a majority of voters are recognized as ranking the team higher than that vote. Using this instead of Borda requires a majority of voters to collude to corrupt the system, no one voter can affect a team's ranking.
AP ballots after 11/27 games counted by IRV
Team Score 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 1 Southern Cal 58 52 63 65 2 Oklahoma 29.25 7 38 65 3 Auburn 26.5 6 29 65 4 California 6.96875 48 62 65 5 Utah 4.00024414 12 42 61 62 64 64 64 64 64 65 6 Texas 3.08203125 5 26 61 64 64 65 7 Louisville 0.78140259 5 36 48 58 61 63 63 63 64 64 65 8 Georgia 0.40859985 1 12 28 42 50 58 62 63 64 64 65 9 Miami (Fla) 0.25488281 6 13 31 44 52 55 58 63 65 10 Virginia Tech 0.21095276 6 14 18 29 41 49 53 59 62 64 65 11 Boise State 0.22411639 4 17 24 37 40 46 49 52 56 58 61 62 62 63 64 64 64 64 65 12 Michigan 0.03759766 2 6 16 27 48 59 65 13 Iowa 0.12957764 1 2 8 10 16 21 35 47 56 65 14 Louisiana State 0.06851196 1 2 7 9 21 33 41 54 61 63 65 15 Tennessee 0.07585144 1 2 2 4 9 14 19 28 32 48 58 65
Team Score 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 16 Florida State 0.00264764 1 8 20 36 54 58 64 64 64 64 65 17 Wisconsin 0.00321043 2 4 8 13 34 49 57 62 63 63 63 64 65 18 Virginia 0.00054526 6 14 38 49 54 58 60 62 64 64 19 Pittsburgh 0.00019860 2 3 1 2 28 40 46 49 51 55 20 Florida 0.00023127 1 2 4 11 17 28 36 41 46 50 54 21 Arizona State 0.00013620 2 9 17 25 29 31 35 38 45 22 Texas A&M 0.00020313 1 1 1 4 4 8 11 19 27 33 43 47 23 Texas Tech 0.00006294 2 3 4 9 14 20 29 40 46 24 Boston College 0.00004804 1 4 13 20 27 31 36 40 25 West Virginia 0.00005162 1 1 2 3 3 6 14 23 26 34 Ohio State 0.00004464 1 5 12 16 23 27 28 31 Toledo 0.00001287 1 3 6 8 11 16 18 Purdue 0.00000995 1 2 4 7 8 12 15 Colorado 0.00000721 2 3 6 9 1 13 Oklahoma State 0.00000292 1 2 6 9 12 Memphis 0.00000542 2 2 3 6 9 10 Fresno State 0.00002515 1 2 2 2 4 4 4 6 8 Miami (Ohio) 0.00000125 1 2 5 8 Navy 0.00000829 2 2 2 2 4 4 7
Team Score 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 UTEP 0.00000787 1 1 1 1 1 1 2 4 Syracuse 0.00000429 1 1 1 2 2 2 2 New Mexico 0.00000054 1 1 1 2 Northern Illinois 0.00000018 1 2 Clemson 0.00000763 1 1 1 1 1 1 1 1 Alabama 0.00000381 1 1 1 1 1 1 1 Bowling Green 0.00000048 1 1 1 1
Teams / Spot 41 3 3 3 3 3 7 1 8 1 9 9 9 11 12 10 12 13 14 16 14 15 18 18 19 20