The Random Politic

After this past Saturday night’s upset of USC (Sorry OJ), rumors were flowing regarding who would play the Ohio State Buckeyes in the college football BCS National Championship Game on January 8, 2007. (For those of you out of touch, college football in the United States does not use a playoff format; instead, it uses a dumbfounded formula to decide which two colleges are the top two in the country for championship game purposes.)  This formula is known as the Bowl Championship Series (or BCS).  At the end of the college football season, it is used to determine the teams that will play in the BCS bowl games (the Rose Bowl, the Sugar Bowl, the Fiesta Bowl and the Orange Bowl), but most importantly, which teams will play for the championship.  Clearly, this system is flawed. (For example, if there are three undefeated teams, this system will only allow for two of the three teams to play for the championship — see Auburn, 1994 [Actually, it was 2004 -- good call Bill].)  At the end of this year’s college football season, there were two undefeated teams, Ohio State and Boise State.  (Boise State did not finish in the top two of the BCS, thus, cannot play in the championship game.)  The number two slot came down to two legitimate one-loss contenders — the Florida Gators and the Michigan Wolverines.

To the chagrin of both Ann Coulter and Dick Gephardt, the Gators finished second (and will play the Buckeyes) and the Wolverines finished third (earning a trip to th Rose Bowl versus USC).  So, how did this happen since the Wolverines are obviously (Note: my opinion) the better football team?  It happened for two reasons:  the BCS does not incorporate a playoff and two-thirds of the BCS formula can be manipulated by the “human” element.

The current BCS system — as it stands today, December 4, 2006 (check back for updates!) incorporates three “sub-systems” in determining the top 25 teams in the nation.  The top two play in the championship game, at a neutral site (this year, at the site of the Fiesta Bowl in Glendale, Arizona). The current BCS formula’s three sub-systems are the (i) Coaches’ Poll, (ii) Harris Poll and (iii) computer rankings.

The Coaches Poll. The USA Today Coaches’ poll is probably the most common college football Division I-A top 25 ranking poll.  (Although the AP poll is also pretty well-known, it is no longer used as part of the BCS.)  This poll ranks teams based on the votes of college football coaches — currently, 63 head coaches at Division I-A colleges.  (Yes, the college football coaches vote for the top college football teams in the country.)  Each week, including and most importantly at the end of the college football season, the votes are cast, translated into points.  Then, for BCS purposes, each team is assigned a percentage of the possible points it could have received. This year, Ohio State received 1550 of a maximum 1550 points (1.000%), Florida received 1470 points (.948%) and Michigan received 1444 points (.932%).

The Harris Poll. The Harris Interactive College Football Poll is the “other” human poll used by the BCS for ranking the top 25 college football teams in the nation. Instead of taking its name by the title of the voters, this poll takes its name from its administrator, Harris Interactive, a market research company that specializes in Internet research. This poll recently replaced the AP poll as part of the BCS. The Harris Poll is comprised of various “humans,” and includes, among others, former players (Terry Bradshaw and Boomer Esiason), college administrators, and current and former media.  (Is this the Oscars?)  Like the Coaches’ poll, each voter picks the top 25 teams each week.  These votes are also converted into points and, again, for BCS purposes, each team is assigned a percentage of the possible points it could have received. This year, Ohio State received 2,824 of a maximum 2,824 points (1.000%), Florida received 2,670 points (.945%) and Michigan received 2,632 points (.932%).

The Computer Average. The non-human component of the BCS includes six computer based ranking systems. These computer systems rank the top 25 teams – assigning the number one team 25 points, number two team 24 points, and so on. To compute the final computer average to be used in the BCS, the highest and lowest ranking for each team is dropped. Therefore, if a team receives five first place votes and one third place vote, it will still end with 100 points (4 x 25). As with the prior two polls, a percentage is produced to incorporate into the BCS final standings. Probably the most important aspect of the computer average is that the individual computer systems take into account the strength of each teams’ schedule. This year, Ohio State received 100 points (1.000%), Florida received 94 points (.940%) and Michigan, tied with Florida in this poll, also received 94 points (.940%).  Apparently, the computers realized that Michigan also played a difficult schedule and only lost to the best team in the country — the Buckeyes.

Okay, I know that this is a lot of information, but at least now the BCS should make some sense.  (Hopefully to you as I am lost again.)  Once each team has been assigned a percentage from each of the three “sub-systems,” the three are averaged and the teams are ranked based on that average.  At the end of the regular season (this past weekend), the top two teams are “chosen” to play for the college football championship.

This year, Ohio State (still undefeated) will play Florida (with one loss to Auburn), and the Michigan (with its only loss to number one, undefeated, Ohio State) will have no chance to win the championship, even though its only loss was to the top team in the country — whereas, Florida lost to a two loss Auburn.  Since the final BCS results were released, many have speculated that Michigan was ousted because their only loss was to Ohio State last week – albeit by only three points in Columbus, Ohio — and the human voters did not want a rematch between the two teams (which also play in the same conference, the Big Ten).  Prior to this past weekend, Michigan was ranked third and Florida was fourth (trailing Michigan by a somewhat large margin in all three polls). On this past Saturday, after number two USC lost to UCLA (knocking itself out of the number two spot in the BCS), Florida defeated Arkansas.  While this was going on, the Michigan Wolverines were watching the games on TV — at no fault of their own — since their season ended two weeks prior with the three-point loss to Ohio State. Sunday evening, when the polls were released, Florida jumped Michigan.

In my view, because of the foregoing, the system does not work for college football.  Either we need a playoff or someone should just call me and ask who I think are the best two teams.   (Who wouldn’t want to see Michigan and Rutgers in the championship?)  Now that I have wasted your time with the details of a system that liekly will not last, let me at least provide some weak insight.  After analyzing the BCS framework and reviewing some of the raw data for various years — maybe this system could be used for something.

To stick with the current theme of prior articles – why not use the BCS to narrow the choices in a primary election?  Think about it.  We could use a combination of polls to determine which two of the numerous candidates we want to vote for in a primary election. This could eliminate the dirt that is thrown around within the same political party (Bush v. McCain) — often hurting the party — during the primaries.  With only two primary candidates chosen quickly, costs would be cut and the overall election process could be streamlined to allow for increased information dissemination to the public prior to the primary vote and election day. So you may ask, how would the polls work?

A Coaches-like Poll.  Instead of coaches, the actual candidates’ campaign managers would vote and rank each candidate in the primary. Each candidate would receive points based on their ranking from the other candidates, and a percentage would be assigned to each.  What could be more democratic than having the competing candidates in a particular political party vote for each other?  (Actually, I guess having the people vote would be more democratic.)

A Harris-like Poll. I believe this is what Harris does anyway – so we would not even need to change the name.  An array of people could be assigned to participate in the vote, similar to the Nelson ratings.  Obviously, this poll could be expanded to represent the population – thou shall not disenfranchise.   

The Computers. Affectionately called the modem-poll,” six computer ranking systems could be set up to incorporate various factors in ranking the candidates.  (I would definitely want a chance to set one of these up.)  The computer rankings could various factors, including, experience, voting record, campaign finances, prior misdemeanor and felony convictions, tenure in civil service, filmology, etc.

Okay, maybe this is going too far.  I tried to mask my unhappiness in the BCS with politics.  Maybe the BCS is just useless.  Regardless, I would rather see it make the primary ballots smaller and get rid of those seven-plus-person debates, than diminish the quality of college football in early January…