On October 23rd, college football teams and fans alike will be eager to see the first 2000 edition of the Bowl Championship Series rankings--the system that has determined the participants in college football's national championship game for the past two years with minimal controversy. The triumph of the BCS in producing a Tennessee-Florida St. national championship in 1998 and a Florida St.-Virginia Tech national championship in 1999 have made the BCS a superficial success.

But the sparkling record of the BCS thus far has merely covered up the system's glaring holes. It's only a matter of time before the system will deny a deserving team of the chance to play for the national title.

College football statisticians claim that the BCS can't fail. They say that they have tested the BCS with data from the past 20 college football seasons and that the BCS produces the desired national championship every time, so therefore it must work. They fail to acknowledge, however, the common sense fact that a system's success with a few specific examples from the past does not guarantee its success with realistic outcomes in the future.

The BCS index is calculated by adding together four factors: composite poll ranking, composite computer ranking, strength of schedule ranking divided by 25, and total number of losses. The two teams with the lowest index play for college football's national championship in either the Rose Bowl, Sugar Bowl, Fiesta Bowl, or Orange Bowl, depending on the year.

The BCS has fallen under a great deal of controversy since its inception, since people have a natural aversion to using computers to determine their fate. Most arguments against the BCS have focused on the computer rankings which favor teams that run up the score. But the biggest problem with the BCS is its overemphasis on strength of schedule.

Without a doubt, strength of schedule is an important consideration in determining the relative abilities of two college teams. That is why both voters and computer rankings take strength of schedule into considerable account in their respective ranking criteria.

Although every voter and computer will place a different emphasis on strength of schedule, the combined opinions of each should, in theory, produce rankings which give the optimal weight to strength of schedule. The problem, however, is that by using strength of schedule as an additional factor in the rankings, it is given far more weight than the average desired weight of computers and voters, and possibly far more weight than any single computer or living human being would ever want.

The BCS designers didn't even intend strength of schedule to have so much weight. Early BCS literature considered strength of schedule to be only of minor consequence, since it could only differentiate teams by at most four points (since there are about 100 division 1-A teams, and the BCS adds in the ranking divided by 25). But in practice, when two teams are ranked one-two or two-three in the polls and computer rankings, the strength of schedule points can far outweigh the effects of the polls.