hich is the best college football team in the country?
Fans debate it at games, in bars and on sports talk radio. The opinions of the news media are revealed in the Associated Press poll. The views of coaches are on display in the USA Today/ESPN poll.
In more scientific, presumably less passionate fashion, individuals and organizations like The New York Times produce computer rankings, each of which has its own way of using raw data to determine the top teams.
As it has since it introduced its ranking in 1979, The Times computes its ranking by using won-lost record, the quality of the opposition (expressed by the opponents' records), the site of the game and the margin of victory.
"Winning counts the most, in an overwhelming way," said Marjorie Connelly, a staff editor in The Times's news survey department.
The Times starts each season with a clean slate. What Miami, Florida State, Nebraska, Oklahoma, Virginia Tech, Kansas State or any other school did last season does not carry over into the next season.
"The computer doesn't know a team has a really great quarterback, or that a team always starts slow and will improve," Connelly said. "At the start, all Division I-A teams are created equal, so reputation does not count. Sometimes a poll takes into account weather, or that a game was pulled from the jaws of death. But the computer doesn't know that."
Sometimes, The Times's ranking is at odds with the polls, which can be explained as the difference between the objectivity of the former versus the subjectivity of the latter. For example, the computer will penalize a team with a weak early schedule, but coaches and journalists may rank the same team higher because they sense that it is stronger than its schedule at that time indicates.
The Times's ranking is not published until there is a database of four or five games; each succeeding game at that point is worth a bit more than the preceding one. Each bowl game is worth one and a half times more than the last regular- season game.
Through last season, The Times's ranking was one of eight included in the elaborate Bowl Championship Series mathematical formula that is used to set up the national championship game between the top two teams. The B.C.S. also uses the news media and the coaches polls in its calculation.
But before this season, the B.C.S. eliminated The Times's and another ranking, the Dunkel Index. The B.C.S. is moving consistently further from margin-of-victory measurements. When The Times chose not to alter its formula at the B.C.S.'s request, it was dropped.
"Some people feel that margin of victory encourages the running up of scores," Connelly said. The Times's computer model scales back runaway victories.
"Margin of victory is still in there, but we've attempted to lessen the impact, which has come mostly from the media, criticism that people ran up the scores to impress the rankings," Roy Kramer, the commissioner of the Southeastern Conference, said. "I think the subjective polls were more influenced by that than the computers were."
Richard Billingsley, whose computer ranking is one of those still in the B.C.S., said he no longer counted margin of victory. For years, it accounted for 40 percent of a team's ranking. "I was 19 and naïve then," he said. Eventually, the statistic's impact was diminished to 20 percent, then to 5. Now it is gone.
In Billingsley's formula, like The Times's, recent performance counts more than earlier games. But, he said, his calculations depart in several ways from The Times's. His starting point each season is where a team finished the previous season.
"If a team finishes No. 1 in 2000, in 2001 they start at No. 1," he said. "It's obvious that all teams are not created equal, so how can you start them out equally?"
While most rankings are weighted heavily to victories, Billingsley said, his is tilted toward losses.
He said another difference is that the strength of the opponent is not based on victories or losses, but by the rank of the opponent. "If you go by the record of opponent, Notre Dame is worth the same as Nevada because both are 1-3. Well, hello, is anybody in there? It's obvious the Irish are better."
