Thursday, October 05, 2006

Computers Scare Me! Pass the Gravy!

Tony Barnhart, writing under the humble moniker "Mr. College Football," cannot understand why the SEC is ranked behind the Pac Ten and Big Ten in the Sagarin rankings. Now, there are decent arguments to be made on this front, but Tony doesn't make them:

This is why I don’t like computer polls and I don’t think they should be a part of the BCS standings-except to break a tie in the human polls. The SEC has four teams in the Top 10 of the AP, coaches, and Harris polls. But Jeff Sagarin’s computer rankings, which are published in USA Today, have the SEC ranked third in conference strength behind the Pac-10 and the Big Ten. There may be perfectly sound science to back up those rankings but it just doesn’t pass the common sense test. But what do I know? I barely passed freshman Algebra.

So let me get this straight, Tony: you don't like computer polls because they reach results that are different from human polls. Barnhart cites to the fact that there are four SEC teams in the top ten of the human polls as if this is good evidence of the conference's strength, but that fact is actually evidence against the human polls. There's no rational way to argue that Georgia has shown itself to be a top ten team so far this year. They have played five games, none against a ranked opponent, and in their last two games, the Dawgs have won nail-biters against arguably the worst teams in the Big XII and SEC. Georgia can't move the ball on offense and the defense looks good, but hasn't yet been tested by a competent offense. A computer can look at that evidence and decide that Georgia isn't a top ten team. (Sagarin, incidentally, has Georgia at #18, which must be right since I have them at #17. Res ipse loquitur.) Your average human voter ranked Georgia in the top 15 before the year and has moved them up as teams in front of them have lost, which is no way to evaluate teams, probably without having watched Georgia's offensive gotterdamerungs against Colorado and Ole Miss. So thanks, Tony, you've illustrated that the objectivity of a computer ranking can be better than a human pollster.

And one other point: Barnhart uses Sagarin as a reason to criticize the inclusion of computer ratings in the BCS formula, but Sagarin is prevented from using his most reliable, empirically-validated formula in the rankings because it includes margin-of-victory in the calculation. MOV is a prime reason for why Georgia is not a top ten team, but Sagarin can't use margin-of-victory in this manner because of the knee-jerk reaction to Miami blowing our Nebraska in the 2002 Rose Bowl (as if Oregon and their inept pass defense would have done much better against Ken Dorsey, Andre Johnson, and Jeremy Shockey). So Barnhart is criticizing Sagarin's rankings even though the people who run the BCS have prevented Sagarin from using his best formula...and that formula would prove Barnhart to be even more incorrect than the computer rankings that are used by the BCS do.


Anonymous said...

Margin of victory should be used to determine the national champion in my opinion. Just add up all the points at the end of the year. Ohio State should never have even been on the field with Miami in '02. Blow-out city.

Michael said...

Do I detect a hint of sarcasm? OSU would have been the title game in '02 regardless of whether MOV was taken into account because they were one of two major conference unbeatens and they had played a very challenging schedule.

Fox said...

It is absolutely absurd to exclude margin of victory for the calculation. Does any sane person think that a 1-point victory says as much about a team as a comfortable 3 touchdown win? Obviously they need to cap it to avoid 70-10 games but the current system is indefensible. Well, the whole system is indefensible but this part is worse.

Anonymous said...

Bobby P says:

GGB, be fair now. We all wanted to see Nebraska get their butt's beat in that game.