The Harris Poll - Revisited

The series so far:

Intro
Voters
Analysis

A few other points that I've been mulling over or were raised in comments:

I. Replacement Voters

Yesterday 6 voters were added to the poll to replace the ESPN conflicted voters. They are:

Bob Pruett: Marshall HC and player. Asst. Coach at Florida, Mississippi, Tulane, Wake Forest.
Ray Goff: Georgia HC, assistant and player.
Jim Mandich: Michigan player.
Joe Jacoby: Louisville player.
Denny Aldridge: Texas player. USA Today article says he was an assistant coach, but I can't find anything.
Frank Sadler: South Carolina assistant. Kentucky player? High school coach.

I don't think this changes all that much from my earlier analysis. It adds another lifetimer, though I'm not so sure Ray Goff is as die hard a Dawg as some of the others (somewhat bad terms). Pruett has weak ties to Florida, so it costs me the good line about the big Florida schools, but the undercurrent remains - Florida is still underrepresented. Most of the new guys have ties to the South.

II. The Feminine Mystique

A few columnists and others have brought up the readily apparent fact that there are no women to serve as voters for the Harris Poll. While I tried in this series to focus on the facts, this issue kind of forces me to have an opinion. I don't think the failure to include women is necessarily a flaw in the voting process. I think of it as more of a public relations problem for the poll, and something that gives columnists have an easy column to write. My opinion has nothing to do with whether women should be voters, or whether they are able to be a good voter.

The point is that the flaw with the poll is inherent biases - relationships to schools, employment, geography. All voters have this, no matter the race, gender, etc. A woman is no less likely to have these flaws than a man. Now, if all women were to have a particular bias - as in toward a single school - then the lack of women might make the poll even worse. But the fact of the matter is that women don't have universal biases when it comes to college football. If the poll were only made up of women, and those women were to have the same connections that the current voters have, the poll would be just as screwed up. So in terms of the validity of the poll, and not whether it's right or wrong to include women, the gender makeup of the poll doesn't really matter.

III. Solutions

One of the problems I have with my own analysis is that I don't offer any solutions. I purposefully avoided saying "the poll sucks, the system sucks, let's have a playoff" because I don't really think that way. I think the system in place now isn't good, but it's all we've got. The goal should be to get the least biased voters who actually watch the most games and know enough about the teams in order to have an informed opinion. Coaches obviously are biased, since they have their own interests to deal with, and they have something else to do on Saturdays rather than actually watch the games. The Harris Poll voters are almost all biased, and who knows how many of them will take the job seriously. I don't even think the AP voters are all that much better, since a good percentage of them are covering a particular team's beat and therefore aren't watching several games. Unless we breed a species of precogs that only watch college football 24 hours a day, polling is a pretty bad way to crown a champion. But it's the way we have. And I don't like pointing out flaws without offering alternatives (although I do it all the damn time).

So here's my alternative: The Gunslinger Poll. I tracked this last year, and it didn't end up much different from the actual polls, but it does have the potential to do so. Here's how it works. Every team at the beginning of the year looks at their schedule and sees 11 or 12 games it has to win in order to have a chance at being considered Champions. Nobody knows how good every team on the schedule is at the onset. Nobody knows anything. But every team knows they have to win every game. Now, once the season goes along, we can see which teams face tougher schedules compared to others. So the system is this: Rank teams first by won-loss record. More wins a team has, the higher ranked it is. An 8-0 team is ranked ahead of a 7-0 team. The more losses a team has, the lower ranked it is. Among teams with the same exact record, strength of schedule is the tiebreaker, with one exception. Among two teams with the exact same record who have played each other, and where the team with the worse strength of schedule has beaten the team with the tougher strength of schedule, the team with the weaker SOS will be ranked immediately ahead of the stronger SOS. In the event of three teams which have the same exact record and have played each other, with each team winning one and losing one, the SOS will control. Schedules are set years in advance, and the kids competing on the field have nothing to do with scheduling, so I think won-loss record should be the first and foremost category. Does this lead to some odd results? Yes. It may favor easy schedulers, but we'll see. I'll give the Top 10 of the poll each week, but not until decent SOS ratings come out, probably in mid-October.

That's my alternative, and I know it's not one everyone will support. But hey, it's what I've got. At a minimum, it takes bias out of the equation, which the BCS system can't say.

Again, thanks for all the links and visits. Feel free to leave comments. I've got a thick skin, so tell me if I'm an idiot.