November 24, 2008 · 6:47 AM

Horn tootin’ time

With the Oklahoma-Texas-Texas Tech waltz in his mind, Dennis Dodd frets about conflicts of interest and the Coaches Poll.

Ahem… there’s a modest proposal we here can suggest that he may want to keep in mind.

4 Comments

Filed under Media Punditry/Foibles, Mumme Poll

4 responses to “Horn tootin’ time”

  1. Macallanlover
    November 24, 2008 at 1:26 PM

    I have to support Dodd on this one. Not only is there a huge conflict of interest here, there is just no way an active CFB coach has adequate time to follow teams around the country closely enough to evaluate their relative strength. I think from what I read, CMR makes an effort to stay up late and watch highlights but that isn’t really enough, and few coaches make even that effort. I realize there are some buffoons with the media who vote in the AP Poll as well, but at least they have the time to do the job if they choose to take it seriously.

    While we are at it, time for someone to challenge the “objectivity” claims of the computer polls as well. If they are totally “objective”, and all “subjectivity” is removed, what accounts for the wide variance in results? When I ask a calculator for the square root of 50 I always get the same answer, but when I see the computer polls I get a variety of answers. Think ol’ Jeff Sagarin doesn’t have a few biases when he sets up the program, or massages the results? If you put the same true factors in, how do you get different garbage out? Inquiring minds want to know.

    Like

  2. Hackerdog
    November 24, 2008 at 3:05 PM

    You get different results because there is no universally accepted right answer. It’s like asking computers to choose the best tasting pie. The answers will depend on the factors analyzed and the weight given those factors.

    The same thing happens with the human polls, BTW. The coaches and writers don’t agree who the top 25 teams are, or in what order they should be listed. Why hold computers to a different standard?

    Like

  3. Macallanlover
    November 24, 2008 at 3:52 PM

    Hacker, I was not looking for the “right” answer (since there is no way of knowing that), just looking for the “same” answer. It seems to me if you start with the same 119 schools WITHOUT BIAS as a baseline, enter all the hundreds of interactions (scores) every week of the season, then juxtapose those against all the previous interactions, you should end up with the same “power”, or “performance” ranking of the 119 schools regardless of how many computers do the work. In my technologically, naive mind, that is what I felt the computers were supposed to do because it would be too much computation for a person to take on.

    If the variables were all the same, why wouldn’t all computers spit out the same results? My point was that someone is monkeying around with the weightings which negates the alledged objectivity of the computer polls. What is the difference with human bias if it occurs when setting the start points on a computer program, or occursduring and throughout the season? I was all for the computer poll (s) until I saw it was just another person’s opinions disguised as something that was supposed to be indisputable. If it worked correctly, we should only need one other as a check on the orginal. Obviously I was wrong in my assumption so I have the same regard for the computer polls that I do the human ones. Their opinions are no better than my own, or yours.

    Like

  4. Hackerdog
    November 24, 2008 at 5:55 PM

    The different computer polls use different formulas to rank the teams. That’s why you get different results. It’s the same reason the AP poll uses lots of different writers instead of just one. It kind of evens out the biases.

    Computers are better, IMO, than pollsters because they don’t have the bias that pollsters do. Steve Spurrier said he always ranked Duke #25 in the first coaches poll because that was his first job. Hal Mumme inspired the Mumme poll by ranking Hawaii #1 in the country, so that a WAC school could go to a BCS bowl. Some pollsters will rank teams with an East coast bias, or a West coast bias, or a major conference bias. Whatever. Computers don’t do that.

    Computers analyze wins and losses, strength of schedule, common opponents, and other factors to achieve a ranking. There is no bias because every single team is subjected to the same formula. Now the formula may not be perfect, but it’s the same formula for all the teams.

    I like that better than saying that a West coast 8-3 team is better than an East coast 8-3 team because your alma mater was a West coast team and you just like them more.

    Like