A Solution To The Human Polls

In the BCS ranking system, 1/3 of the ranking is determined by an average of 6 computer polls. Of the remaining 2/3, half is determined by the Harris poll of former players and such and half is determined by the coaches poll. Does it seem a little backward that the polls most open to error and bias count for 6 times as much as one computer poll? Me too, so here’s a solution.

First, let’s make the computer polls count for 50% of the formula and the human polls count for 50% of the formula. You have 6 computer polls right now. The highest and lowest ranking for a team is dropped and the remaining 4 get averaged to determine your computer ranking. What we need is 6 human polls that act exactly the same way. Highest and lowest ranking get dropped, remaining 4 are averaged.

So what 6 polls could be used? It seems every website has their own rankings and there are lots of official polls out there not currently used by the BCS. I propose we continue to use the coaches poll, only we allow all 120 coaches to have a vote. If you are going for a representative sample, I don’t know why we have 60 something coaches, that’s already more than half of them. Might as well just use them all, it can’t be much harder to get a few more interns to fill out their coaches weekly rankings for them.

The next poll I would use is a journalist poll. Every newspaper has a local sports section with plenty of staff. You know they have someone covering the university’s athletics programs. This person will be pretty familiar with their university as well as others so that they can cover their sports. The AP Poll is supposed to be like this, but they refuse to let their poll be used in the BCS formula. It can’t be hard to make another reputable journalist poll, if one is not already out there.

The third poll I would use would be the scouts poll. NFL teams have lots of scouts. They are always checking out the college talent. Not many people are more familiar with college football than pro scouts. They are already ranking the players and everything. I think a scout poll would add a lot of diversity, value and perspective.

The fourth poll I would use would be a players and fans poll. After all, college football is only about these two groups of people- those who play it and those who pay to watch it. Most might not take it seriously and fans would usually vote for their teams and all that garbage, but it would be fun to let the players and fans help decide the BCS rankings. After all, highest and lowest rankings are dropped, so if they went crazy it wouldn’t matter much.

The fifth poll I would use is a media poll. This would be distinguished media personalities, such as commentators, sports anchors, bloggers and others. The guys reporting on these games probably watch more than anyone. Same with their stats people behind the scenes. These folks should get a poll together.

The last poll I would use would be a legacy poll. These are all the former players, coaches, directors, commissioners and others. For example, all the past Heisman winners get a vote for the Heisman each year. Everyone who votes for the Heisman winner should also get a vote in a poll for weekly rankings. You know people who were in the game for years are still closely attached to it. Let’s use them in a poll. This is similar to the Harris poll now.

I believe if you have these 6 human polls, you really negate the possibilities of bias creeping in. All that bias, opinion and error balances itself out. Plus you always drop the highest and lowest, so it would be really hard to intentionally inflate or deflate a team’s rankings.

If you have the average of 4 human polls and 4 computer polls evenly weighted, the BCS formula becomes a lot more fair, balanced and transparent. Even if these human polls are not eventually used, I believe the BCS should move towards a ranking model like this to improve their system.

Comments