Could the AP Poll employ a better voting method?

 A few weeks ago, I was scrolling through TikTok and got into an argument with a Virginia Tech football fan about their aggravations with the Associated Press College Football power ranking poll (AP Poll). They felt that the poll was not reflective of actual results, and that pre-season perceptions were being given too much weight. This triggered me to look a little further into how the rankings are constructed on a week-to-week basis. After reading about how the voters are chosen and how the votes are totaled, I tried to tell the VT fan they should look into it too, and they responded “I’ll learn it when they get their **** together”.

Thanks to public choice, I get to blog about this interesting application of Borda voting. The AP poll uses the Borda method to rank the top 25 (of 130) college football teams weekly, with first-place votes being worth 25 points, second-place votes being worth 24 points, and so on. Borda voting does not satisfy the Condorcet criterion where the winner of all head-to-head matchups is the winner of the aggregate election. To get an idea of how the Condorcet (head to head) best team may not actually end up in first place in the poll, think back to Mr. Coppock’s example of the Louisiana Gubernatorial election, where the Condorcet winner was knocked out of contention. This example works the exact same way but with more "candidate" teams and more voters. In this respect, I agree with the VT fan that the poll’s method is not perfect, so I will explore some alternatives. 

The VT fan is correct that the poll could be better, and under a ranked pair voting system the weekly victor of the AP Poll would be guaranteed to be the Condorcet winner. The problem with this, however, is that a ranked pair ballot for 130 teams would force sportswriters across the country to consider over 15,000 hypothetical matchups each week, a prohibitively expensive proposition. Other options for the poll would include other single-round non-Condorcet methods, such as ranked-choice. The reason that Borda is preferable to ranked-choice here is that Borda lets voters rank the team they most believe to be the best, and 24 others, rather than all 130. 

For the reasons stated above, I do believe that the Borda Count voting method certainly has its imperfections, but is the best voting method to accomplish the very unique task of ranking college football games each week less than a day after the games happen.