(10-17-2023 01:02 AM)JSchmack Wrote:  ... The #1 reason for inexplicable poll stuff is that it all starts with the preseason poll and voters ADAPT FROM THAT based on each week's results. ...

Yes ... I wouldn't argue that the voting polls are consistent. But its already a result in social science that you can add up the votes of people with different, but consistent, preferences and get inconsistent choices as a result.

And a lot of votes are plagued by the anchoring bias you describe. Joel Klatt on a recent podcast was explicit that he doesn't "punish" a school for having a bye week, when a rational ranking would sometime say, "oh, with the extra information from seeing the games played this week, I have to change the position of that other team that I didn't see play."

Then you add on top the voting system for a lot of voters is "it feels like it to me" after watching some but not all of the games, where adding up the gut reactions of people who have seen different sets of games and have "heard about" others, and then expecting them to be consistent is asking more than they could ever consistently deliver.

The CFP committee is going to suffer from the same group choice paradoxes as the voting polls, but at least the members of the committee will have seen more of the games, and the discussion will include people who have seen play by all of the teams up for consideration.

But we still might be cautious about confirmation bias ... just because there is going to be irrationality in some of the rankings, doesn't mean that every single one that strikes us as odd is necessarily an irrational ranking. Sometimes, we are the ones who would be voting an irrational ranking if we were one of the AP voters.

(10-16-2023 10:11 AM)ken d Wrote:  There are three things that matter which computer algorithms can't take into account: weather, injuries and intensity. Beating a team when their star QB isn't playing, or beating a passing team in a heavy wind or driving rain probably shouldn't count as much as beating them in health and perfect weather.

When you look at rivalry games (like Army-Navy), where the adage is that records don't matter, it's clear that playing with intensity through the final whistle matters. No doubt, when a ranked team is playing a decent unranked team, the unranked team plays with more intensity. Coaches hate that about their teams, but it's human nature. So when a G5 team punches above its weight against a P5 team, fans of the G5 (and not just those of the particular G5 team) want to place more credence on the relative strength of the G5 teams than is probably warranted.

The fact that G5 teams tend to get ranked in the 20-25 range if at all isn't an indication of voter bias, no matter how much G5 fans want to believe it is. If anything, those teams are likely over-rated, not under-rated most of the time.

Injuries, weather, intensity and turnovers are all part of the game.

I don't think that one can separate or remove those from the game outcomes.