that are probably not taken fully into account. I have not read up on Indy Coug's analysis but I'll just give some thoughts without any prior knowledge.

1. No matter what the statistical accounting is an actual Win should carry a very heavy value. Maybe that is already happening but I suspect it is still weighted too heavy elsewhere. There are almost never rematches and certainly not in the regular season. A win is a definitive and unchanging point of data which carries the overriding significance of being the entire point of the sport being played. The more wins against other teams that also win should therefore carry significant weight as well.

2. This is probably also taken into account but the more a team wins the more that has to carry weight. The closer a team is to undefeated the better they are and this bears out with just a glance at any past season. Losing one or two games, no matter the competition is an incredible accomplishment which very few teams achieve.

3. The loses are almost as telling as the wins and the most relevant factor is how close to undefeated the other team you lost to was. Frequently the margin of error in a game is simply turning the ball over a couple times. You could statistically be the best team in the country and never had turned the ball over and lose the game to a bottom quartile team when you have multiple turnovers. Losing or winning against good and great teams is way way more relevant then the score or yardage stats that come with the game particularly if there are outliers there.

4. I would be intrigued when the computer polls start compiling data. If anything from a prior season is used or anything from recruiting rankings then the whole system is built on a stack of cards. No way human bias isn't coming into play in those circumstances.

5. If #4 is not happening a major issue is that you only have 12 major data points to compile information from. Given the number of changing data points within those 12 major ones (players, coaches, injuries, systems, conferences, facilities, away and home games, weather, bye weeks, distance and duration of travel, and much more, no data can be trusted only assumed to be accurate in a year to year comparison based on the same limited principles. If post season and bowl game data are also used then you have only a portion of the teams with a significant variant on which to judge standard deviation.

Finally since the public is given no expectation that the committee is committed to some kind of data review it seems hypocritical to base any ranking on such computer analysis. The lack of disclosure on such things actually drives behavior from teams, players and coaches. No one is blaming a coach for running up a score at the end of the game if game stats and margin of score are significant factors in moving your team up the rankings. Right now there is perception in the public and coaching that doing so is classless but all that goes out the window when winning by an extra touchdown moves the needle on a national scale the could effect your ability to play meaningful games in the post season. Shoot, I could actually see coaching contracts change with such data. I could also see offenses and defenses change because as observed in other posts, a slow burn defense is likely a liability if yards gained, time of possession, field goals allowed, and 3rd down % is all on the table. Starters will stay in the game and game management becomes a whole new expectation because its not enough to win and that should be made clear to everyone.

I would love to see something more similar to NET but even that has some flaws. The conspiracy theorist in me thinks there probably is some heavy use of data and many of the big money teams are aware of the parameters while the rest of college football has to luck into hitting stats the committee cares about.