It is a mafia of people manipulating the system for money. The Human element will always skew logical decisions when it comes to money. We fans need to demand a system based on data and a commissioner of football, which will never happen. Looking at Sagarin's data, for example, clearly tells the story of SEC teams in the top 30 with way more difficult schedules. That, combined with wins over top 30 wins should carry way more weight than going by Human polls. Big 2's record is 3-0 for both of them. The highest rated SEC teams are 5-0, 4-1, 6-2 & 6-1. Even Vanderbilt has played double the number of top 30 teams than any in the big 2. I still don't understand why us fans put up this this farce.
Well first, I've said all along that the playoff is about inclusion. The reason fans put up with it is ultimately that's what they want. They don't want things to actually be fair, to be logical, to be consistent. They want them to be slanted towards inclusion, so the most deserving are left out in favor of inclusivity. Just look at the current rankings, BYU at 11? Utah at 12 ahead of Vanderbilt? The committee should be doing better than any computer, instead they are doing demonstrably worse.
As far as Sagarin, I rely heavily on them, but I would note that even the computers are outdated. When the SEC added Oklahoma and Texas to what was already the strongest conference in football they created an unprecedented situation.
A side effect of that by the way is BYU, Utah, and Texas Tech ranked so high. It's pure coincidence that a conference without it's best teams magically sees the remaining teams get much better. Or you know, it's just that their schedules are now much easier. Anyway, part of the issue is that the models still favor wins and losses a bit too much. This is a relic when the best teams were spread out among six conferences, and now the power is largely concentrated within 2. Losses are an inevitability that mean less than they used to.
The second issue is the models were never really built to take into account conference championship games. It's all percentage based, so if you play 4 games, or 20, it still ranks teams as though what they did was somehow equivalent (check their 2020 rankings). This isn't true though, if Alabama plays in a conference championship game against Texas A&M, it's more than just a percentage increase in the formula. It's a whole entirely new game, but none of the models really know what to do with that because they're not calibrated properly. The SoS should function as a multiplier, for instance each top 10 play you get a X multiplier, each top 20, top 30, so on, and then you apply that to the other ranking components.
Unfortunately it's just not done that way so for a team like Alabama, the SECCG is only an oppurtunity to lose in the rankings, which to reiterate are skewed far too heavily to punish losses.