**** this ****.
Incestuous conference play/minimal noteworthy inter-conference play (across all of college football, not just within the SEC) and a short schedule make it impossible to accurately rank teams from different conferences against one and other. Add in preseason rankings which basically act as weights and you create a bias towards whoever happens to be ranked highly at the beginning of the season based on nothing but intuition (recently this has been the SEC) because nobody has done anything this season.
The strongest way to make accurate rankings, accepting the fact that you can't have 50 game football schedules, would be to kill preseason rankings, remove "conference play" altogether, create schedules focused on playing as many games as possible, randomly, between the top 64 (or however many) teams to build a highly connected network of results, and harness the powers of the transitive property (A>B & B>C therefore A>B) within that network.