In theory, right now, you can design a ranking system that minimizes instances of a team that lost to another team being ranked ahead of that team. If that's all you do with your system, you will end up with about an 8% error (meaning 92% of the games played would fit into your rankings). As a predictive structure, this type of model gets you to about a 65% win rate.
Most good computer models, which use scores and results and home-field considerations and sometimes even statistics, have about a 15% error and more than a 70% win rate.
An experienced better, who can analyze match-ups, handle injuries and massage the numbers, can get as high as 80%. These scout-types don't bother with full rankings. It's a pointless exercise.
Schedule strength is incredibly difficult to resolve in college football. You have 128 FBS teams. You have another 80 FCS teams playing them once or twice. You have a 12-game schedule. Home-field gives you a 15% advantage. The difference between the best teams and the worst teams varies a lot more than in the NFL. Marshall would probably go 2-12 if it played the entire SEC.
Back in the "old days," everything was about winning your conference. A bowl game was the reward for a truly great season. Now, bowl games are awarded for 6-6 seasons, and we try and determine a national champion based on very little information. A four-team playoff is only slightly more valid than a two-team playoff. But any playoff is better than no playoff because at least those games are on more neutral sites. I'd like to see a 24-team playoff, with the top eight conference-winners receiving a bye.
Auburn lost at Mississippi State and Mississippi State is undefeated, so the way we do polls today, MSU is #1. But the computers mostly say Auburn is better, because Auburn has gone 7-1 with a schedule that is much tougher than anyone else's. Not that MSU hasn't had quality wins (it won't help this perception or the computer rankings, though, having their FCS opponent come up next weekend while Auburn hosts A&M). Human polls are more politics than analysis. And I think this new committee, especially since it isn't allowed to use scores of games, will create more controversy than it solves.