I would rank top 9 as follows, beyond that I don't know. The 3-7 ranking is difficult. Ultimately even though I classify Clemson and Michigan as having bad losses, they have better wins in my opinion over the teams I rank 5-7 giving them the edge.
1. Alabama 10-0 - 6 wins vs teams with winning records - best wins vs 7-3 USC, at 7-3 LSU, at 7-3 Tenn, vs 8-3 WKU. Questionable performance, none.
2. Ohio St 9-1 - 4 wins vs teams with winning records - wins vs 7-3 Tulsa, at 8-2 Oklahoma, at 8-2 Wisc, vs 8-2 Neb. Quality loss at 8-2 Penn St (perhaps even fluky loss). Questionable performance against 5-5 Northwestern.
3. Clemson 9-1 - 5 wins vs teams with winning records - best wins vs 9-1 Louisville, at 7-3 FSU, vs 8-1 Troy, at 7-3 Auburn. Bad loss vs home vs 6-4 Pitt as 21 pt favorite, Pitt had just lost by 23 week prior. Questionable performance vs 5-5 NC St who missed chip shot FG for the game.
4. Michigan 9-1 - 4 wins vs teams with winning records - wins vs 6-4 UCF, vs 8-2 Colorado, vs 8-2 Wisconsin, vs 8-2 Penn St. Bad loss at 5-4 Iowa as 24 pt favorite, Iowa had just lost by 27 week prior.
5. Washington 9-1 - 3 wins vs teams with winning records - wins vs 6-4 Idaho, 7-3 Stanford, at 8-2 Utah. Quality loss home vs 7-3 USC. Questionable performance at 2-8 Arizona.
6. West Virginia 9-1 - 3 wins vs teams with winning records - wins vs 6-4 BYU, vs 5-4 TCU, vs 5-4 Kansas St. Quality loss at 8-2 Oklahoma St. Questionable performance at 5-5 Texas.
7. Louisville 9-1 - 2 wins vs teams with winning records - wins vs 7-3 FSU, at 6-4 Wake Forest. Quality loss at 9-1 Clemson, LV was actually favored in the game. Questionable performances vs 4-6 Duke, at 2-8 Virginia, vs 6-4 Wake Forest (WF led the entire game into the 4th qrt).
8. Penn St 8-2 - 4 wins vs teams with winning records - wins vs 9-1 Ohio St, vs 7-3 Minnesota, vs 7-3 Temple, 6-4 Iowa. Quality loss at 9-1 Michigan, although PSU was dominated. Bad loss at 6-4 Pitt where Pitt led big before late comeback bid. Questionable performance at 5-5 Indiana (misleading final).
9. Wisconsin 8-2 - 3 wins vs teams with winning records - wins vs 6-3 LSU, vs 8-2 Nebraska, at 6-4 Iowa. Quality loss at 9-1 Michigan, home vs 9-1 Ohio St. Questionable performance vs 2-8 Ga State.
I personally have always favored a 4 team championship system where only conference champions are eligible for the pool of teams. I wrote a detailed blog on this topic almost 10 years ago on Phil Steele's website where I gave examples of how it would work in every BCS year. This is different from Phil's article in his annual magazine as Phil doesn't have a conference champion only provision. I believe in using the conference races as an ongoing inseason playoff and every team can start the season legitimately believing they have a chance at a national title with the goal of being a conference champion as the first qualifier. I don't care how the rankings are established, it could be all computers for I care, I actually think I prefer a collection of computers like the BCS used over human pollsters.
Two teams from the same conference should never compete for a national title. 2011 LSU-Alabama, 2003 Oklahoma-LSU and 2001 Nebraska-Miami shouldn't have happened as Alabama, OU and Nebraska would not have been eligible. This method also extends access to the pool for teams such as MWC, MAC, Sun Belt, etc. They don't have to be in the top 4. They just have to be among the highest 4 ranked conference champion...which could be #7, or #10, #12, etc. You go as far as you have to for 4 conference champions, but historically the pool is typically filled with teams among the top 10. Back in the day when people argued about Utah or Boise or TCU competing for a title..that model would've included them.
I don't believe in a system that allows for rematches. So then I don't believe in conference championships as they are determined in many conferences currently. I believe in single elimination. If team A already lost to team B, why create a rematch? It minimizes the significance of the first game and magnifies the significance of the second game for no reason other than the second game is played last. Nothing wrong with the old way of doing things...everyone plays every conference opponent and if tie breakers are needed, use tie breakers to determine the winners (and don't outsource the tie breaking process to a poll, if the normal tie breakers can't break the tie then use fewest points allowed, or most points for, or whatever means is necessary to break the tie in house).
But using a conference champion only model for 4 playoff spots takes so much subjectivity out of the equation and reduces the importance of polls. A poll would still be used to determine the top 4 conference champions, but it largely puts a team's destiny in their own hands.