Long post incoming. With the coming importance of SOS, especially if we get the top-4 model that the SEC (and probably Big 12) wants, or a plus-one model, I think we'll start to see a new interesting dynamic in college football. If your team is good, you are going to care a great deal about how well other teams in your conference do in non-conference games, and if they lose too many, that could screw your team over.
Here's
Sagarin last year. Their SOS should be a decent measure. You can see that last year the Big 12 was the highest rated, then SEC very close behind. Then big gap, then B1G, PAC 12, Big East, etc. We'll see why that is soon.
Scroll down to where they show ratings and SOS for teams grouped by conference. Notice anything odd?
Isn't it remarkable how similar the SOS is for each conference? For the ACC, they may be better than the little sister FBS conferences, but compared to the big 4, their SOS blows ass. Despite finishing with the best record of 11-3, VT couldn't crack the top 25 because of that crappy SOS. Notice how everyone in the Big 10 is OK, nothing special, but not bad either? (and contrasted with VT, they have 2 teams with 3 losses, both of whom were easily top-25) Now look at the Big 12. Everyone's got a crazy-good SOS. 3 losses got OU a top-4 rating. Scroll down to the SEC, and you can see they have the 2nd-best SOS's, in the teens and 20's, over there 3 losses would have gotten you rated close to top-10.
Obviously the computers are giving credit to strong conferences and less to weaker conferences, obviously being able to play games and only lose 3 in the Big 12 and SEC means a lot more than in the ACC. How did the computers determine that? Did they just somehow "know" that the Big 12 and SEC were tougher without being able to see the games and have a human's intuition and judgment of talent from whats happening on the field? Obviously not.
2011 Non-conference winning percentages:
Big 12 .844
SEC .814
Big 10 .625
Big East .568
PAC 12 .529
ACC .488
The Big 12 lost only 5 OOC games, and the SEC, playing more OOC games, lost 8
Not perfect, the PAC 12 is generally rated higher than the Big East for SOS, they do look at how good the OOC teams are, and the Big East scheduled a lot weaker than the PAC 12, but generally you need, NOT just YOUR team to win non-conference games, you need everyone else in your conference to also win non-conference games. 4 OOC games are better than 3, because if a few teams drop 1 or 2, they've got more of a margin of error to win the other 2 or 3.
This is also going to present an interesting, brand-new dynamic. Before when you had polls you just relied on writers and coaches to believe their eyes, and when you had BCS v1.0, everyone had an auto-bid so it was all cool. You might be vaguely interested in the conference scoreboard, but if a bunch of people in your conference sucked that weekend, big deal, doesn't hurt you much, if at all.
NOW, you should be interested in the non-conference scoreboard a great deal. If you have a really good team, and you think your team has a chance to get in the playoffs and win a title, then you want everyone in your conference to win every non-conference game, and every time a team in your conference loses a game to another conference, it hurts your team's SOS. People wont realize this right away, but in a few years I can picture fans from the strongest team in the conference coming over to the board of someone who lost to a CUSA team to bitch about how much that loss might hurt.
In order to be a really good, awesome conference, and be regarded as such by the computers, you need a .800+ non-con winning percentage. In order to be at least treated respectfully by the computers and still get in the playoffs with 1 or 2 losses, your conference probably needs at least a .650ish OOC winning percentage. If 12 teams are playing 3 OOC games each, you don't want to see more than 12 or 13 losses. (or no more than 7 if you want the conference to be elite that year) If you are playing 4 OOC games each, you can afford up to about 17 total OOC losses, or up to 9 or 10 losses to be considered elite.