Actually, in all BCS-accepted computer ratings, all that matters is winning and losing. How much a team won or lost by doesn't matter at all. Of course, that only covers 1/3 of the actual BCS rankings. The other 2/3 are covered by a coaches poll, in which what conference a team plays in matters about as much as how well a team plays in its games, and both of which matter far less than winning or losing a game. The other third? The Harris poll, which might actually be the rating which depends on scoring the most, as it's designed to be a statistically broad representation of the entire NCAA. It's still people voting, which is far less objective than running the numbers. So, in essence, a team's sole goal in a year is to make the championship game. To do that, all that really matters is to win every game. Whether a team wins 13-6 or 45-3, all that matters to conference standings and computer ratings is whether or not it was a W or an L. Conservatively, 90% of what matters to the other 2/3 of the BCS ratings is whether or not it was a W or an L. Which, really, means that for all intents and purposes all that a team should be concerned with in a game is a W or an L, since that's what will determine who plays for the championship, and who is watching. As to how that affects which team is actually better or worse, to take a sole sample of scoring differences is inherently flawed as a model, since that is not the goal of any team. Sure, better teams are more apt to win by a greater margin, and blowout wins and close losses are better signs of a good team than close wins and blowout losses, but beyond a one point victory, it really doesn't matter to a college team in the grand scheme of things, so there's absolutely no incentive to run up any score.