From the mailbag, regarding the RPI:

Why is it a flaw that the RPI does not take into account margin of victory? What do you think would happen if it did mean something? Coaches would be running up the scores and that is not good for the game.

True, you wouldn't want an official system like football's BCS, which gives teams an incentive to run up the score. Plus, in college basketball, late-game substitutions after the outcome has been decided often mean that the final score doesn't reflect how close the game really was.

There is a difference, however, between losing by one in overtime and losing by thirty. For certain games, the margin of victory should affect how much weight we give it. It is just another subjective factor to throw in the mix on top of the basic RPI calculation.

Note also that ratings systems based on margin of victory are actually, according to Sagarin, a better predictor of future games.