They always made sense to me in college basketball. Maybe the difference is that there are 3-4x as many games to create data from in BB, so the results are more likely to pass a sanity check.

I was looking at this week's Sagarin ratings which has Kansas (7-0) as #2 overall, even though their schedule is 117th. Hawaii's (also 7-0) is 170th but they are rated #49.
As an aside, UNI's (N Iowa, not a Div I-A team!) schedule is 128, about the same as Kansas to this point.

According to Sagarin's "predictor" rating, Kansas is a favorite against every team in the country--including LSU and Ohio State--except Oregon. Oregon is a spread favorite against EVERY team in the country even on the road.
Would bookies really have Oregon favored over OSU (or LSU or whomever) on the road?

I realize BCS does not equal Sagarin, but the BCS does use the ELO_Chess equation [similar to "the predictor" above] in addition to the polls.

[Note: I'm sure it's obvious from my question that I don't bet on sports.

I think they will shape up here real soon. None of those undefeated teams has faced a team in the top 30 in the BCS, or one of the top three teams in their conference. The computers will get closed to reality once the heavy hitting starts. The rankings aren't generally that far off.