(12-01-2013 03:11 PM)Caelligh Wrote: (12-01-2013 02:48 PM)georgewebb Wrote: (12-01-2013 02:22 PM)texd Wrote: Who said a coin flip should determine which team is in the championship game or Liberty bowl?
I said it, sort of: I said that (given that the difference in the computer rankings is statistically insignificant and that the coaches poll is so obviously subject to self-serving manipulation) deciding by a coin flip would be no less rational*, and would have the added virtue of not being manipulable...
...While I stand by those observations, I certainly expect others to disagree -- and I acknowledged as much in my post....
I think you are spot on, George. Complexity does not necessarily improve accuracy. A coin flip in this case would be fair, transparent, and easily understood by observers. By the time you get to the 3rd or 4th level of a tie-breaking protocol, you've already established that the teams are fairly even in skill and equally worthy of hosting the game.
I disagree. When you have two teams that, in spite of being in the same conference, have very little in overlap, getting to the third level of a tiebreaker between divisions is no sure indication of parity. This is a function of conferences growing in size.
Criterion 1: Record - Because one division can be much stronger than the other in a conference, records may not reflect the actual strength of the teams. For example, had Rice beaten North Texas, I could still make a reasonable argument that Marshall is better. So would home field for 8-0 Rice have been "fair"?
Criterion 2: Head-to-head result - Unless there is an exception for a rivalry or a scheduling quirk, teams in different divisions play each other in the regular season a maximum of twice in each four year period. Head-to-head is quite often not applicable.
It was very simple for us to get to the third level of this process without determining anything about the teams involved, and even then, it could have easily ended at the second determining factor with an outcome that would have understandably rankled some people. Rice has nothing to complain about in this case. A coin flip would not have been equitable, and though the polls are not the ideal mechanism for solving this, they did a better job than a coin flip would have.
Also, in the polls, I think that there is an assumption that any two coaches will both have their buddies voting for them; teams are thus only separated when voters who are not tied to the schools in question submit their rankings. That being case, Marshall got the votes from somewhere because they probably deserved them in the eyes of a few "neutral" observers. That will suffice for resolving this scenario.
All this being said, I agree with the statement that complexity does not necessarily improve accuracy. That is for certain.