We were clearly more talented than Florida and Arkansas.
That's not looking at a spreadsheet, although 99gator has even done that for us. That's looking at the real world outside the back door. You say Florida is supposed to be more talented than us. Maybe that would be the case if the rival rankings over the last four years could still be weighted equally.
Your model says we are more talented than Oklahoma. I'm willing to say we were pretty even against them. Which means, Butch clearly majorly underperformed to allow them to win the game!
The model says Georgia has more talent than us. It's not clear to me, and it's especially not clear when the most important member of their football team - the QB - is also the biggest question mark on the team. Throw in they lost their best player overall on the first play of the game, and I think the model prediction is very much in question.
I certainly have no problem with the idea that talent is the most important factor in determining the outcome of college football games. I would say it's almost infallible with teams of different quintiles.
However, we are talking about a model that is a third wrong. Where? If it is nearly infallible between teams of different quintiles, where does it fall over?
Clearly within quintiles.
I'll be honest, daj. You're in a place that I've been for a long time. Does a democratic society have any meaning in a data driven world? If the "veracity of the data" can't be argued, then democracy has no meaning. We should let the 10 best technocrats run the whole show. We shouldn't even play these games. We should roll the 12-sided dungeon and dragons dice and assign the most talented team numbers 1 - 8. Or 1 - 4 and 9 - 12. It's all the same. Keep in mind too, all this "data" is absolutely subjective. It's good subjective data, but guess what, F = G m1*m2 / r^2 is just an approximation as well.
