[BostonCard]

I would guess they're also a little more forgiving to a team losing on the road to the top team in the country than humans are. Granted $C lost horribly, but most human polls oddly punish teams for losing games they're expected to lose. The computers tend to punish the number in the loss column less than who they lost to or how they lost. Of course, since it was a blow out, I don't fault any humans for throwing them out of the polls. However, I would not have dropped UCLA from the polls for losing in OT to a team that they weren't actually favored against.

I disagree.  Even if UCLA was marginally expected to lose the game prior to playing it, it was not disfavored overwhelmingly.  You might assign TAMU an a priori 60% chance of winning and UCLA a 40% chance of winning.  Once the game is played, the uncertainty goes away, and TAMU has 1 win, while UCLA has 1 loss.  So, even if the loss wasn't unexpected, the uncertainty is now gone, and UCLA no longer has the possibility to upset A&M.  Thus, you would expect the computers to punish UCLA.  Moreover, the spread was only a couple points, and UCLA lost by a touchdown.  Although it was overtime, and therefore can be considered closer than the score, that I know of, most computers only look at score differential.  So I would expect UCLA to be punished, at least a little bit, by the computers.

From the humans perspective, it boils down to a ranked team losing to an unranked one.

BC

[BostonCard]

I would guess they're also a little more forgiving to a team losing on the road to the top team in the country than humans are. Granted $C lost horribly, but most human polls oddly punish teams for losing games they're expected to lose. The computers tend to punish the number in the loss column less than who they lost to or how they lost. Of course, since it was a blow out, I don't fault any humans for throwing them out of the polls. However, I would not have dropped UCLA from the polls for losing in OT to a team that they weren't actually favored against.

I disagree.  Even if UCLA was marginally expected to lose the game prior to playing it, it was not disfavored overwhelmingly.  You might assign TAMU an a priori 60% chance of winning and UCLA a 40% chance of winning.  Once the game is played, the uncertainty goes away, and TAMU has 1 win, while UCLA has 1 loss.  So, even if the loss wasn't unexpected, the uncertainty is now gone, and UCLA no longer has the possibility to upset A&M.  Thus, you would expect the computers to punish UCLA.  Moreover, the spread was only a couple points, and UCLA lost by a touchdown.  Although it was overtime, and therefore can be considered closer than the score, that I know of, most computers only look at score differential.  So I would expect UCLA to be punished, at least a little bit, by the computers.

From the humans perspective, it boils down to a ranked team losing to an unranked one.

BC