Biggest weakness in the combo of computers or polls is that is merely a snapshot in time and the subjective part is choosing what time. The Alabama win over FSU as you point out would not qualify but, there is not really an objective way to say whether that was not a Top 10 win at the time they played. Just a hypothetical but, say Alabama loses one early in season and then suffers massive injuries later in the season and fades down the stretch. The team that beat them early in the year clearly would be entitled to a Top 10/20 win but, might miss points because of the late swoon. I think there has to be some combination of objective and subjective criteria and the broad based committee approach is probably the closest they can come to fully considering all the objective and subjective criteria.