fundamentally an ill-posed problem. For example, consider whether BYU or SMU should be ranked higher: they have the same record, but BYU has a loss to an unranked team, while SMU has a loss to a ranked team. Of course, that ranked team happens to be BYU. You can go round and round, but at the end of the day, some person, somewhere has to decide whether head to head victory + a bad loss to someone else is better than having lost only to a ranked team.

You could make this decision beforehand and risk having the computer spit out something that will seem wrong, or you can make this decision ad hoc based on the teams involved.

At the end of the day, the regular season does not provide enough information to definitively rank the teams. We have to start with priors and then the season provides sporadic updates. I think it's unsolvable, but it would be nice if we didn't always err on the side of anti-BYU bias.