Without an eight team playoff I propose this system - First, understand that OOC conference schedules are really too small a sample size to definitively conclude anything about how strong a conference is. Second, schools set up OOC conference schedules years ahead of time and have different priorities when they do so. Third, we need to understand and differentiate between who "deserves to go" (their record) and who we think "is the best" (eyeball test, margin of victory, and how they are playing lately). Too often these two concepts get muddled together with bias creeping in and giving more weight to one over the other in any otherwise sane discussion. Just separate them completely.
The four spots should all go to power 5 conference champs regardless of their records. Teams with fewer in-conference losses make it over teams with more (yes, all conferences need to play same number of in-conference games). If teams have an equal number of in conference losses then the first tie-breaker is how many wins they have against OOC power 5 conference teams. If there are still unsettled spots then team's names are just drawn from a hat.
For this year that means we would have:
IN
FSU (no losses)
tOSU (no losses)
Two drawn from the proverbial hat
Oregon (one loss, one OOC power 5 win)
Alabama (one loss, one OOC power 5 win)
TCU (one loss, one OOC power 5 win)
OUT
Baylor (one loss, no OOC power 5 win)
No committee even needed so far. I'd be tempted to add another tie-breaker before the hat and that would be "number of power 5 OOC games scheduled".
Once the teams are established, a committee can then rank them by how strong they think they are. So teams that are higher on the list of qualifiers might not be seeded as high as teams that are lower on this list of qualifiers.
Anyways, this system removes the bias to a large extent and rewards teams for winning their conferences and playing tougher OOC games.