Over the last 25 years we keep changing who decides or how many we choose, but we never address the data that drives decisions. College football generates bad data, and expansion has only made that problem worse. The only reason we can even sort of pretend any of these solutions have worked is because most years see one or maybe two teams so stacked that virtually any method would have generated a plausible result.
If someone really wanted to address this, then address the data. That means shrinking the set of all teams down to a manageable number such that you create a web of comparison across all teams. 130 is way, way too many. Half of that may work if they only played each other. Conferences need to shrink by about half so everyone can play a true round robin with balanced scheduling. Major powers from conference A need to schedule major powers from conferences B,C,etc.
This isn’t radical thinking, it’s more or less how basketball works (and most American sports for that matter). Most of the conferences play double round robins, and major powers frequently play each other either by home and home or through early season tournaments. They still suffer from all of the short comings of a single elimination tournament, the worst possible tournament structure, but they generate much better data to input into that structure. Because of that their arguments are more about fairness of selection rather than eliminating teams in the process whom had a decent chance of actually winning those 6 games in a row.