From my biased and limited observation of the rankings, the "accuracy" of the CFP rankings drops once you hit 15, probably because the CFP realizes that the rankings are intended to determine the playoff spots (the most important thing) and NY6. I would bet 65% of the time spent discussing rankings is focused on the playoff spots. 30% is spent discussing #5-12 to determine NY6 spots (although, I'm sure they're aware, really their votes for rankings doesn't matter as much because there are guaranteed spots for P5 champs and the highest rated G5).

After that it seems like they just put the same teams as the weeks before.

One of the major pluses of the AP poll is that you get a large voter pool, and while you get some morons (like people who were putting Utah in the top 10) who have no clue what they're doing or who are working on a bias, you get enough people who actually base it off results beyond the W/L column that it averages out and IMO makes a pretty accurate result.

Unfortunately, I think that this is just a symptom of the biggest problem with the CFP - any team that isn't one of the top four (maybe 8, because they have a chance to make it in the top four) doesn't matter.