The question is whether this is just a sample size issue, or indicative of a wider problem in the polls/committee?
To be honest, with proper seeding, a 1 seed should win the whole thing more than 25% of the time, but less than 100%. The question is how close to those two extremes is "correct", and that question boils down to just how much difference there is between the strength of #1 vs #2, #3, and #4. Still, a #1 seed in my opinion should be expected to win the whole thing maybe 30% of the time at minimum, or 40% of the time at maximum.
This year, for example, Vegas has LSU as a 14 point favorite over OU, and ESPN's FPI has LSU's win probability at 67.5%. If we make the assumption that LSU, as the higher seeded [and therefore supposedly stronger] team than OSU/Clemson would have a >50% win probability in that game, it would still require a 60% win probability in the final (over a team perceived significantly stronger than OU this year) to reach 40% win probability for the #1 seed.
So we've had 5 instances of the CFP, and the expected number of times for a #1 seed to win is somewhere between 1.5 and 2 times. That it's been 0 in only 5 instances might be simple due to small sample size.
Or it might not... You all know my thoughts on how to crown a champion in college football. Some might argue that if a committee made up of CFB experts, whose sole solitary job is to find the best four teams in the land (and seed them according to strength), and they can't pick the #1 team, how could pollsters be any better? Even worse, if the #4 team has a winning record and has won 2 out of 5 so far, might it be that they're actively seeding BADLY such that the consensus weakest team of the 4 is the second best performing seed, it suggests that maybe the "experts" don't know as much as we thought.
It's for that reason that I eschew the idea that only the "best teams" deserve to be in the BCS or CFP. Because we're notoriously bad at determining who the "best teams" are. So I fall back on the hybrid system of "most deserving" teams plus a few "at large" teams that are deemed worthy but didn't meet the objective criteria.
Go to an eight-team playoff. Either the 5 power conference champs plus 3 at-large, or the 5 power conference champs plus 1 highest ranked G5 conference champ, plus 2 at-large.
Because if even the committee can't get seeding right, how do we even know they're getting the best 4 teams right? ESPN FPI has five teams higher than OU's FPI this year. Are we really sure they're the fourth best team in the land?