I posited that the coaches want the best players to play because their jobs depend on it.
Your position is to throw skepticism at this with some half-thought examples. Joe Alt wasn't ready to be a starter when he first went in, he HAD to play and he seasoned quickly. I think most believe, retrospectively, he would've earned a starting spot as a RS freshman after he had enough time to develop. His development was redlined out of need and he grew into the position. Justin A. was probably the more consistently sound player last year. Botelho would've probably given us more splash plays and more gaps unaccounted for. This is what all of the coaches have been saying for 2+ years.
I'm rooting hard for Botelho, seems like a guy who has worked his tail off to get better and play under control but they are still talking about this 3 years after matriculation. That means he's still learning.
Of course there are players whose games elevate under the bright lights but there aren't players who miss their assignments who all of a sudden become assignment sound.
Retrospective analysis is always so easy, it's predictive analysis applied to the human element that's difficult. The coaches are taking in far more data points than we are. They are charting wins/losses/assignment correctness and then applying their own assessment and making a determination on who gives them the best chance to win. No one gets to learn on the job unless there are no better options.
How many busted assignments would you give up per sack? How many per QB pressure? Are they an even trade? Does one potentially have a bigger determination on the game? What about when the game is close? 3rd quarter? 4th quarter? These are the equations constantly being run through spring/summer/fall and the season.