I spent way too much time reading this thread, so I might as well make a few comments:
1. Players shouldn't have a CA boost because they played well. They played well with the CA level they have already. Similarly if a player suddenly plays a bad season, they shouldn't have their CA lowered, as this will create a downward spiral, and they will play worse next season. They were perfectly capable of playing bad with their current CA. If players get a boost to attributes after a data update because they have been playing well, it is only because the human researcher got it wrong in the first place. Guessing how much a player improved last season or will improve next season is really difficult for researchers.
2. Players also learn from their mistakes. Sometimes they train harder after a bad game. Therefore a bad performance shouldn't stop them from learning.
3. Players spend so much more time in training than they do in matches. So they practice technical attributes (long shots, free kick taking etc.) 100 times more in training, than they do in game. Similarly with physical attributes. So match time shouldn't improve these much. I do however believe, that games improve some of the mental attributes in a way, that training cannot. Since the CA is tied between these categories (this is the best model we have, since it allows researchers to model players to play like they do in real life), match experience also matters in the development model today. Perhaps there could be separate CA for these categories for new-gens.
4. One reason a lot of players haven't experienced development after the age of 24, is probably because the player manager is so good at developing them before that age, and are mostly buying youth instead of older players for this reason. This is myself included. I usually buy young and sell early. But I have seen improvements in reserves older than 24, who I have suddenly had to rely on. In my experience, really high PA players keep improving through, unless they are as good as Mbappe when they are young. If you want to experience more Jamie Vardy's and Luca Tony's, you need to be a small club that takes a gamble on an over-performing lower league player, and give him playing time, even when he loses form
Is SI's model perfect? Probably not, but it is pretty good, and I'm sure they continually improve it and listen to feedback from us.