i think shoe's argument here is pretty persuasive. we have nothing like the one and done freshman of real life, not from recruit gen (IQ primarily) not from EEs, etc, i don't think its generally good to pick and choose parts of today's real life underclassmen flow to NBA model to implement without taking into consideration the whole. so it doesn't exactly follow to base the examples on today's college ball.
that said, i think the idea is that in general, from a game design standpoint, it is beneficial to have random events for users to deal with, like a 50/50 on a recruit or a potential roll that is a fairly known quantity, etc - while it is detrimental to have events that are both highly likely/unlikely and severe - the combination is what is killer. such events are very hard to plan around (like severe injuries / player death, a 96% EE roll, a 4% EE roll, and so forth). that is definitely opinion to some extent, but that's my opinion and in general, the game designers made some efforts to avoid those things (like the unrealistically low severe injury rate), knowing the impact on user frustration (high).
i don't think there's a great 'because of reality...' argument for the change, but i think there's a pretty reasonable 'for game design purposes...' argument here, about those 90% juniors returning, and on those low-board juniors who may only get a roll 10% of the time, yielding a 3% real chance of leaving, that sort of thing... with both high/low odds involvement and high severity, i definitely get how this leads to user frustration.