More than likely it's because EA is constantly making little tweaks that have huge and unforeseen (by them) ripple effects. They get feedback that progression is over the top and after a few years in dynasty, all the teams are A+ caliber. So they scale things back. Recruits come in rated a little lower in key categories, dropping their OVR rating of course, and the amounts they can increase are reduced. This tests out as fixing the issue. But they don't have multiple testers sit down and play 10 years in a dynasty to see the long-term effects. And ideas that sound good on paper (freshman have low awareness - as most new freshman would) end up playing out badly when combined with their other changes (if individual ratings increase only a little bit, low awareness as a freshman means low awareness forever).
From personal experience, I know the drill: they encounter a problem, they assign a developer to do something about it, and it gets done. But there's not a lot of flowthrough testing to see how it affects everything else. Especially with the economy like it is, and EA having some lean-ish times recently, they probably don't hire many testers and therefore they don't perform due diligence to see how every change they make will impact the rest of the program. Users are their testers. And issues don't get fixed (or broken in a different way) until the next iteration of the software.
If they really want to solve this thing, they probably need to stop guessing at the starting ratings for recruits and advancement potential for players, and program both things to work off of the mathematical mean. Use the starting roster as a baseline: you can calculate the mean ability for each class (freshman, sophomore, etc.) and caliber of player from that. And then make sure that incoming recruits roughly match the existing mean, and that advancement points are doled out such that each year the classes conform to previous averages. So standard advancement isn't an arbitrary number like 2 or 3, it's just whatever makes the new seniors resemble last year's seniors, and on down the line. And awareness for incoming recruits isn't "50," it's the average awareness for default roster freshmen. Until they do something like this, or get really lucky with their existing model, long-term dynasty players won't be satisfied.
Luckily for me, I seldom play a dynasty past 4 years anyway.