I guess it's pretty natural that the top draftees will be 18 given how skills and training works in this game. I think the training system would need to be modified a bit to make 19-yo's (and why not even older then?) potential 1st picks. Anyway, picking late may be a good enough reason to scout 19-yo's. Especially those you know could be good. You know teams ahead of you will pick the best 18-yo talent in the first round. (Sure, theoretically one or two such players could be missed by others even in an active league, but to draft successfully you just can't rely on that.) So, if you pick late, why spend too much to find the best of the best during scouting? The teams who pick first will try to find them, too. Perhaps it is best to go for the okay ones to get something decent from the draft.
Here's some speculation:
There are basically just three types of players of some value that can fall through the cracks in the current system: top 19-yo prospects, high-potential 18-yo's with low rating, and players too short/tall for their suggested position (regardless of age, although 19-yo's have a much better chance of going unscouted). The third group is so-so, but I believe some managers will shun them. Expect every other valuable player to be scouted by most teams that invest in scouting. I believe savvy scouting can potentially give those picking late a better overall result over three rounds than some of the early picking teams going after the top talent -- especially in a fairly weak draft. That makes the draft interesting.
perhaps this could be solved by having a draft of 96 players created and that way 48 are selected by teams and 48 remain undrafted.
That's a huge change and that would be a long discussion (I think this may have been discussed already?). Such change has some merit and some problems. Anyway, I have understood the BB's don't want to create useless players so this probably won't happen. Even now many drafted players are almost instantly fired.