A couple of points I want to throw in the mix...
About the much discussed and lampooned presets, I actually like most of the presets.
Keep in mind the Eric Persing and his team did not always to try recreate real instruments, for instance, Glass Voices 1 and 2, which are perhaps some of the best pads ever, and used the D-50's samples which some people here are saying are worthless. Same thing with STAR-TREK voices, Marshy Zone, Intruder FX, DND, etc. The now cliche patch, Living Calliope, was obviously not trying to exactly mimic the exact sound of a panpipe with that patch, Eric Persing could have done a better job (and he did). However, Living Calliope is a great patch because it has that uniqueness of having the panpipe "feel", but not really the sound. It sounds more mechanical due to the spetrum wave used.
This brings me to my second point that I have made here before but feel I must reiterate and expand.
The key to the sound of the D-50 is how it tries to emulate something else, but misses the mark
Do people here dislike the Mellotron because it doesn't sound realistic? Do people dislike the TB-303 because it doesn't sound like a bass guitar? The TR-303/808 because they don't sound like real drums? This can go back even further in time. Do people dislike the upright piano because it doesn't sound like a full grand?
Possible problems may arise when you are extremely fond of the technological limitations of the D(5)50 (aliasing, 8bit, lofi, shortloops). That would be an interesting thread on its own tho. The most interesting aspect of such a discussion would be whether Roland would've made the machine with these technological artifacts when there wouldn't be a reason to do so. Just imagine the D50 being made 10 years later.
Well I assume you are talking about in a universe where they never made the D-50 in the first place, right? Because if not, we have already seen the result of that thought experiment.
But if not, you have to make these assumptions;
1) Roland wants to make a profit.
2) A large portion of musicians are not interested in totally new sounds, or creating sounds (this has been shown to be true over the last 20 years).
3) A smaller portion of musicians are interested in totally new sounds, and or creating their own.
4) It is more profitable for Roland to appeal to the first group, for reasons of demand and efficiancy.
In substractive synths, this means few new things ever getting past the same osc types, filter types, we have seen for 30 years. In ROMplers, this means an ever-growing amount of presets, most of which being devoted to emulation. So if the D-50 had been created right now, we would have something like the XV-5080. A large amount of high quality emulations of both synths (but not D-50) and acoustic, and perhaps a little amount of some original sounds thrown in.
However, this isn't really the pertinent question as the D-50 DID get produced, missed the mark, and that "charm" pleased the 2nd group. It was made popular by that group, and is now fair game to be emulated and absorbed into that general process all over again. This is what happened with the TB-303, D-50, and others.
So to close, it is basically a Darwinian process, with the most successful ideas passing along, and the few mutations that arise will be selected. Which ever mutation is most fit to be selected will replicate itself and gets added to the pool of ever growing successful ideas.
Both Don Solaris and CS_TBL are correct.