Spoiler
Show
In a more generic game like, say, GURPS, characters are going to be a heck of a lot more varied than D&D characters, since (A) the abilities are much more granular, (B) more game foci are supported, and (C) point-based systems allow both more specialized and more generalized characters than level-based systems.
By point A, I mean both that abilities aren't interconnected like D&D's level-based stats (so you can improve them individually with independent floors and ceilings), and that GURPS has narrow skills like "good with spears" where D&D has wide skills like BAB and so forth. By point B, I mean that in terms of important given to various mechanics D&D is essentially combat first, exploration and lateral thinking second, talking third, and everything else fourth, whereas a generic system might (and probably does) give as much or more attention to the noncombat stuff as to the combat stuff. By point C, I mean that while D&D rewards specialization, it still requires minimum levels of competence in various areas by virtue of having level-based stats; in GURPS, it's possible both to sink all your points into one thing to be amazing at that and suck at everything else and to spread your points out so broadly that you can't really do anything at all.
So, comparing that to D&D, we see that everyone shares some common stats (BAB, saves, skill maxima, etc.) that go up by level, combat stuff is more varied and defined than noncombat stuff, and everyone needs to have their niche. If combat and social skills both worked the same way, either both are equally simple (roll BAB or Diplomacy, hit, apply effect, repeat) in which case the party face is bored because he rolls Diplomacy rarely and the party fighter rolls attack every round, or both are equally complicated (lots of combat maneuvers, lots of social skills, lots of fiddly modifiers) in which case you implicitly (and explicitly, if you buy combat and social stuff with the same resource) place equal importance on both systems when that isn't (supposed to be) the case, and spend more time on social stuff and less time on combat stuff than is desired.
If you instead separate out the combat stuff and the way it works (one auto-improving combat stat, ablative defenses, at-will default maneuvers) from the social stuff and the way it works (multiple selectively-improving social stats, graduated binary defenses, per-encounter skill tricks) you can give the two different levels of granularity, speed of resolution, importance of focus, and so forth. You can then independently tune the different subsystems to get what you want out of them.
This also helps from a metagame marketing perspective: D&D has plenty of players who want complex mechanics for combat, the bare minimum for climbing and running and similar, and none whatsoever for anything else; they want fiddly combat, but no mechanics to get in the way of roleplaying. Whether you agree or disagree with their viewpoint, it's been fairly well established that some people won't buy a certain edition because they think codified noncombat mechanics turns a roleplaying game into Diablo, some people won't buy a certain edition because they think the lack of codified noncombat mechanics turns a roleplaying game into an MMO, some people think too much mechanical noncombat focus makes the game overly restricting rollplaying, and some people think to little mechanical noncombat focus makes the game into purely hack-n-slash.
While a system like GURPS actually makes it easier for the people who don't want noncombat mechanics in the game (there are no "classes" that trade combat skill for noncombat skill, ignoring the noncombat stuff in your game is equivalent to no characters taking noncombat stuff and the game already accounts for the latter, and so forth), if seeing noncombat mechanics mixed in with the combat mechanics at all will turn them off from the game then the tweakability of noncombat stuff doesn't matter.
Now, in D&D having separate subsystems for noncombat stuff hasn't worked out as well as one could hope: The noncombat stuff was mostly an afterthought in AD&D, the 3e devs ignored any noncombat stuff that wasn't magic, and skill challenges/martial practices/rituals/etc. in 4e are either fairly flawed or missing things that players want, but none of that means the concept is flawed. On the contrary, the popularity of the late-3e subsystems shows that having different mechanics even within the combat sphere is appreciated, and all of the homebrew out there to expand and tweak diplomacy/stealth/crafting/etc. and the varying complexity thereof shows that people have differing goals for the noncombat parts of their games but they value it enough to make it deeper and more interesting.