[storytime]
Long ago, I ran an 1E Exalted campaign. I captured a PC and interrogated his character, trying to force information that he had out of him. The player's reaction: shut up and wait for an opportunity to either be rescued or escape. I hurt his character, he said nothing. I ignored his character for days (ingame), he said nothing. I tried to charm the information out of him. He just spent a willpower to end the effect.
Right then I knew that a more complex social combat system was required and would help, no hinder role playing. I was much pleased with Exalted 2E's social combat system. It still needs work, but it is better than 1E.
[/storytime]
In my experience, a system that promotes role playing puts just as much emphasis on social interaction as it does combat. This mean that the method of convince someone that you're on their side is the same system (or nearly the same system) as swinging a dumb metal stick. It also means that there are roughly the same number of powers as there are for social interaction as there are swinging that dumb metal stick.
D&D/D20 (regardless of edition) is a great example of a system that does not promote role playing. Feats are the best place to start to exemplify this. Using the 3.5 SRD, I count 68 feats that directly affect combat and 7 feats that directly affect social interaction. (Note: I was increadibly liberal with my social count, so whether or not some should have been counted is debateable. Also I'm going to give both counts a +/-10%, allowing for different interpretations, mistakes, etc). I don't own 4E, but the sections I read in B&N lead me to believe that the dispairity is just as bad. The social feats I counted, almost all of them were +2 to two skills. There was no equivilent of power attack where the whole of the person's personality descends upon someone to convince them that you are right and they are wrong. There is no Proficiency with Reverse Psychology, no Stunning Argument, no Quick Whit, no ... You get the idea.
Then there is Good and Evil hardwired into the game. Well, on the good, neutral, evil scale, how do you define V who is doing a good thing (saving her family) for all the wrong reasons and went about it in an evil manner (making a deal with fiends). Or how about Belkar? When He saved Hinjo, he committed a good act. When he rescued Hailey and did not abandon her, he committed a good act. Infact, since getting the Mark of Justice, I cannot think of any truly evil act (save killing the kobold prophet) that he committed. But even then, killing isn't innately evil in D&D since that is what the game is about (killing things and taking their stuff). So even that 1 act isn't necessarily evil. So why is he still considered evil? I believe the BoVD said that actions do not matter, all that matters is what happens when someone is hit with a smite evil.
D&D, regardless of edition, has pretty much been freeform role playing (which is a nice way of saying that it does a slightly better job than monopoly at encouraging role playing). Freeform role playing means that anyone, regardless of Int/Wis/Cha score, can making an equally convincing argument and change someone's mind. There is no system for saying that someone with a 24 CHA is more likeable than someone with a 4 CHA (or that the prettier person has an easier time convincing someone of doing something for them). While the DM should make a judgement call and the player should act the way their ability scores say they should, the rules themselves do not support it. The real limit to a character's social abilities is the player, not the character.
