There’s a neat thread over on the BW forums about reasons to use social mechanics instead of “roleplaying it out” (AKA not using mechanics to resolve social conflicts).
Going to the bigger issues of design and play, functional roleplaying requires all participants to have the chance for meaningful input*.
One thing mechanics do well, is set up a system for “turn-taking”, or generally giving everyone a chance to make that input. A second valuable thing, is that mechanics also put a cap on the conflict.
Just as much as you could sit there for 4 hours describing a never-ending sword fight without mechanics putting a limit to it, you can sit for 4 hours arguing whether to take the Dwarven gold for yourselves or give it to the Dwarven people to rebuild their home. In both cases, odds are pretty good that you’re not going to be able to make that an entertaining 4 hours.
In stories, most social conflicts are resolved rather quickly- the highlighted stances and points are made, and then you move on. One benefit to mechanics is that by putting a cap on it, players have to choose their most relevant points, and not drag it out into a “last word”/endurance argument.
Groups can and do develop social contracts which fit this function, but you do see problems when they try to introduce new players (who then have to learn the implicit rules), or if a players steps out of the bounds, it becomes a big negotiation struggle, as the source of the problem might be completely misidentified. (“My Guy Syndrome” where fidelity to your character = not fun for everyone else involved is a good example.)
Not every game needs social mechanics, though I think in any game where you expect entertaining and meaningful social conflicts to happen, it’s probably a must.
(*Ditto with designing discussion spaces, with the added requirement that you have to develop means to filter the participants from the non-participants, otherwise there is no space for the discussion to happen.)