In my experience, attempting to prop up the social interaction pillar with robust mechanical support counter-intuitively tends to decrease social interaction rather than increase it.
The most functional mechanical system for adjudicating RP is simply IMO some sort of fortune test, preferably one that generates a small degree of success but a simple pass/fail will do fine if you have some means of robustly adjudicating difficulty. Beyond that, no part of the game works better with the GM just working off rulings and gut feel for the situation than social challenges, because any system that realistically models the dynamics of interaction between people will be vastly more complex than a system that realistically models combat, and the simple mechanical churn will detract from RP.
I think the trick is not add too much to the rules for actual interaction. At most I'd add rules for multiple successes necessary to, say, convince an important NPC of something. That isn't adding rules, just a way of keeping score. It would still be up to the PCs to figure out what that NPCs motivations and objections might be. That does a couple of things. One, it provides a concrete handhold for measuring success, which I do think SI interaction needs in some form. Two, it opens up some doors for using other skills, like insight for example, to try and determine what levers are the right ones for that SI task. The only other thing I might add is to decouple the 1-1 skill to ability relationship, at least partially, to allow more characters to participate and succeed who don't necessarily have high CHA. My biggest concerns for actual SI tasks are overcoming the simple binary nature of checks for complex tasks, and keeping more characters active and involved in the whole process. RAW the SI pillar tends to be just one or two characters in the party doing everything, and that's boring for everyone else. Obviously the high CHA PCs will be better at direct CHA stuff, and they should be, but I'd like most characters to have the ability to participate in a meaningful way.
Ironically, the worst thing you can do to encourage good social interaction is provide a system of "moves" and mimic combat mechanics exactly, and yet in my experience that is exactly what games that are attempting to support a social pillar attempt to do. I call this the "banging it with a hammer" approach, in that well you only have one tool in your toolbox and so you assume that it applies equally well to solving all problems. One of the few systems I've seen that looks like it has mechanical support for social combat that might work is 'Dogs in the Vineyard' and that in part because it assumes a hierarchy of conflicts were escalating stakes to a more definitive conflict trumps lower conflicts - thus breaking the symmetry that more niave designers seem to design into the system.
In short, while I agree with the OP that this post shows just how bad the idea is, I'm not convinced that his minimalist approach isn't actually better than a more robust system.
Yeah, I'm with you on the moves thing. Treating SI like combat probably isn't the way to go. I think the ways that SI differs from combat are probably where I'd look to improve things. For example, there are lots of SI related things that exists outside a single encounter. Mostly this indexes things to do with ongoing relationships - so reputation, contacts, favours, leverage, friendship, etc. the DMG has some rules for things like this, but they are really basic, and they don't function as a rationalized system as a whole. Havings system that allows PCs to know, at a glance, where they stand in relationship to various factions and NPCs, and more importantly, what kind game result they can expect if they leverage those relationships. Not specifically, becaue that would require great bloody lists of specifics, but in a general way.
At the NPC level: I can rely on the Guild Merchant for a small favour, but the Duke's secretary owes me big and I could reasonably ask him for a much bigger favour. At the faction level: my reputation with the Guild of Merchants is only +2, so it might be hard to convince the Guild Council of X since I know the idea doesn't have a lot of support. How can increase my reputation there to have a better chance of getting what I want.
By providing some concrete handholds and measures it's a lot easier to make specific plans with specific goals. However, by keeping the abstraction out of the actual encounter level, you aren't distancing the players from what they are trying to do from a role playing perspective. The encounter level stuff doesn't change much, but the PCs can be very specific about longer range planning and the results of those plans are measurable and help the party and the narrative move forward. Keeping score and measuring success are important tools, so my general goal is to add that with the lightest rule set possible and with minimal abstraction to actual SI play within an encounter.
I also think it's a perfectly workable variant.
