Jim responded, "There's one other wrinkle: you have to enjoy using the rules to resolve rules disputes."
See, the same thing applies to resolving rules disputes via discussion. I see a connection to Chris Lehrich's ideas about bricolage ("tinkering") as a mode of discourse within games, and the way that traditional gaming has tended to treat its own rules systems as objects for tinkering. (The most developed version of these ideas which I've seen is at Chris's "personal" LJ, here with followup here.
IOW, in much of traditional/classic gaming, both the rules and the in-game action are approached through a tinkering mindset which the participants enjoy, or have trained themselves to enjoy. On the other hand if you enjoy the mode of discourse offered by Polaris, it becomes natural to apply it recursively to the rules themselves.
What I think comes out of this is that the actual practice of playing a game (not the rules text per se, but the procedures and behaviors engaged in by the participants) have the effect of checking each other and building trust--assuming the game is successful. There may be a connection here to signaling (signalling) games, in that certain behaviors can be taken as proxies for the statement of one's honest intentions. (I wouldn't push that connection, though, without a better understanding of signaling games and their applications.)
And taking this a step further into speculation-land, we might look at some forms of dysfunction as a breakdown in signaling: we're playing a game that, to be enjoyed properly, requires oblique expression of our interests. But if either of us is inept at sending/receiving/acting on social signals, what happens? (Cf: a bad date.)