Calling my statement A, and yours B, both are true. A is probabilistically true (i.e., in most cases) iff the majority of people are idiots (and assuming a normal distribution of "impact someone can have on you"), B is 'strictly' true, well as far as strictly holds in social dynamics.

If you are a really good idiot oracle, i.e. if you're adept at quickly discerning someone's idiot attribute (or the lack thereof), you should follow B (which is a subset of A , "forall X ..." versus "forall X where P(X)"). If you're not, you should follow A, excepting special cases and, as mentioned, actually undesirable consequences (e.g. professional). For example, there are select people on LW whose approval I covet. So I'm not stringently following A (it's hard to follow one's own advice anyways), but I suppose I'm closer to A than to B, which gives me a better worst-case-scenario in terms of "power idiots exert over you".