You assume that online communities die because there are too many (sub-)mediocre people. I admit that this is a reasonable hypothesis that explains part of what can be observed, but it is a rather big leap to get to "a community will not fail as long as its populated by the elite". Even the elite has a finite attention span, after all, and any sufficiently large community is likely to splinter into sub-communities anyway. Either that, or people have to stop posting all but their best posts - and that does not seem likely.
I think the take-away point is that this system would provide a way to have (one or more) communities within the community. The various subcommunities would overlap and cross-pollinate, so to speak, but if there were a small community of serious thinkers, they'd generate interesting talk on their own. This doesn't exclude the larger community - in fact, it permits the existence of a larger community to continue, and occasionally to cross over into the inner community.
I'm not expressing this well, on reading back over it. But I think you're miscontruing the point. It's not that online communities die because there are too many sub-par people; it's that that many sub-par people dilute the original community and everybody throws up their hands and finds another place. This allows the original community to persist amid the din.
I was specifically responding to what I construed as "this will allow a critical mass of smart people to emerge/collaborate": due to people's finite attention span and inability to keep track of too many different people, there is a natural limit to the size of a community anyway, and it is possible that "elite" communities run into this bound before they achieve critical mass (which is rather nebulous anyway, but I interpret this as "genius happens".)
I'm really not trying to misconstrue your point, BTW; if it didn't interest me, I'd have stopped responding long ago, and I have better things to do than trolling.
