I understand what you're saying qualitatively; I was trying to get at your quantitative estimates. The numbers will constrain your optimal strategy for extracting value from the site.

For example,if for every "good" post there are N "table-thumping" ones and N=20, it's difficult-but-possible to find the "good" stuff. If N=200, it's effectively impossible. If N=2, it's pretty easy.

Conversely, at N=2 it is perhaps worth trying to convince the 2/3 majority to behave differently (the way you seem to be doing, sort of), but at N=20 you probably do better to figure out ways to flag the "good" 5%, concentrate your attention there, and allow the "table-thumpers" to play around on the less-valuable periphery in the hopes that maybe we'll be inspired by your good example. (At N=200 you probably do better to create a different site where the top .5% of LW-contributions can be hosted.)

You're right, of course, that this is a very imprecise way of talking about it. Given that I'm just asking about your off-the-cuff judgments rather than the results of your actual measurements, that seemed appropriate.