On Lesswrong, the readers have a distaste for mindless content, so it doesn't proliferate, but all this means is that the "passing" threshold is higher. So you might (just as an example) still end up with content which echoes things that everyone already agrees with - that's not obviously unsubstantial in a way that would trigger down-votes but it is still not particularly valuable while still being easily processed and agreeable.

At some point, shouldn't content like the latter be identified as either applause lights or guessing the teacher's password? And, theoretically, be documented better in the wiki than the original posts? To me it seems like migrating excellent content to the wiki would be a good way to prevent follow-up articles unless they address a specific portion of the wiki, in which case it can just be edited in with discussion. I haven't spent any time on the wiki, though, which suggests that either I am doing it wrong or that the wiki is not as high-quality as the posts yet.

(Note: In pointing out the shortcomings of the voting system, it should be noted that I haven't actually suggested a superior method. Short of peer review, I'm guessing a more nuanced voting system which goes beyond the binary ⇵ would be helpful.)

If I imagine a perfect rating oracle it would give ratings that ended up maximizing global utility. If it only had the existing karma to work with, it would have to balance karma as an incentive to readers and an incentive to authors so that the right posts would appear and be read by the right people to encourage further posts that increased global utility. It could do that with the existing integral karma ratings, but at the very least it seems like separate ratings for authors and content would be appropriate to direct readers to the best posts and also give authors incentive to write the best new posts. This suggests both separate karma awards for content and authorship as well as karmafied tags, for lack of a better word, that direct authors in the direction of their strengths and readers in the direction of their need. For example, a post might be karma-tagged "reader!new-rationalist 20", "author!new-rationalist 5" and "author!bayesian-statistics 50" for a good beginning article for aspiring rationalists written by an author who really should focus on more detailed statistics, given their skill in the subject as evidenced by the post.