I have significantly decreased my participation on LW discussions recently, partly for reasons unrelated to whatever is going on here, but I have few issues with the present state of this site and perhaps they are relevant:

• LW seems to be slowly becoming self-obsessed. "How do we get better contrarians?" "What should be our debate policies?" "Should discussing politics be banned on LW?" "Is LW a phyg?" "Shouldn't LW become more of a phyg?" Damn. I am not interested in endless meta-debates about community building. Meta debates could be fine, but only if they are rare - else I feel I am losing purposes. Object-level topics should form an overwhelming majority both in the main section and in the discussion.
• Too narrow set of topics. Somewhat ironically the explicitly forbidden politics is debated quite frequently, but many potentially interesting areas of inquiry are left out completely. You post a question about calculus in the discussion section and get downvoted, since it is "off topic" - ask on MathOverflow. A question about biology? Downvoted, if it is not an ev-psych speculation. Physics? Downvoted, even if it is of the most popular QM-interpretational sort. A puzzle? Downvoted. But there is only so much one can say about AI and ethics and Bayesian epistemology and self-improvement on a level accessible to general internet audience. When I discovered Overcoming Bias (whose half later evolved into LW), it was overflowing with revolutionary and inspiring (from my point of view) ideas. Now I feel saturated as majority of new articles seem to be devoid of new insights (again from my point of view).

If you are afraid that LW could devolve into a dogmatic narrow community without enough contrarians to maintain high level of epistemic hygiene, don't try to spawn new contrarians by methods of social engineering. Instead try to encourage debates on diverse set of topics, mainly those which haven't been addressed by 246 LW articles already. If there is no consensus, people will disagree naturally.