Sorry I didn't incorporate this into the solutions page sooner, Luke. I didn't check this thread for solutions. I will add this now. (I made a cliff notes version of the suggestions if you're interested). I question, though, whether changing the karma numbers on the comments and posts in any way would have a significant influence on behavior or a significant influence on who joins and stays. Firstly, votes may reward and punish but they don't instruct very well - unless people are very similar, they won't have accurate assumptions about what they did wrong. I also question whether having a significant influence on behavior would prevent a new majority from forming because these are different problems. The current users who are the right type may be both motivated and able to change, but future users of the wrong type may not care or may be incapable of changing. They may set a new precedent where there are a lot of people doing unpopular things so new people are more likely to ignore popularity. The technique uses math and the author claims that "the tweaks work" but I didn't see anything specific about what the author means by that nor evidence that this is true. So this looks good because it is mathematical, but it's less direct than other options so I'm questioning whether it would work.

Luke, I am not sure there are mathematical approaches to social problems. For the same reason you can't solve Google's problem with just the PageRank algorithm.