That's a good point regarding people knee-jerk upvoting too. I wonder if A/B testing to see whether there's a difference if weighting upvotes higher from people who comment more and probably taking comment length into account - and perhaps factoring in the upvotes of their comments to the weight of their upvotes.
People knee-jerk upvote just as much as they knee-jerk downvote and eliminating one half of the knee-jerks has nasty side effects.