We had some discussion here a few years ago about implementing some scoring method for comments — there were some proponents who thought it would be a useful way to get community input. I’ve always been dead-set against it. It turns out I have scholarly justification now.
Abi Sutherland discusses a psychology paper at Making Light, which examined the effect of up- and down-voting on large user communities at CNN, IGN, Breitbart (oops, there’s a dollop of poison in the database), and allkpop, a Korean entertainment site. Cheng, Danescu-Niculescu-Mizil, and Leskovec proposed to test a prediction of the operant conditioning model, that peer feedback would lead to a gradual improvement in the quality of posts. That’s not what they saw.
