I'm trying to prove this hypothesis wrong :) Obviously it's a big experiment.

My belief is that with the right incentive system, and right messaging & social proof, you can fix the problem. We're not there yet, but we are seeing it work on a very small scale.