Reverse-engineering censorship in China: Randomized experimentation and participant observation
Stephen Williams stashed this in Censorship
Cool details with some surprising characteristics. Wonder how this evolved over time. Surprisingly fair, in a still twisted way. Net: the government is afraid of crowds and use criticism to weed out poor politicians. The overhead must be staggering. This is the original paper: http://gking.harvard.edu/publications/randomized-experimental-study-censorship-china
I agree with you -- this has gotta cost way more than it's worth, and it seems unsustainable.