For scholars of autocracies and conflict, the story which this paper presents may be more familiar. The idea that governments lag behind citizens and industry in controlling information is intuitive, and the particular relationship demonstrated by this data, where government must rely on agents to facilitate repression (imperfectly) parallels discussions of delegation to irregular forces, private militias, and the like. The particulars of information control (or not) highlight the difficulties a repressive state faces in controlling citizens, and its own forces, even when power may seem overwhelming. In some sense, the era of social media, a many-to-many information medium, is an ideal case for testing the limits of repressive capacity.
The discussion at CPRD focused on a few main areas of the paper, and in general, participants came away satisfied with quality of the work. First, the source of the data for this paper is an anonymous member of the censorship team at Sina Weibo, the firm under discussion in the paper. This kind of data, while exciting, can raise concerns about replication and veracity of the information. On balance, this isn't a huge concern, given the importance of the work and the steps Miller is taking to demonstrate the validity of the data. Second, the dynamics among firms in the social media environment was not explored in as much depth in this paper, but discussants agreed that it would be a useful avenue for future research, if feasible. Third, discussion focused on whether the assumption of the CCP's information control as complete and fully discretionary was legitemate. In fact, the paper argues soundly against this assumption, which pervades some work and is persistent in media and punditry around the matter.
The major contribution of this paper is twofold: First, this paper provides scholars with excellent evidence of the imperfect control which governments have over information. Even in a single-party autocracy, the ability of firms to opportunistically shirk their duties, allowing information to pass through censorship filters to serve their own needs, provides a more robust view of how censorship works in the era of digital communications. The paradigms of broadcast and print (one-to-many media) do not translate as well to social media (many-to-many), and Miller's paper shows exactly why. Second, this project introduces a theory of multi-principal-multi-agent relationships which may be transported to other areas of conflict studies, where power is contested and governments must rely on private agents (militias, paramilitaries, etc.) to help secure their goals. While the story here is particular to censorship in China (and censorship in one particular venue), one can easily imagine similar patterns of contested control and decentralized force in other settings