Kumar, Srijan, et al. “An army of me: Sockpuppets in online discussion communities.” Proceedings of the 26th International Conference on World Wide Web. International World Wide Web Conferences Steering Committee, 2017.
Summary
Kumar et al. focus this paper on the identification, characterization and prediction of sock puppets in online discussion communities. They define sockpuppets as extra account(s) created and managed by a single user for influencing or manipulating public opinion, igniting debates or vandalizing content (in case of Wikipedia). They characterize sockpuppets as pretenders vs non-pretenders and supporters vs. dissenters based on their linguistic traits, online activity and reply network structure.
Reflection
Although this study nicely situates itself in the body of work currently done in the domain of deception, I felt that it does not establish a very strong objective of being carried out.
It would also be interesting to see if a sock puppet account, or a pair of such is operated by more than one person interchangeably, which not only makes the concept of one puppet master imprecise, but also weakens the statistics obtained by the data with a single user in mind. The hypothesis of puppet masters leading double life reminded me of Facebook, where spouses access each other’s accounts without any problem, sometimes simply in order to peek into the content of ladies only groups, and even comment or react on different posts just for the sake of fun. Although very different from the topic under discussion, it poses the question of whether a study on online behavior of such individuals produce accurate results because of multiple users associated with a single account.
The authors have also used IP as a means to cluster different sock puppets, I was wondering if users logging in to the social platform using proxy servers would be easy to identify using the same study? What if the puppet master uses a both sock puppets and bots to steer the discussion? In such case, the detection system can be made more robust by incorporating mechanisms to not only to take linguistic traits and activity, but also consider the amount of customization in creating user profile and geographical metadata [1]. This will not only help detecting sockpuppets, but will also be able to identify bots from sockpuppets.
The authors also rightly point out that study of behavior or personality traits would add another dimension to this research. The reasons of having more than one identity online can go beyond sadism, and also be a product of sheer boredom or for the sake of bragging in front of friends. The puppetmaster can also create multiple identities to avenge a previous ban.
[1] Bessi, A., & Ferrara, E. (2016). Social bots distort the 2016 US Presidential election online discussion.
