The VWO Gold Farming team won the award for Best Paper at the Game Behind the Game conference hosted by Rutgers University’s School of Communication and Information. Our paper was titled “Mapping Gold Farming Back to Offline Clandestine Organizations: Methodological, Theoretical, and Ethical Challenges.” The abstract is below:
“Clandestine”, “covert”, “dark” and “illicit” organizations are primarily characterized by the need to engage in coordination and collective action while also emphasizing secrecy and security (Ayling, 2009). However, empirical analyses of offline clandestine organizations’ structures have received scant attention because traditional data collection is difficult by design. Studies of clandestine organizations employ methods which censor their embeddedness within particular historical contexts and larger licit spheres of peripheral and legitimate actors. These studies rely on descriptive, single level methods. However, the explosion of behavioral data available in online databases has opened up new avenues of social research. To the extent that individuals in online worlds operate under similar social and psychological motivations and constraints as the offline world, it is possible to use generative models of clandestine networks from online virtual worlds to test and inform theories of clandestine networks in offline contexts (Williams, 2010). We use gold farmers in massively multiplayer online games (MMOGs) as a case to examine how clandestine organizations assemble and maintain their operations in the face of pressure to remain competitive and secret. We review our recent research findings employing methods in network analysis and machine learning to detect and identify gold farmers in a popular MMOG based on distinct structural motifs in trade exchanges, patterns of behavioral similarity, and appropriation of in-game affordances. Although these findings on virtual clandestine organizations comport with many existing theoretical predictions as well as observations from offline criminal behavior, we discuss how they fail to map from online to offline in other contexts. Finally, we discuss the ethical implications of attempting to develop abstract heuristics for identifying clandestine behavior in data rich contexts and conclude by identifying future directions for analytic and theoretical research.