"The information Sony Online Entertainment provided for the research project was scrubbed of all PII (Personally Identifiable Information) prior to being provided to the researchers. For example, no content of any player chat logs were shared with the researchers. The information shared consisted of data such as which in-game characters chatted with each other and the volume and frequency of the contacts. None of this information was connected to, or linked with, the real names or other PII, of any players. Basically, the researchers looked at the connections between players and how their online networks were built and used, not the content of any actual conversations that these players may have had with one another (that content was not made available to the researchers). Additionally, some EQ2 players voluntarily participated in a blind survey concerning their playing habits and demographic backgrounds. This information was anonymously gathered and shared between SOE and the researchers. Finally, corporations such as SOE and the researchers' universities have controls and safeguards in place to help to protect the privacy of individuals who voluntarily participate in surveys for research projects such as this one."
A group of researchers who went from game developer to game developer looking to acquire data for studying online social interaction
got more than they bargained for. Sony Online Entertainment keeps extensive server logs of everything that happens within Everquest 2
. When the researchers asked if there was anything they could look at, SOE was happy to share the entire EQ2
database — upwards of 60 TB — for their perusal. In addition to basic gender and age queries — who interacted with whom, and when — the scientists are also trying to find ways to track more subjective characteristics, such as performance, trust, and expertise. "To get estimates of them, the team is experimenting with trying to track physical proximity and direct interactions, such as when characters share experience from an in-game victory. To give a concrete example of the data's utility, Srivastava described how he could explore the phenomenon of customer churn
, something that's significant for any sort of subscription-based service, like cell phones or cable TV. With the full dataset, the team can now track how individual customers dropping out of the game influenced others who they typically played or interacted with. Using this data, the spreading rate and influence factor could then be calculated, providing hard measures to work with."
Update: 2/18 at 21:04
: Sony contacted us to set the record straight about the shared information. All information that could identify players was removed from the data given to the researchers. Chat logs were not shared at all. Read on for SOE's full statement.