Regulating access to adult content (with privacy preservation)

In the physical world we have well-established mechanisms for keeping children out of adult-only areas. In the virtual world this is generally replaced by self declaration. Some service providers resort to using heavyweight identification mechanisms, judging adulthood as a side effect thereof. Collection of identification data arguably constitutes an unwarranted privacy invasion in this context, if carried out merely to perform adulthood estimation. This paper presents a mechanism that exploits the adult's more extensive exposure to public media, relying on the likelihood that they will be able to recall details if cued by a carefully chosen picture. We conducted an online study to gauge the viability of this scheme. With our prototype we were able to predict that the user was a child 99% of the time. Unfortunately the scheme also misclassified too many adults. We discuss our results and suggest directions for future research.