Figure 1

Giant Component and deanonymization. (a) Illustration of an anonymous network where the observed layer (blue) represents the network of interactions between anonymous actors; while the deanonmyizable layer (pink) is the subset of interactions where individuals have knowledge about their neighbors that an investigator could exploit. Theoretically, based on the observed network an investigator who knows the identity of node A could use this information to obtain the identity of node E by first determining the identity of one intermediate node (D). However, in practice, based on the deanonymizable layer the investigator would need two intermediate nodes (F and D) to identify node E since D did not reveal their identity to A in their interaction. (b) Demonstration of the process through which deanonymization could occur. In the initial network (i) no users are known. In (ii) after investigators successfully identify node A, this node is marked green meaning it is known and investigators can then use information from node A to try and traverse the network. Initially all links are unknown and thus marked red. Green links represent links where users have provided information about the neighbor at the other end and green nodes are identified nodes. In (iii) node B is successfully identified through information obtained via a link from node A. Next, in (iv) node B provides information about node C, who is also shown as identified in subplot (v). Finally, node C provides information leading to the identification of node D in (vi)