Can pseudonymity really guarantee privacy

One of the core challenges facing the Internet today is the problem of ensuring privacy for its users. It is believed that mechanisms such as anonymity and pseudonymity are essential building blocks in formulating solutions to address these challenges and considerable effort has been devoted towards realizing these primitives in practice. The focus of this effort, however, has mostly been on hiding explicit identity information (such as source addresses) by employing a combination of anonymizing proxies, cryptographic techniques to distribute trust among them and traffic shaping techniques to defeat traffic analysis. We claim that such approaches ignore a significant amount of identifying information about the source that leaks from the contents of web traffic itself. In this paper, we demonstrate the significance and value of such information by showing how techniques from linguistics and stylometry can use this information to compromise pseudonymity in several important settings. We discuss the severity of this problem and suggest possible countermeasures.