Prevention and Reaction: Defending Privacy in the Web 2.0

User data can originate from three sources – the user, the user’s friends and acquaintances, and services with which the user interacts – and users may be harmed by the disclosure of data from any of these sources, some of which are not directly under her control. We argue that, for data disclosed by the user herself, tag-based privacy policy languages are flexible, are easy for users to understand, enable users to express their prefereences succinctly, offer good portability across service providers, and provide a strong foundation for creating sticky policies. We review our previous work developing and evaluating a tag-based privacy scheme. Our past results demonstrate that tag-based policies are easy to use, support short policies, and enable Web 2.0 services to offer automatic privacy policy inference, further reducing the policy maintenance burden on the user. We then consider several possible mechanisms for managing the impact of data published about one user by another. Even if it were possible to automatically determine the target user described by the data, social conventions, such as freedom of expression, preclude giving the target user absolute control over the information. Furthermore, privacy and data usage controls must advance the interests of the parties that implement those controls, and this significantly constrains the design space for effective solutions. We therefore propose reactive mechanisms that users and search engines can use to steer other users away from libelous content.