Empowering bystanders to facilitate Internet censorship measurement and circumvention

Free and open exchange of information on the Internet is at risk: more than 60 countries practice some form of Internet censorship, and both the number of countries practicing censorship and the proportion of Internet users who are subject to it are on the rise. Understanding and mitigating these threats to Internet freedom is a continuous technological arms race with many of the most influential governments and corporations. By its very nature, Internet censorship varies drastically from region to region, which has impeded nearly all efforts to observe and fight it on a global scale. Researchers and developers in one country may find it very difficult to study censorship in another; this is particularly true for those in North America and Europe attempting to study notoriously pervasive censorship in Asia and the Middle East. This dissertation develops techniques and systems that empower users in one country, or bystanders, to assist in the measurement and circumvention of Internet censorship in another. Our work builds from the observation that there are people everywhere who are willing to help us if only they knew how. First, we develop Encore, which allows webmasters to help study Web censorship by collecting measurements from their sites' visitors. Encore leverages weaknesses in cross-origin security policy to collect measurements from a far more diverse set of vantage points than previously possible. Second, we build Collage, a technique that uses the pervasiveness and scalability of user-generated content to disseminate censored content. Collage's novel communication model is robust against censorship that is significantly more powerful than governments use today. Together, Encore and Collage help people everywhere study and circumvent Internet censorship.Ph.D