The Internet will shape the history of the 21st century. Governments around the world realize this, which is why, out of 40 countries studied by the Open Net Initiative, Internet censorship increased from 3 of 40 in 2002 to 26 of 40 in 2007. The U.S. must set effective policy regarding this, but the inherently computational issues lead to a great deal of uncertainty. We are studying the dynamics of modern Internet censorship implementations to understand how their low-level mechanisms (for example, at the level of individual Internet packets) affects the application-level behavior of censorship that web users actually experience.
Internet censorship implementations are inherently computational, thus whether or not content is available to a user can depend on key issues such as routing and protocol dynamics. Current characterizations of censorship are incomplete, not founded on quantitative measurements, and cite experiments that cannot be reproduced. We are addressing this by laying a solid computational foundation for Internet censorship research. Central to this effort is Internet measurement of existing censorship implementations to understand and compare their dynamics at the packet- and application-level. This will impact both policy development and the future design of computer networks to address Internet censorship, a critical issue that directly harms U.S. interests by hindering the ability of human rights organizations and U.S. government agencies to disseminate information. Understanding the connections between Internet censorship and computational issues will also enable the integration of this research throughout the computer science curriculum at UNM and other institutions.
