Not everyone has the freedom to speak openly on the Internet. Many people turn to the use of tools that offer them hidden, protected, or obscured access to the Internet, opening up content that might otherwise be blocked or providing connectivity to blocked services. However, no one has identified a way to measure how well these “circumvention tools” work, when and why they might not work, and which tool to choose. Depending on which tool is selected and how it works, the simple act of attempting to use it could put someone at risk of being discovered.
Creating a common framework for testing circumvention tools is vital in supporting users to understand the risks, and rewards, of circumvention tools. In 2014-2015, with a small prototype grant from the Knight Foundation, OTI and the Open Internet Tools Project (OpenITP) began work on a common testing framework, dubbed the Network Interference Test Environment (NITE) project. This framework prototyped gathering data on where the tools work, where they do not, and the particular network conditions during successful or unsuccessful attempts to use them, both with or without user intervention. This project involved the prototyping of a set of automated, virtualized testing tools to detect and characterize Internet interference.
Although some resources exist about different circumvention tools, there is no information about when, why, and how they fail. Many of these tools are not tested systematically in the countries where they are meant to be used. While users of these products provide ad-hoc reports on their efficacy when used in locations around the world, and organizations such as APC, GISWatch, Freedom House, and others regularly compile information about the most dangerous countries in which to practice freedom of expression, this information becomes quickly outdated as the adversaries who engage in network censorship and interference continually work to thwart the technical interventions that circumvention tools provide.
Assessing the efficacy of circumvention tools in a particular environment upon request becomes an intensive process, one that requires quickly learning as much possible about the user’s technical network environment and the sociopolitical conditions in their location. Providing this service in challenging conditions requires acquiring network access in-country and running single-point or ad-hoc tests. Single-point tests rarely reveal trustworthy information about the network environment. Ad-hoc tests only provide isolated data points at specific moments in time. Because tests are both ad-hoc and manual, they are somewhat akin to anecdotes, and are not systematic enough to analyze in a rigorous way. Practitioners who do this work regularly often supplement these anecdotes with data points gleaned from Twitter and anti-censorship mailing lists. Again, this combined set of anecdotal data is not enough to help users make an informed decision when selecting a tool.
The NITE project and prototype provided an opportunity to analyze and understand the many challenges that face interference measurement projects, beginning with the fact that categories of censorship vary from country to country. As OTI wrapped up the prototype project and took a step back to assess the lessons learned, it is clear that developing such a testing framework remains critical to helping users assess the risks of using selecting a particular circumvention tool. For many users of these tools, the risk of exposure by using these tools remains, at best, a question mark.
Read the full report here.