MDI Faculty Spotlight: Micah Sherr

Written by Carrie McDonald, MDI Journalism Intern 

Professor Micah Sherr’s research is contributing to an “arms race” against censoring states to work toward a future where everyone has access to a free and open internet 

As the Callahan Family Professor of Computer Science and a Faculty Affiliate of the Massive Data Institute (MDI), Sherr’s projects focus on computer security, privacy, and internet measurement. His most recent research focuses specifically on internet censorship, building new methods to counteract nation states that seek to curtail their citizens’ free access to information online.

Sherr is the co-director of Georgetown’s SecLab, an academic research facility that investigates problems involving information assurance and computer and network security, and was the Principal Investigator of the Reliable Anonymous Communication Evading Censors And Repressors (RACECAR) project. A collaboration between Georgetown University, the U.S. Naval Research Laboratory, and the Tor Project, the RACECAR project aimed to develop unobservable and compromise-resistant obfuscation channels for censorship-resistant communication.

“There are a lot of people who are disenfranchised who don’t have access to good sources of information because their government cuts them off,” Sherr said. “Having an open internet affects international relations, it affects perceptions of the United States and other countries, and in general, from a philosophical standpoint that I think aligns with Georgetown’s core mission, having free access to information is required for society to progress.”

The problem of how to encrypt and achieve confidentiality of information has been around for thousands of years, according to Sherr. 

“Even generals in ancient times needed to communicate with their soldiers in ways that wouldn’t be understood by their enemies if the message was intercepted,” Sherr said. 

However, researchers like Sherr today face a tougher challenge of figuring out how to communicate confidential information without alerting a censor that there is any activity, even when a censoring state controls a network and closely tracks its traffic. 

“It’s very much a constant arms race between the free internet advocates who are developing these censorship-resistant technologies and the censors who are trying to detect and quash their use,” Sherr said. “This back and forth is essentially a game of Whac-A-Mole where you come up with a new way of getting around the censorship system and then you can see them respond to that.”

However, one of Sherr’s recent results reveals that current approaches to resisting censorship are “more brittle than previously believed” and contain “fundamental weakness.” Currently,  the general approach used by many censorship-resistant technologies involves an intermediary (third party) in a non-censored country that volunteers to receive and redirect traffic to someone in a censored country. 

“Even a censor who is only partially able to detect when this happens can over time accrue evidence that this [communication] is going on,” Sherr said. “If the censor suspects that a server in a free country is relaying traffic, it takes just a very short amount of time to detect that it’s doing this and then permanently block it.” 

An investment in more dynamic approaches that do not rely on fixed or stationary network positions is needed to surmount this challenge, Sherr said. This method could involve active networking, known as polymorphism.

“Think of [these methods] as non-stationary servers that go from one IP address to another, or hop around, so that this dynamism is too fast for a censor to block,” Sherr explained. 

Looking forward, Sherr believes that the most exciting opportunity in the field is the potential to use machine learning techniques to generate new types of censorship-resistant protocols. However, the potential for censors to use the same machine learning techniques to detect the presence of censorship-resistance technologies poses the most significant challenge ahead.

This paradox may intensify the “arms race” between censors and resisters. Nevertheless, MDI equips Sherr with “a way to gather large datasets and collaborate across disciplines on those datasets as a way to do groundbreaking science.”

“Studying things myopically as just a computer scientist is unlikely to lead to breakthroughs that really better society, whereas taking an interdisciplinary approach and tackling a data-driven problem from multiple perspectives, is much more likely to yield results that are far better,” Sherr said. “And the opportunity to engage in that is what attracted me to MDI.”