How Cryptography Lets Down Marginalized Communities

One of the perennial highlights of the International Association for Cryptologic Research’s Crypto conference is the “invited talk.” For an hour each year, a prominent scholar shares a big idea or new perspective on the protocols, algorithms, and math problems that underlie cutting-edge encryption. It’s usually a deeply technical bacchanal, but this year was not. Prolific academic cryptographer Seny Kamara of Brown University had something other than formulas and theorems on his mind.

“So an actual question then is OK, well, what am I doing here, right?” Kamara asked the livestream attendees. “Why am I giving a talk at Crypto if I’m not talking about technical things? And, you know, basically I’m here because Ahmaud Arbery was killed in February, because Breonna Taylor was killed by a police officer in March, and because George Floyd was also killed by police officers in May.”

The talk, dubbed Crypto for the People and given on August 19, examined the question of who really benefits from encryption technologies and advances in cryptographic research. It sounded a call to reexamine research priorities that today largely serve the interests of governments and corporations instead of marginalized people, be they racial minorities, immigrants, women, the LGBTQ community, or others. As an immigrant and Black American—and one of the few Black academic cryptographers in the world—Kamara pointed out that even the open source community and movements like the cypherpunks largely don’t directly work to address these needs. They are focused on taking power from corporations and developing technologies to defend people from mass government surveillance and digital intrusion, but generally not on developing encryption technologies and new areas of abstract theory that are specifically motivated by the needs of underserved communities.

“As long as I’ve been studying and working in cryptography and computer science, about 20 years now, it was always very clear to me that my own work and other people’s work was disconnected from my life experiences,” Kamara tells WIRED. “I believed it could have an impact on people’s privacy as a whole, but I didn’t think I would have cared about any of it when I was 13 or 15 and growing up in New York City. And that disconnect always bothered me.”

So much of cryptographic research is abstract and mathematical—divorced from real-world conditions—that it can be easy to simply let all lines of inquiry exist only in that theoretical space. And Kamara argues that even when encryption technologies are brought to underserved communities, they arrive retrofitted from other research projects, rather than conceived based on the needs of the vulnerable and the specific threats they face.

“As academics working on policy questions, we motivate our work in grant applications and so on by arguing that it benefits the people in some way,” says Abdoulaye Ndiaye, a macroeconomics researcher at New York University who discovered Kamara’s Crypto talk on Twitter. “However, the consumers of our research are other academics, government institutions, and, in some fields, businesses. There is this underlying assumption that these entities will implement the research and it will trickle down to the underserved people. Dr. Kamara highlighted that in cryptography the incentives of the government and the business are not necessarily aligned with underserved people, the missing link in this trickle down.”

Encryption technologies do provide protection to vulnerable groups around the world like political dissidents, activists, and journalists. Kamara’s talk made the case, though, that purpose-built cryptography could accomplish so much more.

In his own research at Brown, for example, Kamara and his colleagues have done work motivated by law enforcement databases in the United States that track alleged criminals like possible gang members. In a 2015 audit of a California state platform called CalGang, for example, 42 people entered in the database were under the age of 1 year old. In a sample of 100 entries from the database, 13 of the people represented should not have been in the database at all, and 131 of the 563 evidence points used against the 100 people were not supported.

Leave a Reply