An early warning system for the internet
My new feature story takes us down the rabbit hole
On a 2018 United flight to South Florida, Adam Sohn, a broad-shouldered businessman, found himself sitting next to a lanky man transfixed to his laptop. Sohn’s eyes popped when he noticed what was on his neighbor’s screen — intricate word clouds with racist and antisemitic language. Words most people never think, much less leave visible on their computer.
Sohn was an open-minded guy. A former Wall Street trader who left that life after 9/11 and had since pursued an eclectic career working for the AARP, Jeb Bush, and Charles Koch, Sohn wasn’t afraid of going against the grain to do work he thought could make a difference. And in this case, he was determined to figure out if his airplane neighbor was a terrorist. “You look like a pretty smart guy,” Sohn recalls saying to the stranger. “But are you gonna crash the plane? Or maybe save the world?”
When Joel Finkelstein told him what he was up to, Sohn was floored. Finkelstein was a neuroscientist whose research had led him to study how hate speech spreads on the internet. By tracking the proliferation of racial slurs and memes on far-right platforms such as 4chan and Gab, Finkelstein had discovered that hate speech activity spiked on social media before hate crimes and other attacks occurred, allowing him to anticipate events such as the 2017 Charlottesville “Unite the Right” rally.
As Sohn heard this, he knew he and Finkelstein were going to be working together. A few months later, in May 2018, Sohn and Finkelstein co-founded the Network Contagion Research Institute (NCRI), a nonprofit research group based in Princeton, with Sohn its CEO and Finkelstein its chief scientist. Since then, the group has released white papers that have had an outsize impact on U.S. policy. A month before the Jan. 6 attack on the U.S. Capitol, NCRI’s December 2020 report on QAnon examined the conspiracy movement’s role in amplifying the false narrative that Dominion voting machines had rigged the 2020 presidential election in President Joe Biden’s favor. And when the dust settled on the Capitol riot, NCRI’s follow-up research informed the ensuing House investigation.
NCRI’s research interests range widely, from prompting the Apple app and Google Play stores to take down misbegotten dating apps for teens that served as the playground for Nigerian “sextortionists,” to understanding the extent of social media posts supporting Luigi Mangione, charged with the December 2024 murder of United Healthcare CEO Brian Thompson. Perhaps most significantly, NCRI’s work was invoked by Congressional leaders in the passage of the TikTok ban bill, entailing a forced sale or ban of the app that was affirmed by the Supreme Court and postponed three times by President Donald Trump — shortly after he took office in January, again in April, and then for 90 days on June 19.
NCRI is not without controversy. Some data researchers have criticized its methods and questioned why most of NCRI’s work is not peer reviewed, all while the landscape of social media data research has shifted seismically in the past few years.
How did a neuroscientist studying addiction and mind control in mice turn to researching how hate and misinformation metastasize across the internet? Despite Finkelstein’s relatively recent transition to the field, the connection between the two subjects is clearer than one might expect. For Finkelstein, it all started with puppets.
To find out what happened next, read my latest feature for Princeton Alumni Weekly, “Where Hate Hides.”
If you read anything I’ve written this year, it should be this—one of the most sophisticated pieces of reporting I’ve ever done.