The audio on the otherwise shaky body camera footage is unusually clear. As police officers search a handcuffed man who moments before had fired a shot inside a pizza parlor, an officer asks him why he was there. The man says to investigate a pedophile ring. Incredulous, the officer asks again. Another officer chimes in, “Pizzagate. He’s talking about Pizzagate.”
In that brief, chilling interaction in 2016, it becomes clear that conspiracy theories, long relegated to the fringes of society, had moved into the real world in a very dangerous way.
Conspiracy theories, which have the potential to cause significant harm, have found a welcome home on social media, where forums free from moderation allow like-minded individuals to converse. There they can develop their theories and propose actions to counteract the threats they “uncover.”
But how can you tell if an emerging narrative on social media is an unfounded conspiracy theory? It turns out that it’s possible to distinguish between conspiracy theories and true conspiracies by using machine learning tools to graph the elements and connections of a narrative. These tools could form the basis of an early warning system to alert authorities to online narratives that pose a threat in the real world.