2020 was a fruitful year for conspiracy theories: QAnon gained followers, COVID-19 misinformation proliferated in viral YouTube videos, and in November, President Trump helped proliferate the entirely false narrative that the election he’d lost was, in fact, stolen.
The details holding these falsehoods together get complicated quickly. But according to a group of researchers at UCLA and the University of California, Berkeley, even the most convoluted of conspiracy theories has a distinct structure. That’s different from real-life scandals, which tend to unravel as new evidence emerges—take former New Jersey Governor Chris Christie’s ‘Bridgegate’ scandal, a completely verified event in which several of the governor’s staff and appointees colluded to close toll bridge lanes during morning rush hour, intentionally clogging traffic to the town of Fort Lee, New Jersey.
The researchers wrote in the journal PLOS One in June that applying machine learning tools to conspiracy theories reveal them to be less complex than things that actually happen. Often, just a few important details are erroneously combined that don’t necessarily line up—Pizzagate, for example, combines narratives about Democratic party politics, a pizza restaurant in Washington, D.C., sex trafficking and Wikileaks. Pizzagate has been widely debunked, including by Washington, D.C. police. Meanwhile Bridgegate, an actual conspiracy in which multiple people were arrested, stayed firmly in the realm of New Jersey politics.
Ira talks to UC Berkeley’s Tim Tangherlini, a co-author on the research, about how these analyses might help actually disarm dangerous conspiracy theories.