Now more than ever, people are standing up for what they believe in and marching for it. Unfortunately, as they have throughout history, these demonstrations sometimes end in violence. But now, researchers might have found a way to foresee just when this will happen.

Researchers at the University of Southern California believe that they can use social media to predict whether or not protest will turn violent.

In the study, published in Nature Human Behaviour, researchers found that a “moral convergence” over an issue is typically the cause of violent protest. This convergence refers to the idea that violence is more likely to take place when people regard something as a moral issue and believe that others share their belief.

With this in mind, the team created an artificial intelligence that scanned through some 18 million tweets posted during and before protests in Baltimore in 2015 after the death of Freddie Gray. The algorithm looked for a correlation between “moralized” language (i.e. posts regarding issues that the user found either definitely right or wrong) used on the social media platform, and arrest rates (which they used a proxy, albeit a controversial one, for the level of violence) during the protest.

They found that arrests made during the demonstrations were directly reflected in the moralized tweets posted in the hours leading up to the protest and that these tweets nearly doubled on days when protesters and police violently clashed.

To put it simply, moral posts on social media equaled violence in the real world.

“By tracking moralized tweets posted during the 2015 Baltimore protests, we were able to observe that not only did their volume increase on days with violent protests, but also that their volume predicted hourly arrest rates, which we used as a proxy for violence, during the protests,” Joe Hoover, a USC PhD student who led the study, told Digital Trends. “To further unpack these effects, we conducted a series of controlled behavioral experiments and we consistently observed the same effect of moral convergence.”

The information will no doubt be useful to authorities, allowing for better preparation against demonstrations that could potentially escalate into violence—perhaps even, depending on how law enforcement handles the situation, curbing that escalation before it takes place.

One minor setback with the algorithm however, as pointed out by Android Headlines, is that social media platforms are currently working to limit controversial/hate speech. Moving forward, this would potentially keep this “moralized” speech from ever being posted which would make this research obsolete.

For now, we can only wait and see.


Cover Photo by Damien Checoury

Photo 1 by Monica Melton