Critical Slowing Down of Societal Anger predicts Peaks of Social Movement Activity

Social movements like #BlackLivesMatter, which have a strong representation on social media, often display abrupt peaks in activity. These peaks are important because they mobilize large populations, but they are also very hard to predict and explain. The current project focused on collective dynamics of anger as a predictor of online peaks. Specifically, we examined whether delayed recovery in anger online in the days preceding activity peaks, termed critical slowing down, predicted the occurrence of peaks. We analyzed the full archive of #BlackLivesMatter tweets between 2015-2020 (N = ~60 million). Congruent with critical slowing down theory, we found that increased autocorrelation in anger but lowered emotional variance were associated with a greater likelihood of peaks. These results suggest that delayed emotional recovery paired with lowered temporal emotional volatility predicts consecutive peaks in activism. By integrating emotional dynamics with complex systems theory, this project opens new avenues for forecasting tipping points in collective behavior.