The Popularity of extremism: A computational exploration of extremist political speech on Twitter

We define political extremism as a duality of internal ideology and external performance and explore whether its performative component on digital platforms is associated with greater popularity. We select three measures of extremism performance supported by prior literature and examine their ability to predict human-labeled extremism in crowd-sourced data. Then we observe how the social media posts' popularity is dependent on these measures. We find that, out of the three measures, sentiment has a statistically significant association with popularity, but could not yield evidence to confirm it to be a good predictor of extremism.