I will present our work on the Twitter influencers and the increased polarization during the last two U.S. presidential elections.
This is a collaboration with Stuart Feldman, James Flamino, Alessandro Galeazzi, Brendan Cross, Zhenkun Zhou, Matteo Serafino, Hernán A. Makse and Boleslaw K. Szymanski.
Abstract: New social media are decentralized, interactive, and transformative by empowering users to produce and spread information to influence others. In short, they changed the ways information and influence spread. Here, we analyze dynamics of political polarization among Twitter users using hundreds of millions of tweets that we collected over the 2016 and 2020 US presidential elections. From this data, we recreate news diffusing Twitter retweet networks segregated by political orientations. We identify top influencers in each news category in terms of their ability to spread information. The top influencers are classified into those affiliated with a traditional media organization, or with a political organization, or unaffiliated. Most of the top influencers were affiliated with media organizations during both elections. We measure the strength of polarization among them using different methods to assure robustness of the result. We find a clear increase of their polarization from 2016 to 2020. We also find that 75% of the top 100 influencers of all media categories in 2020 were not there in 2016, demonstrating how difficult it is to retain influencer status. The majority of influencers affiliated with traditional media shrunk their fraction by 10% from 2016 to 2020. The replacement came mostly from influencers affiliated with political organizations with center or right orientations. Independent influencers advanced too, overtaking about one third of the majority drop. These and other results create a foundation for understanding how new social media increase polarization and transform the election process.