Polarização virtual: estudo da dinâmica de cenários politicamente polarizados em sites de redes sociais

Abstract

The political discourse in different countries shows distinct levels of polarization, which tend to be stronger in social networking sites, given the ease of connection and engagement with likeminded individuals. One way to reduce polarization would be through a balanced distribution of information showing distinct sides of the same reality among individuals with opposing political views. This study investigates whether this applies in the context of national elections in Brazil and Canada. Political tweets were collected before and after the votes in each scenario, on which the polarization dynamics were analyzed through retweet networks separated weekly. Over these networks, nodes able to distribute information among users with divergent political orientations, called “bubble poppers”, were identified. A new centrality metric called “bubble popper” was proposed and applied for this task – a fundamental metric for this study. Then, the degree to which users, polarized or not, were involved with content shared by bubble poppers, its source (domain), and its theme (topic) was analyzed, considering their political orientation and from who retweeted. As a result, it was identified that users with the highest value for the bubble popper metric were from accounts representing news media with neutral political orientation. While these accounts disseminate information that evenly reaches groups with distinct political orientations, users tend to engage considerably more with content that aligns with their political view, regardless of topic. This same phenomenon was observed in the Brazilian and Canadian political situations, despite having different levels of polarization. This result shows that efforts made by neutral news media to cover different political views in a balanced way may not contribute to reducing the polarization in social networking sites.