Alignment of Polarised Structures in Trending Topic Discussions in the German Twittersphere

The next speaker at the Indicators of Social Cohesion symposium is Eckehard Olbrich, whose focus is on the evidence for polarisation in the German Twittersphere. This seeks to evaluate the claims about the role of social media as a driver of polarisation, and to address the negative impacts of such polarisation if such polarisation is indeed present. Polarisation might exist at issue, ideological, or affective levels, and these levels also intersect with each other, of course.

Taking the German Twitter data as a starting point, then, what are the issues that are being discussed, and what evidence for polarisation is there at this level; do these align to form broader patterns of ideological polarisation? Past research suggests that there is only a weak correlation between individual issues. The present project drew on Twitter’s trending topics list, and gathered tweets from the past two days for the five top trending topics over a period of more than two years from 2021 to 2023. Key topics related to sports, politicians, COVID-19, party politics in Germany, the media, holidays, and ways of life, in particular. Individual topics also clustered together, though, and these clusters covered areas like the Ukraine war, sports, politics, holidays, and COVID-19.

The mapping of retweet networks from these datasets is a valid indicator of polarised structures in the data. Through a process of mapping each trend and selecting and paring down the largest weakly connected components of these networks it was possible to identify whether these themes were polarised (two distinct network clusters) or not (one single major cluster), and there were clear trends for which trending topics were producing polarisation (politics, talk shows, etc.) and which were not (sports, weather, holiday greetings, but also some topics like Putin or Erdogan).

Where trends produced polarised clusters, then, how persistent was the user composition of these clusters across topics? Which topics aligned with each other in terms of their user membership? The project here also distinguished between influencers (who were most retweeted) and multipliers (who were most active at retweeting others – these indeed also be responsible for the creation of trending topics through their activity, of course). Both influencers and multipliers break down into a smaller left-wing and a larger right-wing group, who tend to appear in the same polarised clusters in trending topics.

Across trends themselves, there seems to be a strong alignment between COVID-19, climate, migration, and gender identity topics, while discussions relating to the unprovoked Russian attack on Ukraine stand out as being very differently polarised. This indicates some level of partisan sorting into broader ideologies for these four topics, while attitudes towards the Ukraine war are more complex.