Community dynamics and echo chambers: a longitudinal study of the Belt and Road Initiative in the Twittersphere during COVID-19 pandemic

The “echo chamber” effect has been widely decried as a destructive consequence of social media on public discourse, leading people to repetitively encounter their pre-existing perspectives without rebuttal in online political discussions. In this research, we have tested this proposition by using text analytics and the hybrid Louvain method and K-mean clustering to categorize the diffusion patterns of more than 158,000 reposted threads about the Belt and Road Initiative, a massive and controversial Chinese-led infrastructure development strategy with profound economic and geo-political implications, transmitted before and during the early stage of the COVID-19 pandemic. We found that while echo chambers persist temporally within certain like-minded communities, the overall public opinion is not necessarily polarized and highly sentimentalized. The magnitudes of opinion polarization are contingent on topics of information diffused by opinion leaders, and we estimate that the most severe isolation from dissimilar opinions only existed among <15% of users, who retweeted highly positive opinions on a singular topic. Temporally, we captured the counter-polarization effect among the majority of users during the pandemic, which poses a challenge to the common belief that social media echo chambers act as a mechanism for intensifying divisions among people.