Characterizing the Dynamics of Conspiracy-Related German Telegram Conversations during COVID-19

Abstract

Conspiracy theories have long drawn public attention, but their explosive growth on platforms like Telegram during the COVID-19 pandemic raises pressing questions about their impact on societal trust, democracy, and public health. We provide a temporal and network analysis of the structure of conspiracy-related German-language Telegram chats in a novel large-scale dataset, which captures a significant proportion of COVID-19-era conspiracy discourse in Germany, Austria and Switzerland. A preliminary assessment of the dataset reveals that 37% of shared links point to sources rated untrustworthy by NewsGuard, a proportion substantially exceeding those reported for other platforms and comparable discourse contexts, therefore attesting to the high prevalence of low-credibility information within this corpus. Conspiracy-related activity spikes during major COVID-19-related events, correlating with societal stressors and mirroring prior research on how crises amplify conspiratorial beliefs. We find that the top 10% of chats account for 94% of all forwarded content, portraying the large influence of a few actors in disseminating information. However, these chats operate independently, with minimal interconnection between each other, primarily forwarding messages to low-traffic groups. Lastly, we show that the dynamics of attention are much slower on Telegram than on algorithmically moderated platforms.