Link-Traced Amplification: How Telegram Redistributes YouTube Political Content at Scale

Abstract

Telegram is a lightly moderated platform hosting, among others, fringe and politically extreme communities, many of which migrated after being deplatformed from mainstream social media. While prior research has mainly analyzed political debate on Telegram through in-platform text, less attention has been paid to how external media content flows into and circulates within this ecosystem. We study the circulation of YouTube videos shared in 43,000 public Telegram chats surrounding the 2024 U.S. presidential election, analyzing 686,625 English-language videos as traces of cross-platform connectivity. Using topic modeling, supervised classification, and multi-dimensional toxicity measures, we characterize which narratives are amplified, how they diffuse by analyzing reach, recirculation, persistence, and transfer time, and whether toxicity is related to amplification. We find that Telegram functions as an agenda-redistribution layer for YouTube political content. Political videos diffuse in bursty, short-lived patterns that track high-salience events. Toxicity is only weakly associated with reach; instead, more toxic content tends to persist longer within narrower thematic circuits, reinforcing segmented information environments.