From Attention to Dialogue: Does Audience Engagement Reinforce Constructive Cross-Party Communication?

Collection of recent research (arXiv, Mar 24 2026) covering social-media dynamics, platform auditing, underground forums, and ML methods with security-relevant implications. Highlights include: empirical audits of TikTok personalization and YouTube comment dynamics showing algorithmic steering and asymmetric rewards for hostile cross-cutting attacks; evaluation showing link-deletion is a limited mitigation for information diffusion; analysis of a long-running Russian–English invite-only underground forum using unsupervised topic/clustering to surface language-bound ‘dark jargon’ and sub-commun