The Structure of Participation and Attention in Arabic-Language Hezbollah Discourse on X

A new computational social science study by researcher Mohamed Soufan, published on arXiv, provides a detailed structural analysis of Arabic-language discourse about Hezbollah on X (formerly Twitter). The research examined a dataset of 15,767 tweets posted by 8,148 unique users during a one-week period in March 2026. The findings reveal a starkly unequal distribution of audience attention: while thousands of users participated by posting content, engagement metrics were overwhelmingly concentrated among a tiny elite. Specifically, the top 1% of users captured 61.5% of all engagement (likes, retweets, replies), and the top 10% accounted for a staggering 96.2% of total engagement.

Despite this concentration of attention, the conversation was broadly participatory in terms of content creation. Non-media users—those without media-related keywords in their profile metadata—made up 89.6% of the user base and authored 79.9% of the tweets in the dataset. However, accounts identified as 'media' wielded disproportionate influence, receiving an average of 41.32 interactions per tweet compared to 30.84 for non-media users. These media accounts were significantly overrepresented among the most-engaged users, suggesting they act as central hubs that shape and amplify the discourse. The study, which employs network analysis and statistical methods common in the fields of computers and society (cs.CY) and social and information networks (cs.SI), highlights the tension between open participation and concentrated influence in online political ecosystems.