The Qasmiya Bridge on the Litani River, which connects southern Lebanon to the rest of the country, was destroyed by the Israeli army on March 22, 2026. Photo by Megaphone on Wikimedia Commons (CC BY 4.0).
By Mohamed Soufan
Social media platforms such as X (formerly Twitter) are often treated as reflections of public opinion, with journalists using social media trends and posts to infer broader sentiment, especially during political conflicts.
Trending topics and viral posts can give the impression of broad, representative debate. But this perception can be misleading. An analysis of online discussions related to Lebanon showed that a small minority of users can dominate visibility and interaction, with the top 1 percent accounting for more than 60 percent of total engagement.
If a handful of highly active accounts shape most of what we see, what does “public opinion” online actually represent?
A small minority
Engagement — likes, reposts, and replies — is not evenly distributed. A small group of highly active or highly visible accounts captures a disproportionate share of attention. As a result, a limited number of voices can shape narratives, amplify specific viewpoints, and set the tone of discussion.
This is not always obvious. Repeated exposure to similar messages can create the impression of consensus, even when it is driven by a narrow segment of users.
This becomes clear in a dataset I compiled of Arabic-language posts about Hezbollah on X. The dataset includes 15,767 posts from 8,148 unique users collected between March 1 and March 8, 2026. Posts were identified using Arabic-language keyword variations of “Hezbollah,” and engagement was measured as the sum of likes, reposts, and replies. Accounts were classified as media or non-media based on profile metadata, supplemented by manual review of the most engaged accounts.
The results reveal a sharp gap between participation and visibility. Non-media users make up 89.6 percent of accounts and produce 79.9 percent of posts, yet the top 1 percent of users capture 61.5 percent of total engagement. The top 5 percent capture 90.6 percent , and the top 10 percent capture 96.2 percent.
Media accounts, although only 10.4 percent of users, account for 29.6 percent of the top 1 percent most engaged users and receive roughly 34 percent more engagement per post than non-media users — about 41 interactions per post on average, compared to 31. At the same time, most total engagement (74.8 percent) still comes from non-media users because of their much higher volume of posts.
Beyond social media
The implications go beyond social media itself. In practice, journalists, researchers, and audiences often turn to X during fast-moving events to get a sense of what is happening and how people are reacting. Trending topics, widely shared posts, and visible engagement can quickly become shorthand for public sentiment.
But these signals are not neutral. Research by Shannon McGregor shows that journalists sometimes rely on social media trends as indicators of public opinion, even though the users driving those trends are not representative of the broader population. A widely shared post or a trending hashtag can give the impression of a dominant public mood, even when it reflects the activity of a relatively small and highly engaged group.
Data from the Pew Research Center shows that political activity on X is concentrated among a small share of users. When the same accounts and viewpoints dominate what is seen, they can influence how events are framed, which voices are amplified, and which perspectives are overlooked.
This dynamic can reinforce itself. Content that gains early visibility is more likely to be seen, shared, and amplified again, making certain narratives appear more widespread than they actually are.
At the same time, social media data still provides valuable insights into how information spreads and narratives form, but it requires careful interpretation. This shifts the focus from ‘What is everyone saying?’ to ‘Who is being seen and heard — and why?’
In conflict contexts, this matters. When a small group dominates what is seen, it can create the impression of consensus where there is none. For those following events in Lebanon, highly visible reactions may reflect amplification rather than widespread agreement, making online discourse a misleading signal of public sentiment.
