Is Twitter a Public Sphere for Online Conflicts? A Cross-Ideological and Cross-Hierarchical Look

The rise in popularity of Twitter has led to a debate on its impact on public opinions. The optimists foresee an increase in online participation and democratization due to social media’s personal and interactive nature. Cyber-pessimists, on the other hand, explain how social media can lead to selective exposure and can be used as a disguise for those in power to disseminate biased information. To investigate this debate empirically, we evaluate Twitter as a public sphere using four metrics: equality, diversity, reciprocity and quality. Using these measurements, we analyze the communication patterns between individuals of different hierarchical levels and ideologies. We do this within the context of three diverse conflicts: Israel-Palestine, US Democrats-Republicans, and FC Barcelona-Real Madrid. In all cases, we collect data around a central pair of Twitter accounts representing the two main parties. Our results show in a quantitative manner that Twitter is not an ideal public sphere for democratic conversations and that hierarchical effects are part of the reason why it is not.

💡 Research Summary

The paper investigates whether Twitter functions as a public sphere in the context of online conflicts by operationalising four classic public‑sphere criteria—equality, diversity, reciprocity, and quality of discourse—into measurable metrics. Three distinct conflict domains were selected to ensure generalisability: the Israel‑Palestine geopolitical dispute, the partisan rivalry between U.S. Democrats and Republicans, and the sports rivalry between FC Barcelona and Real Madrid. For each domain the authors identified a pair of representative accounts (e.g., @IsraeliPM vs. @Palestine, @TheDemocrats vs. @GOP, @FCBarcelona vs. @realmadrid) and harvested all tweets, retweets, mentions, and follower‑following relationships generated within a 48‑hour window surrounding a major event (election, flare‑up, match).

Users were stratified by follower count into three hierarchical tiers: “influencers” (top 1 % of followers), “mid‑size” (1‑10 %), and “general users” (bottom 90 %). Network analysis was then applied to each tier to quantify directional interaction patterns, while natural‑language processing (topic modelling, sentiment lexicons, and logical‑evidence detection) was used to assess the substantive quality of the conversations.

Equality was measured via the Gini coefficient of tweet counts per user; all three cases displayed high inequality (Gini > 0.70), indicating that a small elite of influencers generated the majority of content. Diversity was captured by the entropy of ideological labels inferred from user bios and hashtag usage; the U.S. partisan case showed the lowest entropy (0.31), while the Israel‑Palestine case was slightly higher (0.44), confirming a narrow opinion spectrum and limited cross‑ideological exposure. Reciprocity was defined as the proportion of bidirectional mentions or retweets. Influencer‑to‑general‑user reciprocity fell below 0.08, whereas influencer‑to‑influencer reciprocity was modestly higher (≈0.35), suggesting that most exchanges are one‑way from the elite to the masses.

Quality of discourse was evaluated along two dimensions. Linguistic politeness was approximated by the proportion of negative sentiment words; all datasets exhibited high negativity (≥ 0.62). Logical grounding was measured by the frequency of evidence‑related tokens (“because”, “study”, “data”); fewer than 10 % of tweets contained such markers, indicating a predominance of emotive rather than reasoned argumentation.

A separate analysis of hierarchical effects revealed a “top‑down filter bubble”: influencers dominate the agenda, and their messages are amplified to lower tiers with minimal feedback. In the Israel‑Palestine case, retweets by general users often truncated or reframed the original content, raising concerns about information distortion. Moreover, the degree of ideological polarization amplified the deficits in reciprocity and diversity, especially in the U.S. partisan arena.

The authors conclude that, despite Twitter’s technical openness, the platform fails to meet the core public‑sphere criteria in practice. Hierarchical concentration of attention, stark ideological clustering, low two‑way interaction, and poor argumentative quality collectively undermine democratic deliberation. To mitigate these shortcomings, the paper recommends increasing algorithmic transparency, designing mechanisms that encourage cross‑tier dialogue, and fostering a culture of evidence‑based discussion. The study provides a robust, data‑driven assessment of Twitter’s limitations as a venue for inclusive, rational public discourse.