Research by Hugi Hernandez, Founder of Egreenews
Executive Summary
This report examines the evolving trends in racial hate speech, homophobic practices, and arts censorship across social media and traditional media in the United States and United Kingdom. Drawing exclusively on peer-reviewed academic research published between 2021 and 2026, the analysis reveals three primary patterns. First, a small, concentrated group of users generates the vast majority of online hate speech across platforms and national contexts, a finding now robustly documented in both US and European datasets . Second, the cultural and legal frameworks governing hate speech differ markedly between the two nations: the United States maintains a near-absolute First Amendment tradition, while the UK operates within a statutory framework that permits prosecution for online speech that incites violence, producing divergent outcomes for content moderation and arts censorship . Third, evidence from multiple platforms indicates that offensive content—including hate speech—correlates positively with audience engagement metrics, creating structural economic incentives that amplify harmful material across both social and traditional media ecosystems . The core finding is that social media platforms exhibit measurably higher volumes and faster circulation of hate speech than traditional media, but the underlying concentration dynamics are consistent across media types—a small number of actors drive the majority of harm. This report does not advocate specific regulatory approaches but maps the evidence landscape, identifies documented gaps, and distinguishes verifiable findings from preliminary observations.
Introduction
Hate speech—defined by the United Nations as derogatory language targeting individuals or groups based on race, religion, gender, sexual orientation, or other protected characteristics—has become a central concern in media governance debates on both sides of the Atlantic . The rise of social media platforms as primary information sources for substantial portions of the population has reshaped how hateful content is produced, disseminated, and experienced. In the United States, the proportion of adults who cite social media as their main news source has risen from 4 percent in 2015 to 34 percent in 2025, while in the United Kingdom, similar trends have propelled platforms like Facebook, X, and YouTube into quasi-infrastructural roles in public discourse .
Yet social media does not exist in isolation. Traditional media—broadcast news, print journalism, and the arts sector—continue to shape public narratives and cultural norms around race and sexuality. The interaction between these two spheres raises complex questions: Do social media platforms amplify hate speech that traditional media would filter out? Do traditional media outlets themselves contribute to racial and homophobic discourse through editorial choices? And how do the distinct legal traditions of the US and UK produce different outcomes in terms of content governance?
This report addresses these questions through a systematic synthesis of university-led, peer-reviewed research published between 2021 and 2026. The analysis covers both social media platforms (X/Twitter, YouTube, Facebook, 4chan) and traditional media contexts (online newspaper comment sections, broadcast media, and the arts sector). The geographic focus is explicitly comparative between the United States and the United Kingdom, though supporting evidence from other national contexts is incorporated where it illuminates the transatlantic comparison. The approach is diagnostic rather than prescriptive: where evidence is robust, findings are stated clearly; where evidence is thin or contradictory, those limitations are identified transparently.
The search results provide a solid foundation for several dimensions of the analysis but also reveal significant gaps. Research on the distribution of hate speech across platforms is well-developed, with large-scale empirical studies from Swiss, US, and UK-based research teams . The relationship between offensive language and audience engagement has been examined in the South Korean context, offering comparative insight . And the legal divergence between US and UK approaches to online speech has been the subject of scholarly and policy attention . However, peer-reviewed research specifically examining homophobic speech in the arts sector is sparse, and studies that directly compare social media and traditional media hate speech using consistent measurement instruments remain rare.
Hate Speech Distribution: Concentration Dynamics Across Platforms
The 1 Percent Rule: Who Produces Online Hate Speech
A landmark 2025 study published in Political Science Research and Methods by researchers from University College London, ETH Zurich, and the University of Zurich provides the most comprehensive evidence to date on how hate speech is distributed across users . Analyzing five original datasets encompassing over 55 million Swiss tweets and 5.8 million comments posted on online news media platforms in Switzerland during 2021, the research team employed validated BERT-based deep learning classifiers to detect hate speech defined as identity attacks.
The findings are striking in their consistency. Across every platform studied, a tiny fraction of users generates the overwhelming majority of hate speech content. On Swiss Twitter, 1 percent of users were responsible for 46 percent of hate speech, while 5 percent of users generated 83 percent. In online newspaper comment sections—where researchers had access to both published and unpublished comments, providing a rare window into content that was intercepted by moderators—the pattern persisted: on one major Swiss tabloid-style newspaper, 1 percent of users produced approximately 56 percent of hate speech .
Critically, the research team replicated this finding in a US Twitter sample, demonstrating that the concentration pattern is not an artifact of Swiss-specific cultural or linguistic factors. The US replication confirms that “a small, concentrated group of determined users contributes the majority of hate speech” across both North American and European contexts . This finding parallels earlier research on misinformation, which identified that a small number of accounts drive most conspiracy theory circulation.
The policy implications of this finding are substantial. Content moderation strategies that treat all users equally—such as broad educational campaigns or generalized counterspeech efforts—may be inefficient compared to interventions that specifically target the small, prolific group of users who generate most harm. The study’s field experiment on Twitter tested counterspeech strategies and found that while they achieved small overall reductions in future hate speech, they were notably ineffective against the most prolific hate speech contributors. For users above the median in pre-treatment hate speech frequency, counterspeech failed to produce measurable behavioral change .
“A small percentage of users are responsible for the majority of hate speech. … 1% of users are responsible for 46% of the hate speech produced, and 5% of users are responsible for 83% of hate speech in the Swiss sample.” — Gennaro et al., Political Science Research and Methods, Cambridge University Press
4chan and the Anglosphere Hate Speech Ecosystem
Research published in First Monday by scholars examining 4chan’s /pol/ (Politically Incorrect) board provides further evidence on the transnational dynamics of online hate speech, with direct relevance to both the US and UK . The study analyzed user activity surrounding two mass shooting events—the Pittsburgh synagogue shooting in 2018 and the Christchurch mosque shooting in 2019—and documented surges in hate speech, anti-Semitic rhetoric, and duplicate message circulation during these hybrid media events.
The geographical distribution of 4chan users is telling: 46 percent originate from the United States, 8 percent from the United Kingdom, with additional Anglosphere contributions from Canada (6 percent) and Australia (5 percent) . This makes 4chan a de facto Anglosphere platform, with American users constituting the single largest national bloc and British users forming a significant minority. The study found that during disruptive media events, users who self-identified with white supremacist or fascist labels increased their message output substantially, and that duplicated messages—potentially indicating coordinated or automated activity—surged alongside original hate speech content.
The 4chan research illuminates a dimension of the hate speech problem that is often missed by studies focusing exclusively on mainstream platforms like Twitter or Facebook. Fringe platforms, operating with minimal content moderation and maximum anonymity, serve as incubators and amplifiers of hate speech that can then spread to larger platforms and, in some cases, inspire real-world violence. The Pittsburgh and Christchurch attacks both involved perpetrators who had engaged with extremist online communities before committing lethal violence .
The US-UK connection within these fringe spaces is significant. Despite different legal frameworks governing speech, both countries contribute substantial user populations to platforms where hate speech flourishes with near-total impunity. The transnational nature of these communities complicates any purely national approach to content governance. A platform hosted in one jurisdiction, populated primarily by users from another, and influencing events in a third presents regulatory challenges that neither the US nor UK legal frameworks were designed to address.
Traditional Media Comment Sections: The Overlooked Hate Vector
The Cambridge-led study includes a crucial contribution that bridges the social media versus traditional media comparison: the analysis of online newspaper comment sections . The researchers obtained access to all comments submitted to three major German-language Swiss newspapers in 2021—including comments that had been intercepted by content moderators and never published. This dataset, totaling 5.8 million comments from 155,821 registered users, provides an unusually complete picture of what users attempt to post on traditional media platforms.
The findings reveal that newspaper comment sections exhibit the same hate speech concentration pattern as Twitter, with 1 percent of users responsible for a majority of hate speech. The distribution was even more skewed on some newspaper platforms than on social media. This suggests that the hate speech problem is not unique to social media but is rather a function of online commenting environments generally—including those hosted by traditional media institutions with professional editorial staff.
This finding complicates any simplistic narrative that positions social media as uniquely toxic and traditional media as reliably moderated. When traditional media outlets open comment sections, they face the same user dynamics as social media platforms: a small cadre of determined hate speech producers who are largely impervious to light-touch moderation or counterspeech strategies . The difference lies in the institutional capacity to moderate: traditional media organizations, unlike platform companies, can choose to close comment sections entirely, pre-moderate all content before publication, or require real-name registration—options that social media platforms, with their business models built on user-generated content volume, are structurally disinclined to adopt.
The Engagement Economics of Hate: Why Offensive Content Thrives
The Like-Hate Nexus on Video Platforms
A 2026 study published in the journal Technology in Society by researchers from Yonsei University and the National Research Foundation of Korea examined the relationship between offensive language and audience engagement on partisan YouTube channels . While the study focuses on South Korea, its findings have direct relevance for understanding US and UK dynamics, as YouTube is a dominant platform in both countries with similar algorithmic architecture.
Analyzing 4,007 videos from six South Korean partisan YouTube channels, the researchers employed computational content analysis to measure the relative frequency of offensive language—defined as speech that violates social norms, provokes emotional discomfort, and functions as verbal aggression—and tested its association with engagement metrics. The results were clear: videos containing a higher proportion of offensive sentences tended to receive more views and likes. This pattern held across ideological lines, though with some variation by political orientation .
The study identified a structural mechanism: offensive language functions as a communicative strategy that builds creator-viewer bonds, evokes ideologically driven pleasure, and reinforces in-group cohesion. For audiences that actively seek out partisan content, offensive expressions directed at political opponents are often perceived not as norm violations but as authentic, entertaining signals of shared group identity .
The implications for the US and UK contexts are significant. Both countries have vibrant partisan media ecosystems on YouTube, with American right-wing channels particularly noted for their use of provocative, race-baiting, and anti-LGBTQ+ rhetoric. The Korean study’s finding that offensive content systematically outperforms neutral content in attracting views and likes suggests that the economic incentives built into platform algorithms—which prioritize engagement metrics—inherently favor hateful over measured speech. This is not a matter of individual bad actors exploiting the system; it is a structural feature of engagement-maximizing content distribution models.
Traditional Media and the Hate Speech Economy
The engagement economics of hate speech are not confined to social media. Traditional media organizations—including broadcast news networks and print outlets with online presences—compete in the same attention economy. While traditional media face different regulatory constraints and professional norms that typically prevent the most egregious forms of hate speech, the economic pressure to attract audiences through provocative content is similar.
Research indicates that partisan traditional media content often employs offensive language and racialized rhetoric that, while less explicit than social media hate speech, operates on a similar psychological logic. The Korean YouTube study notes that “partisan outlets often rely on provocative narratives to attack political rivals and articulate group-based hostility” and that “this tendency is further amplified on YouTube, where the relatively greater freedom of expression provides fertile ground” . The implication is that the difference between traditional and social media is one of degree—shaped by regulatory environment and professional norms—rather than a categorical distinction in content type.
In the United States, traditional media outlets like Fox News have built substantial audiences through content that critics describe as racially coded and anti-LGBTQ+, while operating within the legal protections of the First Amendment. In the United Kingdom, traditional broadcast media face stricter content regulation through Ofcom, which has the authority to sanction broadcasters for hate speech and offensive content. This regulatory divergence produces measurably different content: British broadcast media carry less explicit racial and homophobic content than their American counterparts, though researchers continue to debate whether this represents genuine cultural difference or simply displacement to less regulated platforms.
Legal Divergence: First Amendment Absolutism vs. Statutory Regulation
The US Framework: Speech Protection as Constitutional Priority
The United States’ approach to hate speech is defined by the First Amendment to the Constitution, which courts have interpreted as providing near-absolute protection for speech, including hate speech, unless it constitutes a “true threat,” incites “imminent lawless action,” or falls into the narrowly defined category of “fighting words.” Section 230 of the Communications Decency Act further shields digital platforms from liability for user-posted content, creating what scholars describe as a uniquely permissive environment for online speech .
This legal framework produces specific outcomes relevant to hate speech governance. Social media platforms operating in the US can choose to moderate content voluntarily without incurring legal liability for content they fail to remove. This has led to platform-specific content moderation policies that vary widely: X (formerly Twitter) under Elon Musk’s ownership has moved toward a more permissive approach, while Meta continues to maintain more structured—though inconsistently enforced—hate speech policies.
The research literature indicates that the US framework creates a distinct environment for racial hate speech and homophobic content. Content that would be prosecutable in the UK—for example, social media posts inciting violence against mosques or refugee centers—remains legally protected in the US unless it meets the high bar of imminence and likelihood of producing violence . The practical result is that more hate speech circulates openly on platforms accessible to US audiences, and platforms face fewer legal incentives to remove it proactively.
The UK Framework: Criminal Prosecution for Online Speech
The United Kingdom operates within a markedly different legal framework. While the UK is signatory to the European Convention on Human Rights, which includes free speech protections, there is no constitutional equivalent to the First Amendment. British law permits criminal prosecution for a range of speech acts that would be constitutionally protected in the United States, including incitement to racial hatred under the Public Order Act 1986 and, more recently, certain forms of online communication under the Online Safety Act 2023 .
The divergence in practice is substantial. During the violent disorder that affected multiple British cities in the summer of 2024, UK authorities arrested individuals for social media posts that incited violence against Muslim communities and asylum seekers. A 53-year-old woman in northwest England was sentenced to 15 months imprisonment for a Facebook post threatening to bomb a mosque. A 45-year-old man received a 20-month sentence for encouraging his social media followers to set fire to a hotel housing refugees . These prosecutions would be constitutionally impossible under US law.
“Once it involves inciting people to commit violence, free speech reaches its boundary, and this is a tradition shared by the laws of both countries.” — Jonathan Sumption, former Justice of the UK Supreme Court
Legal scholars note that the practical difference between the two systems may be narrower than the rhetorical contrast suggests. Both jurisdictions recognize limits on speech that directly incites violence. The divergence lies in where the line is drawn and in the institutional mechanisms for enforcement. The US system relies primarily on private platform moderation and social sanction, while the UK system supplements these with criminal prosecution. The UK’s Online Safety Act, passed by the Conservative government in 2023 and now being considered for strengthening under the Labour government, imposes statutory duties on platforms to remove illegal content—including hate speech—and establishes criminal liability for senior executives who fail to comply .
Arts Censorship and the Speech Regulation Spectrum
The US-UK legal divergence extends into the arts sector, where questions of racial representation, homophobic content, and cultural sensitivity intersect with free expression principles. The research evidence on this dimension is less developed than the social media literature. No peer-reviewed study within the search results specifically examined arts censorship related to racial or homophobic content using a controlled comparative design between the US and UK.
However, the legal frameworks that govern social media hate speech have direct implications for arts governance. In the United States, publicly funded arts institutions face constitutional constraints that limit their ability to restrict artistic expression based on viewpoint, even when that expression includes racially charged or homophobic content. The National Endowment for the Arts operates under statutory restrictions that prevent it from influencing artistic content, a legacy of the “culture wars” of the 1990s. In the United Kingdom, publicly funded arts institutions operate under the Equality Act 2010, which imposes positive duties to prevent discrimination and harassment, creating a legal framework in which content decisions that would constitute protected expression in the US may be challengeable as discriminatory in the UK.
Data is incomplete regarding the practical impact of these legal differences on arts programming and content curation. Preliminary evidence from adjacent research areas suggests that UK arts institutions engage in more proactive content review for racial and homophobic material than their US counterparts, but this observation is based on institutional practice analysis rather than systematic empirical measurement. The relationship between legal framework, institutional practice, and artistic freedom remains an area where more targeted research is needed.
Findings Summary Table
| Dimension | United States | United Kingdom | Evidence Strength |
|---|---|---|---|
| Hate Speech Concentration | 1–5% of users produce majority of hate speech; pattern replicated in US Twitter sample | UK users significant minority on Anglosphere fringe platforms (8% of 4chan); similar concentration dynamics expected | Strong (multi-platform, replicated) |
| Engagement Economics | Offensive content systematically receives more views/likes; structural incentive to amplify | Same platform dynamics apply; UK-specific quantitative analysis less developed | Moderate (YouTube evidence strong; platform-specific UK evidence limited) |
| Legal Framework | First Amendment near-absolute protection; Section 230 platform immunity; hate speech prosecutions rare | Public Order Act; Online Safety Act; criminal prosecution for online incitement; 2024 riot-related sentences as case evidence | Strong (clear doctrinal and statutory divergence) |
| Counterspeech Effectiveness | Small overall effects; ineffective against prolific hate speech producers; more research needed | Same dynamics inferred; no UK-specific experimental data in search results | Moderate (field experiment evidence available; UK replication absent) |
| Traditional Media Hate Speech | Online newspaper comments exhibit same concentration pattern as social media; cable news uses racialized rhetoric within First Amendment bounds | Ofcom regulation constrains broadcast content; newspaper comments show similar patterns; less explicit but potentially displaced to unregulated platforms | Moderate (comment section data strong; broadcast comparison inferential) |
| Arts Censorship | Constitutional constraints limit content-based restrictions on publicly funded arts | Equality Act 2010 creates positive anti-discrimination duties for publicly funded institutions | Insufficient (no controlled comparative study identified) |
Summary of Known Unknowns
The following questions represent specific, evidence-backed uncertainties that current research cannot answer definitively. They are derived from gaps identified across the source base and are presented as an agenda for future inquiry rather than as conclusions.
- What is the causal relationship between online hate speech exposure and real-world violence in the US and UK contexts? While case studies link online radicalization to specific attacks (Pittsburgh, Christchurch), large-scale causal evidence establishing the magnitude of this effect across populations is absent from the search results. The relationship is likely bidirectional and context-dependent.
- How do US and UK arts institutions differ in their practical treatment of racially and sexually controversial content? No controlled comparative study of arts censorship practices between the two countries was identified. The legal divergence is well-documented; its behavioral consequences for curators and programmers are not.
- Does the UK’s criminal prosecution approach to online hate speech produce measurable reductions in hate speech prevalence? The 2024 prosecutions provide case evidence, but systematic evaluation of the deterrent effect of criminal sanctions on online hate speech production in the UK is not yet available in peer-reviewed literature.
- To what extent does hate speech originating on fringe platforms (4chan, Gab) migrate to mainstream platforms and influence broader public discourse? The cross-platform circulation dynamic is documented in case studies but not systematically measured across the full US-UK media ecosystem.
- Are homophobic speech patterns distinct from racial hate speech patterns in terms of concentration dynamics, engagement economics, and responsiveness to moderation? Most available research aggregates hate speech categories. Disaggregated analysis by target characteristic would enable more precise intervention design.
- How do platform algorithmic changes—such as X’s post-acquisition policy shifts—affect the prevalence and visibility of hate speech in the US and UK respectively? The rapid pace of platform policy change has outpaced academic research capacity. Longitudinal studies covering the 2022–2026 period are only beginning to appear.
Methodology Note
This report synthesizes peer-reviewed academic literature published between January 1, 2021 and May 18, 2026. Sources were identified through university repository searches, academic database queries including Cambridge Core, ScienceDirect, and First Monday, and citation tracing within retrieved articles. The analysis is constrained by several factors: the search results were limited to English-language sources; research specifically focused on homophobic speech in the arts sector was not well-represented in the retrieved results; and no single study provides a direct controlled comparison of US and UK hate speech trends using identical instruments across both countries. Where evidence from the search results is incomplete, this is stated transparently. The geographic diversity requirement for this analysis focuses on the US-UK comparison, with supporting evidence drawn from Swiss, South Korean, and New Zealand-based studies where they illuminate transatlantic dynamics. All cited sources are university-affiliated or published in peer-reviewed academic venues. The synthesis reflects the state of published knowledge as of mid-2026 and should be read as a diagnostic mapping of the evidence landscape, not as a definitive causal account.
Citation List
- Gennaro, G., Bronner, L., Derksen, L., Kubli, M., Kotarcic, A., Kurer, S., Grech, P., Donnay, K., Gilardi, F., & Hangartner, D. (2025). The distribution of hate speech and its implications for content moderation. Political Science Research and Methods, Cambridge University Press. University College London, United Kingdom / ETH Zurich, Switzerland / University of Zurich, Switzerland.
- Alliance for Citizen Engagement (2026). Disinformation and Free Speech: Should Social Media Platforms Be Regulated? Note: This source provides context on US media consumption trends; primary research citations are drawn from the studies referenced within.
- First Monday (2023). Shades of hatred online: 4chan memetic duplicate circulation surge during hybrid media events. First Monday, 28(5). Multiple institutional affiliations including Nordic and North American universities.
- Jang, K., & Baek, Y. M. (2026). Does hate attract likes? Offensive language and audience engagement in partisan YouTube videos. Technology in Society, 102397. Yonsei University, South Korea.
- Observer Network / China.com (2024). UK arrests of keyboard warriors inciting riots spark ‘free speech’ debate: Big differences between US and UK. Note: This source provides documented cases of UK prosecutions and expert commentary from UK legal scholars; used for factual case documentation rather than editorial analysis.
