This article identifies and analyses social media racist discourse, using as a case study a stabbing incident in Dundalk, a small town in Ireland. This fatal incident was for a brief period of time considered a terrorist attack, based on speculation on the accused’s nationality. Analysing the online discourses on this event, the article aims to show that conceptual and operational distinctions between organised, extreme racism and ambient, banal everyday race talk, do not in fact hold, and may even contribute to the circulation of racism in digital environments. Recent studies and reports have observed surges in racism, hate crime and discriminatory practices. In Ireland, in the first six months of 2017, there was an increase of 33% in (reported) hate crime, compared to the previous six months, and overall a consistent rise since 2013 (Michael, 2018). Of the incidents reported, over one third were instances of online racist hate speech. In the UK, the Organization for Security and Co-operation in Europe (OSCE) reports that 95,552 instances of online racist hate speech recorded by the police in 2017, a 15% rise since 2016 and almost a 50% rise since 2013 (OSCE, n.d.). Tell Mama, a UK-based civil society organisation measuring anti-Muslim attacks, recorded 1,223 incidents in 2016, of which 64% were street/offline attacks and 36% online, an overall 47% increase from the previous reporting year of 2015 (Tell Mama, 2017). The spread of (digital) hate has also been noted by the European Commission and the Council of Europe, leading the former to establish a Code of Conduct with social media platforms for taking down illegal hate speech in a systematic and timely fashion, and the latter to launch the No Hate Speech Movement, a youth-based campaign against hate speech.

In response to this rise in racist hate speech and hate crime, offline and online, there is a growing body of work that looks to address the types of online racism (Awan, 2014), the discursive shifts in racist repertoires (Lawson, 2018; Titley, 2016a; Titley, 2016b), the ambiguous role of new technologies and media (Farkas, Schou and Neumeier, 2018; Lentin and Humphrey, 2017), the increasing visibility of organised racist social media pages and accounts (Daniels, 2009a and 2009b; Ekman, 2015; Ekman, 2018) and the spillover from the offline to the online and vice versa (Klein, 2017). Building upon this research, the current contribution seeks to understand the differences between offline and online hate speech, if any, and the connections between the everyday banal, ‘ambient’ digital racism (Essed, 1991; Sharma, 2017) and the organised and/or extreme kind that is typically associated with the far right in Europe and the US. Essed (1991) writing in the 1990s, in the post-civil rights world where racism was seen as something of a past era, pointed out the importance of everyday encounters as a crucial locus for the reproduction of racism. Similarly, Lentin (2016) makes a distinction between ‘frozen’ and ‘motile’ racism, whereby the first is clear to all, but the latter becomes a moving target, the subject of discussion, denial and questioning. In the everyday context, racism becomes ‘banal’, shifts in forms, is experienced in a fragmented manner as a set of micro-aggressions (Sue, 2010) and becomes elusive and consistently questioned and denied (Titley, 2016a). On the other hand, most forms of organised, frozen, overt and explicit racism are seen as unacceptable and an aberration, as found for example in definitions of ‘illegal hate speech’ adopted by the European Union and most of the nation-states that comprise it (EC, 2008). Similarly, social media corporations have developed their own terms of service (Twitter), or ‘community standards’ (Facebook), which draw upon these understandings of illegal hate speech, but which also allow certain contents to circulate under rules governing freedom of expression and public debate. In this manner, we obtain a dichotomy of what effectively constitutes ‘acceptable racism’ and what is deemed unacceptable or even illegal.

Drawing on a dataset of posts following an event in a small town in Ireland, this article will seek to show that there are very few differences between these two forms, which are often combined and circulated in tandem. The article will begin with a review of the literature on digital and online racism. It will then move on to discuss how the dataset under analysis was generated and the methodology by which the analysis was conducted. The analysis will show the ways in which in practice very little separates the forms of racism that circulate in everyday online discussions and those found in the accounts and spaces occupied by those associated with the far right. Additionally, both kinds, if we accept this division, seem to circulate unhindered and for the most part unreported. The final section attempts to examine some of the implications of the findings, especially in terms of the recent Code of Conduct signed between the European Commission and social media platforms. The focus on illegal hate speech and enforcement appears to allow certain contents to circulate while arbitrarily prohibiting others which form part of the same overall discourse; the result is that the focus is on governing and managing digital racism rather than seeking to eradicate it.

Digital Racism

One of the earliest articles that studied the internet from a socio-cultural perspective was Robert Eatwell’s ‘Surfing the Great White Wave: The Internet, Extremism, and the Problem of Control’. Published in 1996, two years before Google was founded and almost ten years before the rise of social media, the article discussed the white supremacist site Stormfront, the potential of the internet for such groups and the difficulty of controlling online communications. What is important to note here is that the expansion of supremacist groups online began almost as soon as the World Wide Web allowed individuals and groups to go online. While this may appear surprising, especially given the association of the early internet with libertarian virtual communities (Rheingold, 2000), it is noted here to show the intricate and intimate links between the web and racism. The first line of inquiry into digital racism then concerns the connections between technology and racism. Even conceding that racism is always present, the preponderance of racist hate online (Daniels, 2013) still requires explanation and explication especially given the many forms it takes. A second line of inquiry looks at this issue and the new impetus for racism afforded by the digital space. These areas of inquiry foreground the institutional role of the regulators of the internet, which in the first instance include the social media platforms themselves who have become the new ‘custodians of the internet’ (Gillespie, 2018). This constitutes the third line of inquiry as, ultimately, their policies shape the form of digital racism. It is here, at the operational level, that we find the clearest separation between ‘ambient’ racism which is seen as acceptable or at least as something than can be tolerated, and illegal, extreme and unacceptable racism that has to be taken down. This section will discuss these lines of inquiry, concluding that a question: how different are these two racisms?

Technology and Racism

Notwithstanding the links between the US military and the internet, it is the liberal Californian ideology (Barbrook and Cameron, 1996) that shapes current technology and technological policy as well as public understandings of it. This ideology revolves around discourses of freedom, individualism and egalitarianism but holds that these can only be enacted through economic liberalism and individual entrepreneurship which will give rise to a bright technological future. As Barbrook and Cameron put it, the Californian ideology is ‘a bizarre mishmash of hippie anarchism and economic liberalism beefed up with lots of technological determinism’ (1996: 52). But this ideological patina fails to mask the undercurrent of racism and inequality that underpins not only the relations of production of technology and its products but also the codes and algorithms that make up the internet.

As Nakamura (2002) put it, the default internet user is a middle-class white male, with everyone else ‘cybertyped’ from futuristic Asians to ‘techno-primitive’ Blacks. The supposed colour-blindness of technology masks the racialised reality of its production, which shows a typical racial hierarchy of white men and a few white women at the top, buttressed by men and women of the global south that do the manual work of assembling the circuits either as immigrant labour in the global north or as outsourced workers in the south (Daniels, 2015; Dyer-Witheford, 2015; Gajjala, 2004). Well-meaning studies of the racial digital divide that understand race as a variable end up creating a deficit model in which minorities are lacking, forever trying to play catch up (Zuberi and Bonilla Silva, 2008). Perhaps because of these structural inequalities, the coding language itself is racially loaded. Moreover, in a context of constant surveillance and securitized technocapitalism, technologies often become instruments for further subjugation and control. A 2016 investigation by Angwin et al. showed that risk assessments in penal contexts conducted via algorithms systematically discriminate against minorities, for example denying them bail or contributing to higher sentences. Algorithms can only be trained on existing data, which are derived from societies that are already racially structured, hence reproducing the current racial logics. Another example here is Microsoft’s attempt to use Twitter crowds to train a chat-bot: in less than a day, the bot was trained to use racist and misogynistic language (Vincent, 2016).

Technological artefacts, apps and platforms are further used as a means of racial differentiation. Boyd argued a contributory factor to the demise of MySpace was its association with Black users (2013). A flipside to this is the rise of Black Twitter, a space created by Black publics within Twitter, which constitutes a Black cultural artefact (Brock, 2012; Sharma, 2013). While Black Twitter has been seen as empowering, it is also a means of foregrounding racialised practices. As Brock (2012) notes, Twitter’s affordances are mediated differently depending on the racial affiliation of those using it. On the one hand, Twitter acts as a facilitator of cultural deficit racial stereotypes, such as, for example, that Black people are not as technologically advanced as Whites; on the other hand, it can be seen as an important forum for civic action (Brock, 2012). Sharma (2013) understands race and Twitter as an assemblage whereby racialised subjects multiply, contest and transmit different ways of being raced online. In Sharma’s view, neither the technologies nor the users are pre-constituted but emerge out of their coming together, opening up new ways of being.

This brief exposition outlines some of the ways in which new technologies reproduce dominant racial understandings, but also offer the potential for to subvert these, through the practices and appropriations of technologies by racialised subjects. However, looking at the kinds of contents and representations that circulate in the online platforms that have come to structure the internet it seems that racial hate prevails.

Online Racist Hate

There is a large body of research that has identified online racist hate speech, focusing on those who develop and share racist hate, on the specific targets and on the various internet and/or digital genres that host such hateful speech. Taken together, these aspects form a very bleak picture of a digital environment that is not only racialised, in the sense that it is becoming inimical to, and marginalises or ghettoises non-white, non-Western people, but which also actively robs these people from participating in the construction of a common digital future.

In terms of the tropes of racist hate, important work has been conducted by think tanks such as the UK-based Demos. Using data mining through complex algorithms and qualitative content analysis of selected content, Demos has published studies on the prevalence and patterns of use of racial and ethnic slurs on Twitter (Bartlett et al., 2014; Bartlett and Krasodomski-Jones, 2016); on Islamophobia spikes on Twitter in the immediate aftermath of news events, especially terrorist attacks (Miller and Smith, 2017); on the volume of derogatory and/or hateful anti-Muslim tweets; as well as the impact of the EU Referendum on xenophobia and racism. Demos (Bartlett et al., 2014) estimated that about 1 in every 15,000 tweets contains an ethnic or racial slur but included terms such as ‘whitey’ which do not carry the same force as some of the derogatory terms used for black, Asian or Jewish people. Demos further found that Islamophobic tweets surge after specific incidents, such as terrorist attacks. Following the Nice attacks in July 2016, Miller and Smith (2017) reported an average of 286 Islamophobic tweets per hour. In a study of the aftermath of the EU referendum, Miller et al. (2016) found that about 40% of the tweets on Brexit, in the period 19 June–1 July 2016 used a term that showed a xenophobic or anti-immigrant attitude. Studies such as these show clearly the extent of problematic contents circulating, contents that may not qualify for removal as ‘illegal hate speech’ and which therefore remain online.

Increasing volumes of racist hate were also observed by other monitoring organisations, such as the UK-based Tell Mama. In its 2017 report, the organisation reported a clear escalation, with a 31% rise in verified incidents compared to 2016. The organisation also noted spikes following specific incidents: for example, the Brexit referendum led to a 475% increase in reported anti-Muslim incidents compared to the week before the referendum and a 700% surge following the terrorist attack in Manchester, in May 2017, compared again to the week before the attack. Tell Mama monitors both physical and online hate incidents. In terms of the online incidents, Tell Mama highlighted the role of bots and foreign-run accounts. Two of those, @DavidJo52951945 and @SouthLoneStar, were subsequently found to be operated by the so-called Internet Research Agency based in Russia (Booth et al., 2017). Tell Mama noted that these accounts operate transnationally, using Islamophobic hashtags such as #BanIslam and #CultofDeath. In terms of the content of online racist speech, Awan (2014) analysed 500 Islamophobic tweets, identifying some of the dominant hateful discourses, consisting of words or phrases such as ‘Muslim pigs’ (9%), ‘Muzrats’ (14%), ‘Muslim Paedos’ (30%), ‘Muslim terrorists’ (22%), ‘Muslim scum’ (15%) and ‘Pisslam’ (10%). Awan’s analysis took place after two incidents, the murder of Lee Rigby in Woolwich (2013) and the conviction of several Asian men for grooming young women for sex in Rotherham, the extent of which was publicised in the UK media in 2013. The analysis maps the surge in Islamophobia that typically follows such news events, but not necessarily the more banal, everyday expressions.

In the US, it is primarily white supremacy contents that have been mapped and understood, by theorists such as Daniels (2009a; 2016). Daniels (2009a) compared print to digital white supremacism, highlighting the new opportunities for this ideology afforded by the new media. Overall, white supremacists online mobilise anti-Semitic conspiratorial discourses about Jewish control of the media and the complicity of mainstream politicians that does not allow the ‘natural’ superiority of white identity to emerge. Brown (2009) discusses online supremacist discourse as positing a strict racial and gendered hierarchy with Western European men at the top, using biological and similar essentialist arguments, which are described as ‘facts’. Daniels (2009b) further points out the role played by ‘cloaked websites’, which are sites that hide their authorship and links to political causes such as white supremacy, pretend they are neutral, informational websites and peddle racist propaganda, such as, for example, that slavery was an idyllic social system. More recently, cloaked pages were examined by Farkas, Schou and Neumayer (2018) who found that fake Islamist Facebook pages are used to stoke racist hate in Denmark. Taking stock of this body of work, alongside the rise of the alt right, Daniels (2018) shows that the expressed purpose of organised racist groups in digital environments is to push the boundaries of acceptable discourse; Daniels (2018: 64) quotes a white nationalist on Twitter claiming that ‘People have adopted our rhetoric, sometimes without even realizing it. We’re setting up for a massive cultural shift’.

In Europe, the rise of identitarianism associated with political movements and parties such as the Golden Dawn in Greece (Siapera and Veikou, 2016), Pegida in Germany (Vorländer, Herold and Schäller, 2018), the True Finns and Sweden Democrats in Northern Europe (Teitelbaum, 2016) and Jobbik in Hungary, points to the association of white supremacy with nativist and ethno-nationalist ideologies. According to Ahmed and Pisoiu (2017) the three core ideas of identitarianism are: (i) Ethnopluralism, which refers to the idea that cultures should strive to retain their uniqueness, but separately from one another. Ethnopluralism is connected to anti-migrant and strong border discourses; (ii) Post-ideology, which refers to an ideological positioning beyond left and right and which sets up identitarianism primarily as a cultural movement. Post-ideology is associated with a populist critique of all non-identitarian political parties and the political establishment; and (iii) ‘Retorsion’, the idea that the majority is now somehow a threatened minority in its own territory, a situation which needs to be resisted. Retorsion is linked to discourses of ‘replacement’, and to anti-refugee and anti-migrant positions.

While this body of work illustrates the ways in which social media have provided another platform for racist hate, especially by organised groups, another line of inquiry focuses on the ways in which racism is denied in digital environments. Titley (2016a, 2016b, forthcoming) has argued that the proliferation of racist and anti-racist voices in digital environments, the opportunities for discursive exchange and the race-blindness of digital platforms, within a post-racial ideological context, are factors contributing to the rise of the debatability of racism. For Titley, the very existence of these debates contributes to silencing and denying racism through making noise, constantly debating and arguing about what is race and racism. Moreover, their existence is a political issue as it raises important questions as to who gets to define racism. While this debatability is the result of what Lentin (2016) calls the motility of racism, its dynamic, generative and evolving character, along with online affordances, such as the spreadability of contents (Jenkins, Ford and Green, 2018), actively contribute to its current shape.

Taking the above into account, the picture that is emerging shows that digital and social media have enabled both the proliferation of racist discourses and a shift towards a different kind of racism that is spreading racist hate all the while denying its existence. It is within this context that the role of the ‘custodians of the internet’ (Gillespie, 2018) emerges as crucial.

Platform regulation of contents

The policies used by digital platforms are directly responsible for the contents we see. What is less clear however is where these policies are derived and how they are implemented. To begin with, in the US the legal context within which platforms operate is the so-called safe harbour policy, established by Section 230 of the Communications Decency Act of 1996 (Gillespie, 2018). This policy prevents platforms from being liable for contents published on them, on the basis that they are not publishers but provide access to the internet. In a second move, Section 230 allows platforms to remove some contents if they so choose without losing their safe harbour status. This dual provision, argues Gillespie (2018), shaped the platform approach to content moderation. A similar policy exists in the European Union: in the 2015 Delfi vs Estonia case (ECtHR 64669/09), on the extent to which an online news publisher (Delfi) should be liable for contents posted by users in the comments section, the European Court of Human Rights upheld the original decision of the local Estonian courts, making publishers liable. The Court further considered that its judgment was not applicable in the case of an internet forum, social media platforms or comments left on personal blogs. Social media platforms specifically are protected by the limited liability for internet service providers (ISP) in terms of the E-Commerce Directive of 2000 (Directive 2000/31/EC). Social media platforms are considered ‘mere conduits’ and hence not responsible; however, the directive states that an ‘information society service […] upon obtaining knowledge or awareness of illegal activities has to act expeditiously to remove or to disable access to the information concerned’ (Article 46: L 178/6) and that ‘Member States are prevented from imposing a monitoring obligation on service providers’ (Article 47: L. 178/6). These requirements create a context whereby social media platforms only voluntarily impose content controls according to their own criteria and rules, and typically act retro-actively, upon receiving reports of illegal contents, not proactively by pre-moderating contents before they publish them.

While still voluntary, the European Commission (2018a) signed a code of conduct with four platforms (YouTube/Google, Facebook, Twitter and Microsoft) in 2016, while more platforms joined in 2018 (Snapchat, Google+ and Instagram). The basic premises of the code of conduct require that companies must make clear the reporting process, that they must review and if necessary remove illegal contents within 24 hours of notification, that they should form more relationships with civil society and in general be more transparent in terms of how they operate with respect to such contents. The Commission monitors how the code is applied. The most recent report, in January 2018 (European Commission, 2018b), noted considerable progress with 70% of reported illegal contents taken down compared to 28% in the first monitoring exercise in 2016. Notwithstanding the importance of the code of conduct, the whole exercise hinges firstly on what is considered illegal hate speech and secondly on users actually reporting contents.

In the EU, the definition of illegal hate speech is determined by the EC 2008 Framework Decision on combating certain forms and expressions of racism and xenophobia by means of criminal law. The framework decision expects member states to introduce legislation against xenophobia and hate speech that includes incitement ‘to violence or hatred directed against a group of persons or a member of such a group defined by reference to race, colour, religion, descent or national or ethnic origin’; the ‘public dissemination or distribution of tracts, pictures or other material’ that incites to violence or hatred; and the trivialization of genocide, crimes against humanity, war crimes and crimes covered by the Tribunal of Nuremberg (European Commission, 2008: 328/56). In contrast, the US First Amendment does now allow any content-based regulation, with two exceptions: libellous and defamatory contents and/or ‘fighting words’, i.e. incitement to violence (Cornell Law School, n.d.).

Social media platforms navigate this regulatory environment and its tensions by creating their own ‘terms of service’, ‘community standards’ and policies (Twitter, Facebook and YouTube respectively). In general, these policies draw upon the UN Declaration of Human Rights of 1948 and the subsequent International Covenant of Civil and Political Rights, and specifically Article 19 on Freedom of Expression, and Articles 20 and 26 that prohibit advocacy of racial, religious or national hatred that incites discrimination, hostility or violence and establish the right to non-discrimination on grounds such as such as race, colour, sex, language, religion, political or other opinion, national or social origin and equality before the law for all these groups (UN, 1966). For instance, Facebook (n.d.) defines hate speech as:

A direct attack on people based on what we call protected characteristics – race, ethnicity, national origin, religious affiliation, sexual orientation, sex, gender, gender identity and serious disability or disease. We also provide some protections for immigration status. We define ‘attack’ as violent or dehumanising speech, statements of inferiority, or calls for exclusion or segregation.

On the other hand, when the training materials used by Facebook were leaked to the media, it emerged that the procedures used created a two-tiered system, whereby certain contents were (rightly) seen as unacceptable while other equally problematic contents remained. In the leaked documents, it is made clear that while for example followers of a religion (e.g. Muslims) are protected, the religion itself is not. In enforcing its rules, Facebook has made a distinction between protected, not protected and quasi-protected categories. The protected categories are those referred to above. Non-protected categories include social class, appearance, countries, political ideologies, occupations and so on (Guardian, 2017: n.p.). The quasi-protected categories include refugees, migrants and transgender people. In an article in the New York Times, Carlsen and Haque (2017: n.p.) show how Facebook allows statements such as ‘Poor black people should still sit in the back of a bus’ but not statements such as ‘White men are assholes’. This is because ‘white men’ combines two protected categories, race and gender, while ‘poor black people’ combines a protected category, race, with a non-protected category, social status, which according to the leaked training materials means that the protected category loses its protected status. Additionally, the leaked documents revealed various rules that allow, for example, calls for dismissing or excluding whole groups of people, but which delete dehumanising generalizations. The emergence of these kinds of rules points to the need for digital platforms to balance conflicting demands: the demand to allow contents to circulate as this is their main business model with the demand to keep out hate speech (Siapera, Viejo-Otero and Moreo, 2017). In doing so, platforms have created a division between acceptable and unacceptable racist speech, whereby unacceptable racist hate is seen as dehumanising and harmful but acceptable racist hate is seen as ‘allowing for a broad public discussion’ (Guardian, 2017: n.p.).

In this, platforms appear to understand racism as an expression of a set of bad attitudes or beliefs – what Lentin (2016) calls ‘frozen’ racism, often associated with extremists. This is the unacceptable racism that Facebook and other social media seek to control. For example, Facebook has implemented a long-standing ban on the Greek far right party Golden Dawn, and has recently banned the US-based group Proud Boys, a self-proclaimed ‘Western chauvinist’ organisation. In justifying this decision, Facebook noted that they continue:

To study trends in organized hate and hate speech and works with partners to better understand hate organizations as they evolve […] We ban these organizations and individuals from our platforms and also remove all praise and support when we become aware of it (quoted in Price, 2018: n.p.).

However, other forms of ‘ambient’ racism are not considered problematic because they are seen as open to debate, because they are seen as a matter of private discussion between users and because both the circulation of contents-data and issues of ‘freedom of expression’ require that such contents remain.

In this context, the empirical question that emerges concerns the tropes of racist hate encountered in the digital domain and whether indeed there can be any actual differentiation in terms of unacceptable and ‘acceptable’ forms of racism, with the former often referring to organised racist accounts, and the latter to ‘milder’ forms of ambient, everyday or banal racism. The remainder of the article will examine this question empirically.

Methodology

This article focusses on a particular case study of an incident that occurred in Dundalk, a small town to the north of Dublin, Ireland. On 3 January 2018, an 18-year-old man of unknown nationality attacked and stabbed three men, killing one of them, 24 year-old Yosuke Sasaki. While the nationality of the accused is still not determined, there was speculation that the attack had terrorist motives. Our corpus harvested and analysed the tweets using #Dundalk and #Dundalkattacks as Twitter indicated these two as trending during the incident. We harvested in total 6,302 tweets on the day of the incident (3 January 2018), of which we analysed 1,905 original tweets (the remaining 4,397 tweets were retweets that repeated the contents of the original 1,905). In these tweets we looked at the ‘bios’ and other tweets that the user had shared in order to understand the background of the accounts that were involved in tweeting racially-loaded toxic posts.1 Indicators of organised activity were seen as a large number of racially toxic tweets, as well multiple retweets of the same racially toxic tweets on the Dundalk incident, that indicate that the account had a political agenda.

We operationalised user activity, specifically comments appearing in news sites and their Facebook pages as corresponding to non-organised public debate. Here, the analysis included all 215 user comments left on the relevant Facebook post of TheJournal.ie news site and all 409 user comments left on the original story on TheJournal.ie site. We focused on TheJournal.ie as it is the only site in Ireland to allow for user comments on its website, while it is also the most popular digital news site according to the Reuters Digital News Report of 2018. This has resulted in a vibrant comment section. There is no foolproof way of dividing the posts into those clearly coming from organised or strongly partisan right-wing political views and those coming from ordinary users: this operationalisation relies on the affordances of Twitter compared to Facebook pages of news web sites. Recent reports (Berger, 2018; Neudert et al., 2017) found that extreme right views are very prominent on Twitter, while the structure of Facebook, which still relies on personal accounts and insists on real identities, may be seen as more linked to private individuals rather than political activists. Facebook pages of news outlets tend to host debates but because of the real name policy these can be seen as taking place mainly between private individuals. On their websites, news outlets such as TheJournal.ie impose registration policies to users and moderate contents if they are reported to them, in an attempt to discourage the highjacking of the discussion by partisan interests. In this analysis, we present the general outlines of the forms that an organised version of the story takes compared to what appears to be a discussion of the same story among private individuals.

To analyse all contents, we employed a combination of grounded theory (Glaser and Strauss, 2017 [1999]) and discourse analysis (Fairclough, 2013 [1995]). Discourse analysis looks for patterns and regularities in texts and then interprets these in terms of the relationships they engender and the sociocultural practices they are associated with (Fairclough, 2103: 132). Rather than formulating hypotheses and then examining the data to see whether they support the hypotheses, we sought to generate insights from the data in a bottom up manner. We read and re-read the contents to understand the patterns that emerged and sought to order the fragmented narrative of the event as it was emerging from the tweets. In assembling this narrative, we were interested in identifying the parts told by different accounts and the manner in which they were told. Primarily, we wanted to see how the ostensibly different ‘race talk’ of organised or quasi-organised user accounts linked to right-wing political positions compared to that of private users. The findings are presented in terms of the themes emerging from this systematic reading of the contents, the story they tell and the political function they accomplish.

Ultimately, we want to outline the difficulties in the current models of thinking about and monitoring online racist speech, as fragments meld together and are re-signified in specific ways. The next section will discuss the findings of this analysis. In discussing the tweets, we have not identified any of the account handles, except for those of media or politicians, following the requirements for ethical treatment of social media data (AOIR, 2012).

#Dundalk: Re-tracing a narrative

Right-wing accounts

This part discusses the narrative as it emerged on Twitter, with an emphasis on accounts that seem to have a particular political agenda. The first reports came from local media Dundalk Democrat (9.28 am) and Louth Page, a local digital news blog. These spoke of an ‘incident’ (9.43 am), while the Irish Daily Star crime correspondent Michael O’Toole was the first to report on it involving a fatal stabbing (9.56 am). Then at 10.48 am, a pseudonymous twitter account tweeted a link to an RTE article referring to the suspect as of Syrian nationality, using the hashtag #migrantcrisis. When other mainstream media began to refer to the suspect’s nationality, accounts linked to nationalist-identitarian politics began to tweet links to mainstream media references to the Middle Eastern origins of the suspect, then purported to be Syrian, speculating on a potential terrorist attack and simultaneously accusing the mainstream media of a cover up. While mainstream media followed news values, such as immediacy, sensationalism and speculation (Lewis and Cushion, 2009; Harcup and O’Neil, 2017), some users sought to exploit these values for political purposes. Explicitly speculating on the suspect’s nationality before this or any motives for the attacks were known made the media complicit, albeit unwittingly, in furthering the extreme right-wing agenda. However, some tweets both made use and attacked the mainstream media in pursuing their political goals.

These political goals were three fold: (i) to attack the Irish political establishment but also assumed ‘leftists’, especially targeting Katherine Zappone2 and all those considered ‘responsible’ for accepting refugees; (ii) to involve foreign personalities known for nationalist or extreme right-wing views, such as Alex Jones and Paul Joseph Watson (@prisonplanet, editor of InfoWars) from the US, and Katie Hopkins and Tommy Robinson (who is now banned from Twitter) from the UK, building up transnational links; and (iii) to push a xenophobic and Islamophobic agenda within the Irish public sphere, while at the same time trying to push the public sphere to the right.

The first objective involved direct attacks on specific targets such as politicians like Zappone: ‘Zappone must be checking to see if all her Syrian children are at home reading up on transgender theory. #Dundalk’. Scoring political points involved more general attacks, for example:

Given that the guy was apparently from Syria I can see all of the leftists leaping to his defence saying that he came from a horrible background and his mental state was messed up before they tell us to love everyone and we should let more in #Dundalk.

These tweets come from accounts that appear to pursue a particular political agenda, one that is clearly xenophobic and against immigration and multiculturalism.

The second objective was achieved through tagging foreign actors and media that have a similar political agenda. Examples here include: ‘@TRobinsonNewEra Syrian refugee stabs Japanese men to death and attacks two other Irish people in Dundalk and no one mentions the word terrorism’; ‘@RealAlexJones Terrorist attack in Dundalk, Ireland. Syrian man stabs Japanese man to death and attack two Irishmen with an Iron bar!’;

‘First terrorist attack in dundalk Ireland refugee stabs 3 people killing one @TRobinsonNewEra @TheRebelTV @PrisonPlanet @KTHopkins @DVATW @DefendEvropa @mc5506 @BreitbartLondon https://www.thesun.ie/news/2000230/gardai-confirm-terror-attack-is-a-line-of-inquiry-after-man-is-killed-and-two-others-injured-in-dundalk-stabbings/’.

This tactic was successful, as eventually the ‘information’ was picked up by Watson who tweeted to his 928,000 followers (Figure 1). It is noteworthy that he links to a mainstream news article.

Figure 1 

Paul Joseph Watson, InfoWars editor, tweets about Dundalk.

The point of forging a link between the alleged nationality of the suspect and terrorism is to push an anti-immigration and Islamophobic agenda into the mainstream digital sphere. Pushing a xenophobic and anti-immigrant agenda in turn relied on direct attacks against Muslims and/or refugees and attacks against mainstream media. The first tweet by a pseudonymous account using a fake stereotypically Muslim name immediately sought to make a connection between the incident and immigration even when such connections were not made by the media: ‘#BreakingNews #MigrantCrisis #Dundalk https://www.rte.ie/news/2018/0103/930792-dundalk-death/’. Other examples include: ‘Dundalk stabbing assailant reportedly a Syrian migrant which means media blackout is imminent. Always remember #DiversityIsOurStrength’ ‘RTE News at One calling the #Dundalk Syrian man a teenager. They’ve said ‘teenager’ multiple times. This is why people despise the news media

Gardai launch ‘terror probe’ after one man is fatally stabbed and two others injured in early morning knife attack in #Dundalk https://t.co/U5iFF81Nke [LINK TO IRISH SUN] Isn’t multiculturalism wonderful especially the enrichment the #ReligionOfPeace brings?

Similarly, Paul Joseph Watson’s tweet in Figure 1 does not make the link in a direct manner, but many of those replying and retweeting actually did. For example, this was one of the replies:

But do you invite killers into your home? Not all of these refugees are, but a high percentage of that population skews towards terror/violence and a large number are males of military age. There is compassion but there is also common sense.

Linking violence to migration is ‘just common sense’. In further responses, another of Watson’s interlocutors offers a ‘solution’: [Replying to @PrisonPlanet] ‘Deport them all back to home nation @potus @realDonaldTrump #MAGA #MASA #LockHerUp #EndDACA #FakeNews’.

In a significant development, a journalist found a discussion in 4chan3 which instructed users on what kind of ‘common sense’ to mobilise (Figure 2). A user, possibly from the US, as indicated by the US flag, urges others to go on boards.ie, the most popular Irish discussion forum, and Reddit to use this event to ‘red pill normies’, i.e. to make them see ‘reality’. In the second picture, a user tweeted the exact same message.

Figure 2 

4chan user provides text to be used in other platforms.

The political agenda of these posts is made clear in this tweet: ‘If you want to know what political party in Ireland wants to act to prevent future events such as what happened today in #Dundalk, join @NationalPartyIE and follow @JustinBarrettNP’. The National Party of Ireland has a clear identitarian agenda, explained in their website as follows:

We believe in the Irish people, our right to exist as a nation and our right to defend and lay claim to our homeland. We seek an Ireland united, Irish and free. […] We stand against the corrupt and amoral establishment who push replacement level immigration which will in time completely destroy Irish nationality. We stand against the project of a federated European Superstate which is explicitly anti-national and therefore anti-Irish. And most of all we stand against the decay of national spirit at the heart of our society. (National Party of Ireland, n.d.: n.p.)

All this is amplified and spread through multiple retweets, often by the same accounts, which frequently retweet their own tweets. It is also clear in this analysis that none of these tweets qualify as ‘hate speech’ according to formal definitions even though they are stoking up division and even hate against Muslims, and more broadly against immigrants. In re-checking these tweets, we found that some users had been suspended, resulting in the removal of all their tweets, but the majority of the tweets we collected are still online. If this has been the reaction of some of those allied with right-wing politics, was the reaction of the broader publics different? This will be examined in the next section.

The broader publics

While this analysis cannot capture the entirety of the reaction across social media timelines and newsfeeds, the focus on news sites and the reactions ‘below the line’ provides an overview of the range of reactions to and discussions triggered by this event. The story centred on issues of security and purported tensions arising due to migration, in a very similar way to the story emerging from the organised accounts discussed above. In short, the main themes of the discussion revolved around: (i) the deservedness of refugees/migrants and the relevant policies in Ireland; (ii) the character and the culture of foreigners, especially Muslims; (iii) calls to deport or otherwise exclude undeserving foreigners; (iv) and links to broader political issues, specifically Brexit and the hard border with Northern Ireland, Trump and EU politics. Finally, (v) a number of comments were concerned with defining what is racist and what isn’t. This section is elaborating on these themes, using some of the most typical user comments per theme to illustrate them.

On the Facebook post of TheJournal.ie, the comment that received the highest number of reactions was this: ‘He’ll be chauffeur driven to the late late show nextsick of this country ’. This post, which received 64 likes, 2 smiley emojis and two angry emojis, alluded to the case of Ibrahim Halawa, an Irish citizen and son of the most senior Muslim cleric in Ireland, who was arrested in Egypt in 2013, when he was 17, for participating in political protests. After four years in prison, he was acquitted of all charges and upon his return to Ireland he was invited to The Late Late Show on the Irish public service broadcaster RTE. Halawa’s Irishness and alliance to Ireland were disputed among certain circles, who constructed him as ungrateful and circulated false rumours that he burned his Irish passport while in Egypt (O’Rourke, 2018). In the first instance, the focus on ungratefulness operates as a divider, separating those who ‘deserve’ the right to live in Ireland from those who don’t.

Secondly, the Facebook post of TheJournal.ie operates in a manner that criticises Irish asylum policies as being ‘too generous’ and undiscerning: ostensibly there are too few controls, no differentiation among immigrants and ‘everyone’ can get in. Claims that Ireland is too welcoming to refugees and migrants, and lacks proper border controls, are found in several comments:

‘So a guy comes over to Ireland from Syria and repays our hospitality by stabbing a man to death and injuring 2 others, oh yeah thanks lad’.

‘We are just letting everyone into this country. No vetting. Ireland as we know it is being changed utterly’.

‘Only in Ireland would he be welcomed with open arms…. Our law system is a joke…. Our border patrol & checks are lax unless you are from here of course…. We let in paedo’s & criminals to beat the band…. A serious overhaul of the system is needed’.

These views in turn were followed and amplified by calls to enforce not only stricter controls but more targeted ones, and in general to be much more discerning in terms of who is ‘allowed in’. This has clear echoes of the identitarian argument that immigration is ‘out of control’ and will lead to the ‘replacement’ of local ethnically Irish people with ‘undeserving’ non-European foreigners:

‘Here we go with the start of this crap from these people and none of them being checked before coming into the Country. It’s one of the best infiltrations of Europe ever’.

‘This young man was stabbed on his way to work, he was here making a contribution to Irish society causing no trouble. That’s the immigration we want & need. The immigration we don’t want is letting people who have no intention of contributing to society, have a values system incompatible with ours & despise us into our country. Tragic stuff’.

Several comments demand that ‘they’ be excluded or deported in order to protect the nation from both potential violence and potential loss of its ethnic purity and cultural character, repeating the same rhetoric of exclusion found in the right-wing accounts:

‘Get them all out’

‘The current Polish government, elected about 18 months ago, scrapped their imposed quota of 4,000 Syrian refugees and instead replaced it with 400 – insisting Christians only, NO Muslims. That government was elected on 38% of the vote and have since been ever more nativist nationalistic. The latest poll puts them on more than 50% – unheard of in Irish politics. Poland has a future, Ireland does not’.

Divisions between ‘them’ and ‘us’ is the topic of several comments, which mobilise essentialist, stereotypical and openly Islamophobic tropes. Most of this rhetoric is using apparently reasonable language, i.e. language that does not use racial slurs or is outright offensive but relies on what appears to be rational argumentation based on alleged ‘facts’:

‘[…] I suggest you look at examples of the influence of Islam in places like Molenbeek, Blackburn, Malmo and countless other European towns where large populations of Muslim’s are present, you will see for yourself that the idea of freedom and individuality are being strangled by the power of a 1400 year old ideology’.

‘Shipping in large numbers of unvetted individuals such as the “bearded children” from zones where Islamic Jihadists operate and want to attack any western targets will not end well. While some on here don’t articulate this well their basic fear and worry is legitimate uncontrolled immigration from dubious parts of the globe threaten our values and society’.

‘There is simply no peaceful version of a religion founded on the principles of a mass-murderer (Mohammad). So it’s important we know the motivation, and if it’s confirmed to be islamic terror, we can be certain there will be more, probably involving driving vehicles into ‘infidels’. This isn’t ‘racist’, ‘bigoted’ etc – this is being objective. Europe has an enormous enormous problem – now it’s come to our doorstep’.

Some comments draw a line connecting the event in Dundalk with other political issues, specifically President Trump’s rhetoric for border control and critiques of the alleged ‘open border’ policy of Germany and the EU.4 The anti-EU and anti-elite agenda in some of these tweets is almost identical to that pursued by the identitarian-nationalist accounts we saw on Twitter.

‘Gee, Trump does not appear so wrong about defending borders now, does he?’

‘Trump is right!’

‘Leaving the EU may now be a very real option. Staying in requires us to take a certain quota of these mentalists. Please god the Irish public sees sense or we will start hearing and seeing a lot more of this, IF we aren’t the victim the next time that is. Rip poor soul. taken for no reason. A senseless, satanic act whatever way you look at it. And now people may sit up and take heed to the fact we ARE very much a target to terrorists. This is only the beginning  our government should be hung for allowing these nut jobs in our country in the first place. They have blood on their hands.’

‘Merkel and the globalist agenda divide and conquer, drive down wages and working conditions, the results of this nonsense will not effect [sic] them behind their security gates. Wake up people and let the political classes know we have had enough of this mass migration of non EU citizens into Europe.’

These allusions to global politics, and specifically to justifications and support for the shift towards right-wing xenophobic policies, contribute to further polarization on the issue of migration and borders, making it appear as a black and white issue: you either have total control of borders or none at all, and if it is the latter, ‘anyone’ (which in fact is a reference to non-white or racialised people) will enter un-checked.

This issue of controlling who gets in makes clear allusions to race politics. When called out, some posts revert to what Titley (2016a) described as the debatability of racism: open refutation of racism, references to freedom of expression and even ‘reverse-racism’:

‘Then what’s with you dropping the racist card? Personally, I think there should be extreme vetting from ME countries, not only because of the security risk but medical reasons too, TB etc. Simply picking random refugees, economic migrants from camps in the ME and dropping them off in a fancy hotel near a small village community away from Dublin is reckless, wouldn’t you agree?’

‘Freedom of speech only when it suits the liberal agenda. It will lead to a right wing backlash. People can’t express any fear of immigration or they are labeled racist. It’s why Trump/Brexit happened’

‘As far as I can see the most hate-filled and venomous comments on here are the ones accusing other commenters of being racists, bigots, xenophobic etc. etc..’

‘Yeah everyone that says anything about Islam & Islamic terrorism is logically a racist.’ This post received 369 upvotes on the Journal.ie.

This debatability of racism acts, as Titley (2016a) has argued, as a means of silencing. At the same time however, in insisting that it is a valid debate to be had, this trope is significant because it speaks to the values at the heart of social media: that everything can be debated, that everyone is entitled to their opinion, which is considered to represent diversity in the marketplace of ideas, and that platforms are there to encourage and host such debate and the exchange of all ideas. For example, in the introduction to their Community Standards (n.d.), Facebook states that ‘Our mission is all about embracing diverse views. We err on the side of allowing content, even when some find it objectionable, unless removing that content can prevent a specific harm’. In their ‘Hard Questions’ blog, Richard Allan, VP EMEA Public Policy, elaborates further:

We are an open platform for all ideas, a place where we want to encourage self-expression, connection and sharing. […] [W]e have left in place the ability for people to express their views on immigration itself. And we are deeply committed to making sure Facebook remains a place for legitimate debate (2017: n.p.).

This analysis of user comments shows the similarity between the tropes used by those pursuing a specific political agenda and some commenters who appeared to merely mobilise ‘common sense’ discourses. While there is no way of knowing whether some of the comments were written by members of organised or quasi-organised racist/identitarian groups, the discourses used represent an emerging ‘common sense’ around security, a purported incompatibility between European/Christian and other cultural and religious values and a justification of the kinds of politics associated with Brexit, Trump and European populist identitarianism. That such discourses and their specific vocabulary were triggered and readily used by the mere suspicion of a terrorist attack, and in the absence of any concrete events of Islamic terrorism in the context of Ireland, shows at the very least a transnational contagion of racist and supremacist discourses. Such discourses, which share important similarities with the supremacist ideologies in the US discussed by among others Daniels (2009a, 2016), and their circulation as ‘common sense’ in Ireland, point to the fact that if ever there was a moment to effectively distinguish between organised and ambient forms of racism, it has now passed. In fact, as the example of the 4chan-dictated ‘common sense’ used to ‘red pill’ the Irish ‘normies’ in Figure 2, the blending of supremacist and ‘common sense’ banal racism is purposeful. While the direction or flow of such views is not entirely clear, Mudde (2010) suggests that instead of seeing right-wing views as pathological or perverse, it may be more appropriate to view them as gradations of the same mainstream attitudes that are widely shared; this is a case of ‘pathological normalcy’ in which the populist right expresses mainstream views against foreigners, elites, Islam and so on, taken to their logical extreme.

The implications of this blending and emergence of a supremacist common sense throw into question the moderating practices of social media platforms and specifically their tendency to allow the circulation of some discourses as debate but the removal of what they consider as direct attacks and hate speech. Further, the analysis points to the futility of attempts to control and contain certain forms of the extreme ‘frozen’ racism associated with organised racist groups, as their discourses have adapted to, or can also act as the source of the social media ‘common sense’ on race. The analysis shows that both ambient and organised/extreme forms co-exist and mutually reinforce one another.

Conclusions

This article began with a consideration of the relationship between digital media and racism, which was found to be ambiguous. Previous research has shown the ways in which digital media and technologies are not only linked structurally to racialised social relations but are also connected to white supremacist racist hate discourses. In order to address the latter, social media platforms developed a series of policies on content moderation and signed a voluntary code of conduct with the European Commission. The enforcement of these policies and the code relies on making a separation between what is acceptable race discourse and what is considered hate speech and will be removed. In this article we have sought to problematise such distinctions, focusing on the one between organised and ambient race discourses. We have found that very little separates the two. If anything, the ambient racist discourses appear to mobilise a wider variety of arguments against mainly non-white migrants. Both rely on essentialist understandings of culture, race and ethnicity; both posit a racial hierarchy with white Christian natives on top; both link migration to violence and terrorism. Both ultimately justify, support and sustain exclusionary and discriminatory policies and practices towards racialised others.

The analysis points to the ways in which supremacist discourses have blended with racial and ethno-nationalistic common sense. Daniels (2009a) argued that far from marginal, in fact white supremacist views are at the heart of US culture. Mudde (2010), who studied attitude surveys in Europe, echoes the same view: xenophobia and anti-immigration attitudes are at the heart of European culture, not in the extreme margins. The blending we encountered here suggests a similar dynamic in the case of Ireland, which points to a transnational contagion further evidenced not only in shared vocabularies but also in providing verbatim quotes to be used in social media, as the example in Figure 2 shows. This blending makes it very difficult to sustain arguments about acceptable race debates and unacceptable hate speech and makes clear the moral vacuity of rhetoric on diversity of opinion and ideas used by social media platforms to justify their policies.