Here I Am,
Praying to an Egyptian Frog
Exploring Political Fluidity on 4chan/pol/
Sal Hendrik Hagen
MA ThesisUniversiteit van Amsterdam
Programme: Media Studies (Research)
Referencing: MLA 8th edition
Index
Index ... 2
Abstract ... 4
Introduction: “No One Can Step in the Same River Twice” ... 5
1. Theory: Tarde and Why 4chan Is Not One Person ... 13
1.1 Tarde’s imitations and anti-structuralism... 13
1.2 A Tardean view on 4chan ... 17
1.2.1 4chan as an incubator for imitations and innovations ... 18
1.2.2 4chan and the emergence of publics ... 20
2. Method/ology: Tracing and Navigating Imitative Currents ... 24
2.1 Approach: Circulating around differently conceived wholes ... 24
2.2 Data: 4chan as an archived object ... 29
2.2.1 Full sample ... 29
2.2.2 Topic sample ... 31
2.3 Methods: Digital navigation through text mining ... 33
2.3.1 Mapping vocabulary change with tf-idf ... 33
2.3.2 Semantic similarities through word embeddings ... 38
2.3.3 Ontological associations with word trees ... 42
3. Case Study: Digitally Navigating Trump as an Imitated Object ... 44
3.1 Text mining per week: Stopping momentarily to test the waters ... 45
3.1.1 2015, week 45: “Trump is a meme-wizard” ... 47
3.1.2 2016, week 18: “A major zone of sigil magic” ... 51
3.1.3 2016, week 42: In the trenches of the Great Meme War ... 57
3.1.4 2017, week 21: “Identity first, lulz second” ... 62
3.1.5 2017, week 51: “Rationalizations that support emotionally driven conclusions” ... 66
3.2 Informed reflection: Lulzy crowds, extremist publics? ... 71
Conclusions ... 78
Works cited ... 81
Appendices ... 89
I A primer on 4chan’s infrastructure ... 89
II Scoping the amount of users on /pol/ ... 94
III Database column headers ... 97
This text references the online 4chan/pol/ archive archive.4plebs.org. Archived posts are referred to as (Anon #4). Links to the respective posts are included in the “Anonology” underneath the works cited.
Acknowledgements
This thesis is not an isolated unit created in solitude, but rather a coming-together of influences from many people around me. As such, a brief note of gratitude is fitting. My gratitude to Bernhard Rieder for the supervision of this project; I left each meeting with a newfound dose of inspiration. Thanks to my brother Abel for the beautiful cover art representing both my research process and the object of study. Not in the last place, I would like to express my gratitude to the people behind 4plebs.org for dedicating their time to the under-appreciated art of Web archiving and making this research possible. Thanks to my friends and fellow library dwellers who made the daily (and nightly) study sessions bearable, and the breaks too long. Lastly, I am most indebted to my lovely parents, who provided me with all I could have asked for and more - a happy youth, a carefree study career and a well-filled fridge.
Abstract
This research is concerned with exploring change in political sentiments on /pol/ “Politically Incorrect”, a subsection of the anonymous imageboard 4chan. Since its inception in 2003, 4chan gained infamy as a hotspot for offensive humour and trolling campaigns, coalescing into political activism on multiple occasions. While earlier activism emerging from 4chan leaned towards left-libertarianism, the imageboard has been increasingly associated with the far- and radical right after 2014, with /pol/ as the main incubator. This political fluidity, together with 4chan’s anonymity, ephemerality, and subcultural characteristics, has troubled nuanced scrutiny of the space, often leading to generalising equations of the imageboard with “Anonymous”, “trolls” or the “alt-right”. This fluidity means that making claims about 4chan’s political composition requires a perpetual analysis of the sentiments present. This text explores the theoretical and methodological challenges that arise from this task. Theoretically, holistic or metaphysical conceptions fail to grasp the heterogeneity of the imageboard’s users, but 4chan’s anonymity also complicates individualistic accounts. To that end, Gabriel Tarde’s work on imitation offers a suitable framing that shifts the focus from the collective or individual towards repeated “objects of imitation”, without disassociating these objects from subjective relations. Further, Tarde helps characterising 4chan as a particularly “innovative” space, which in turn stimulates the rapid emergence of (political) publics. Methodologically, the Tardean view is also espoused by following his claim that within complex social heterogeneity, dominant imitative patterns can be discerned, legitimising an empirical study into what 4chan “is made of” to demarcate political trends. To study such revelatory “objects of imitation”, this project proposes a set of experimental text mining methods that identify changes in word associations. To test these theories and methods, the research offers a case study on how the word “trump” provides various “points of view” into political sentiment on /pol/ within five weeks from late 2015 to late 2017. It identifies that the 2016 campaign cycle was particularly marked with semi-humorous slang, Trump became more embedded in (derogatory) vernacular, and ontological associations regarding Trump shifted from him being a “meme candidate” to a target for anti-Semitic hate speech. Together, these results suggest a “crowd-like” collective formation during the 2016 election period, and that /pol/ is not “countercultural” but rather marked by a longer presence of extremist publics. Because associations on the word “trump” changed significantly over time, the case study cements the necessity of “testing the waters” of the political currents on 4chan.
Keywords: 4chan/pol/, political fluidity, Tarde, imitation, Trump, text mining
Introduction: “No One Can Step in the Same River Twice”
I can feel a warmth deep within my blossom. It’s energy. It’s high energy. I am feeding from the iridescent pool of MAGA meme magic. Can you feel it? Once in my life I feel proud of my country. Goddamn /pol/ it feels good. (Anon #1) 9 November 2016 was a special day on /pol/, a subsection of the infamous imageboard 4chan dedicated to “debate and discussion related to politics and current events” (“Rules”). After months of trolling, conspiring and “memeing” during the election cycle, its users realised that the electoral tables slowly turned in favour of Donald J. Trump. To the delight of many anons, as 4chan’s anonymous users are nicknamed, their continuous pro-Trump, anti-Clinton content stream was seemingly not in vain. These efforts ranged from spreading pro-Trump memes (Beran; Nuzzi), constructing a conspiracy claiming the Clintons maintained a child pornography network in the basements of pizza parlours (Fisher et al.; Tuters et al.), and creating a pseudo-religious roleplay prophesising Trump’s win (Burton; Lawrence). The former business mogul formed a politically incorrect and, at times, absurd avatar matching the transgressive, comical, and right-wing sentiments common on /pol/. After the eventual results came in, anons both ironically and sincerely claimed they collectively “memed Trump into office” (Ohlheiser; Anon #2).
As a self-contained force, the political agency of these pro-Trump anons during the 2016 U.S. Elections was hard to delineate, but likely to be negligibly marginal (Faris et al.; Phillips, “Oxygen” 5-6; see appendix II). Nonetheless, a consensus arose in news media that the so-called “alt-right” -- often equated with /pol/ -- had successfully won the cultural, metapolitical battle intertwined with the elections (Phillips, “Oxygen” 5). For instance, The Washington Post headlined that “The Only True Winners of this Election Are Trolls” (Dewey). More problematically, 4chan and its malicious users quickly appeared in headlines verging on sensationalism. According to some, the anons, often presented as interchangeable with “trolls”, were “Plotting a GOP Takeover” (Stuart), had “Won the 2017 Election” (Nuzzi) or had already reached the White House (Marantz). In a much-cited piece, Paul Beran even headlined 4chan as the “Skeleton Key to the Rise of Trump”. Enticing as they might be, these claims are questionable considering the far-right political narratives could not have catapulted into the broader political agenda without the amplification by coverage of centre-left press, provoked to criticise or capitalise on the misinformation cooked up in nebulous trolling hotspots like /pol/ (Faris et al. 131; Phillips, “Oxygen” 5-6). Even more pressing, generalising statements on far-right publics -- Hillary Clinton’s “basket of deplorables”1 as the most problematic frontrunner -- recognised and unintentionally
1 Clinton described half of the Republican voters as right-wing bigots in a campaign rally speech, saying: “To just
be grossly generalistic, you could put half of Trump’s supporters into what I call the basket of deplorables. Right? The racist, sexist, homophobic, xenophobic, Islamophobic — you name it. […] [Trump] has given voice to their websites that used to only have 11,000 people — now 11 million. He tweets and retweets their offensive, hateful, mean-spirited rhetoric.” (Blow). Instead of a damning condemnation, deplorables became a shared self-adopted nomenclature for right-wing and far right actors, Donald Trump Jr. included (Firozi).
strengthened the in-group coherence of those it was meant to disavow (Phillips, This Is Why; Phillips “Oxygen”; Phillips et al.).
A main problem in the coverage of the online far right is a lack of a nuanced understanding of the online spaces these actors inhabit (Phillips, “Oxygen” 18). “Trolling scholars” Whitney Phillips, Jessica Beyer, and Gabriella Coleman, each of whom conducted extensive research on (political) collectives emerging on 4chan, emphasise such an understanding is paramount, arguing to refrain from making claims about a transgressive, dehumanising space like 4chan without first “plotting the landscape” and “safeguarding the actual record”, as not doing so risks the propagation of problematic ambiguities and sweeping statements. For instance, attributing the election of Trump to a vague notion of “4chan” or “trolls” “bestows a kind of atemporal, almost godlike power” to what is actually an “ever-evolving, ever-unstable, ever-reactive anonymous online collective” (Phillips et al.). Lamenting this dynamic, Phillips et al. call for an increased scrutiny of politicised, anonymous digital spaces:
Taking the time to map—to accurately map—the repeated, fractured, reconfiguring
mobilizations emerging from anonymous and pseudo-anonymous spaces online allows us to understand where we are and how we got here. (Phillips et al.)
As such, Phillips et al. argue that coming to a better cultural and political understanding of an anonymous space like 4chan helps to “stand up to those who attempt to hijack the narrative” (Phillips et al.).
A lack of mainstream understanding of a space like 4chan is perhaps unsurprising considering 4chan’s obscuring affordances2 and related (sub)culture. At its core, the setup of the imageboard is
straight-forward: users can post in one of 4chan’s seventy boards; digital bulletin boards dedicated to
2 The term “affordance” knows many different (mis)uses (Bucher and Helmond), but is it is understood here as
“functional and relational aspects, which frame, while not determining, the possibilities for agentic action in relation to an object” (Hutchby 444).
Figure 1: An example of a thread on 4chan/pol/. The post with the picture is the opening post (OP) of a thread, while the brown boxes underneath reply to the OP.
specific topics, like /pol/, for political discussion (also referred to as Politically Incorrect). Anyone can create a new thread by making an opening post (OP) or leave a reply in one of the 200 active threads (see fig. 1). These threads exist of simple boxes with text and an (optional) image. No account is needed to participate, meaning almost3 all posts are labelled with the default name: Anonymous. Despite this simplicity, 4chan’s affordances and interrelated cultural norms render the imageboard more arcane than its infrastructure would suggest. Firstly, 4chan’s anonymity disallows a user’s posting activity and impedes traditional classifications by age, gender, identity and so on. Further, anonymity and lax moderation allows (but does not inherently lead to) an environment where “unthinkable” and dark thoughts find their place (Nagle, Kill All Normies 14). Combined with the homogenisation of all users by the shared moniker of Anonymous, these aspects might heighten the outsider’s perception of 4chan as a single, mysterious, and potentially dangerous entity, even though the imageboard consists of a cacophony of conflicting voices (Coleman, Hacker 114-5). Secondly, 4chan’s content is ephemeral: threads can be deleted after only a few minutes4, while every thread is permanently deleted from the
servers after a few days5. This complicates information retrieval for historical research purposes, but,
more importantly, it stimulates proactive repetition and remixture of images and texts as a “locus of memory” working against the quickly moving “volume of posts and responses” (Coleman, “Net Wars”). As a corollary, a considerable time investment is required to stay up-to-date with the many intertextual and intracultural references (Beyer 48). Emboldened by anonymity and ephemerality, anons intentionally and unintentionally use transgressive “gatekeeping” practices to protect 4chan’s reputation as an underground hub, for instance through posting extremist or gory content, covering statements with (often imperceptible) irony, and adopting discourse filled with vernacular terms. Resultingly, many anons are no strangers to “discourse around ‘normies’6 and ‘basic bitches’ who ‘don’t get’ the
countercultural styles of the amoral subculture” (Nagle, Kill All Normies 107), and “[embrace] the mantle of being the ‘tastemakers’ of memetic subculture” (Milner 105). However, the ambiguities in decoding 4chan’s posts applies to anons as well, as jokesters might be unaware of the political sincerity of other anons. For example, an oft-seen claim on /pol/ is that its users are merely “roleplaying as Nazis”, even though the board formed an inspiration and recruitment zone for overt and violent
3 There is an option to insert a name when posting. However, these names have to be inserted each time and can also be co-opted by other users, meaning there is no consistency between actual users and the names. Further, using a different name than the default “Anonymous” is heavily frowned upon.
4 Bernstein et al. reported in 2011 that the lifetime of a thread on /b/ ‘Random’ had a median of 3.9 minutes and a mean of 9.1 minutes, while the shortest thread expired in 28 seconds and the longest-living thread lasted 6.2 hours. I could not find such metrics on /pol/, but a self-scraped sample of all posts from 15 January 2018 01:30:00 to 16 January 2018 01:30:00 found that an average /pol/-thread lasted 1 hour and 17 minutes, with the longest thread lasting over 23 hours and 9 minutes, and the shortest being archived after merely 13 minutes.
5 The deletion of threads differs per board. /b/ ‘Random’ does deletes threads as soon as they fall below the thread-limit, while /pol/ has an archive that stores the threads for a few days (although replying is disabled).
6 Normie is Internet slang for a conventional person who does not understand the unwritten subcultural rues of spaces like 4chan/pol/ (“Normie”).
Nazi’s (O’Brien; Phillips, “Oxygen” 23). Combined with conspiratorial distrust, wild claims are made about covert hidden political agendas of other anons - all impossible to verify (see fig. 2).
Just as 4chan’s affordances can obscure the production of cute cat pictures, they can mystify patterns and shifts in political activity (Zuckerman). Both are not without precedent on the imageboard,7
but the latter is of focus here. Historically, groups of anons managed to rapidly generate political mascots, ideas, and an activist drive (Coleman, Hacker). This is best exemplified by two movements that emerged (or are heavily associated with) 4chan: Anonymous and the alt-right. Anonymous, directly named after 4chan’s default user name, emerged from the /b/ “Random” board around 2007. Initially, its loose set of participants was mainly interested in lulz, the joy of eliciting a (preferably emotional) reaction from an unwitting or unwilling audience (Coleman, Hacker 48-9; Phillips, This Is Why 27-8). However, with “the lulz as a behavioural rudder” (Phillips, This Is Why 57), Anonymous gradually drove towards politically motivated activism, mainly focused on liberal or libertarian issues of free speech and anti-corporatism. For instance, its “members” performed DDoS attacks on MasterCard, VISA and PayPal for boycotting Julian Assange’s WikiLeaks, and later partook in the Occupy movement (Coleman, Hacker). By then, Anonymous had “left 4chan in pursuit of activist goals” (Coleman, Hacker 47), but from 2014 onwards, the imageboard gained infamy for a darker side of grassroots political mobilisation, as anons targeted left-wing liberalism, feminism, globalism and multiculturalism. The main hotspot for these sentiments was not /b/, but /pol/ (Phillips, “Oxygen” 23), created in 2011 as a “containment board” for right-wing extremists (Hine et al. 1), suggesting it was filled with politically distasteful content from the outset. Even though one of 4chan’s global rules was to “keep /pol/ in /pol/” (“/pol”), the board drew broad attention during Gamergate, a controversy
7 In the case of cute cat pictures, anons on /b/ would organise “Caturday”, denoting that Saturday was the day to
post image macros of cats, also known as LOLCats (“Caturday”).
Figure 2: A post on /pol/ indicating 4chan's intra-ambiguities. Quick glossary: MSM is ‘mainstream media’, an oldfag is a long-term 4chan frequenter, a cuck is a subordinate man, BBC refers to male genitalia, and BTFO is “blown the fuck out” (roughly meaning
to desperately lose at the hands of someone else). Derived and screencaptured from
surrounding feminist perspectives on videogames, framed by Angela Nagle as a “galvinizing issue that drew up the battle lines of the culture wars for a younger online generation” (Kill All Normies 24). Gamergate escalated into a repugnant online harassment of female journalists, with /pol/ as one of the “troll” hangouts, cementing the board as a “safe space for self-selecting misogynists and racists whose bigotries were an identity first, source of lulz second” (Phillips, “Oxygen” 23). Despite a ban on Gamergate discussion by 4chan’s founder, Christopher “moot” Poole, the event marked an “ideological crystallization” of previously disparate far-right underbellies (Phillips, “Oxygen” 24). Additionally, it prepared /pol/’s grounds just in time for Donald Trump, whose campaign and antics matched /pol/’s misogyny, racism, anti-political correctness, and absurdities. These online far-right publics were referred to as the “alt-right”,8 a “broad, and for its adherents, a usefully vague” (Nagle, “Goodbye,
Pepe”) moniker for “an amalgam of conspiracy theorists, techno-libertarians, white nationalists, Men’s Rights advocates, trolls, anti-feminists, anti-immigration activists, and bored young people” (Marwick and Lewis, 3). It is troublesome to consider both “Anonymous” and “alt-right” as coherent entities, since they merely form generalising monikers for “ever-evolving, ever-unstable, ever-reactive” collectives, but that does not negate that diametrically opposed political collectives emerged from the same space (Phillips et al.). This means that with 4chan, “nobody can step in the same river twice” (Phillips et al.).
If I follow Heraclitus’s analogy, exploring how to test the waters of the 4chan river becomes an urgent challenge, both because it could aid in identifying and making sense of the peculiar political shifts already transpired, as well as to “prepare” for possible influential collectives emerging in the future. While excellent ethnographical and anthropological studies on 4chan have been conducted (Coleman Hacker; Beyer; Phillips This Is Why; Nagle, Kill All Normies), little research has investigated the space from a data-driven approach. This is somewhat surprising, since the imageboard offers virtually unlimited data fetching all boards, circumventing the platform’s ephemerality, and opening ways to maintain and scrutinise a “data record”. Literal records have been established in the form of archives of various boards.9 While I do not argue a data-driven approach is inherently superior to
ethnographic work, it can provide empirical handles that answer the call by Phillips et al. to “plot the landscape” and “safeguard the actual record” of 4chan. Such efforts are not wholly absent: Bernstein et al. collected two weeks of data from the /b/ “Random” board, analysing the ephemerality of the various threads, and engaged in a categorisation of post content. Hine et al. analysed two-and-a-half months of /pol/ data from the summer of 2016 and traced the spread of hate speech, amongst others. Zannettou et al. classified thirteen million images from /pol/ and tracked their diffusion to other Internet platforms. However, these studies emphasise the resulting metrics in lieu of the conceptual and methodological
8 The “alt-right” was vaguely inclusive but became more loaded after the violent protests in Charlottesville,
Virginia, increasingly becoming a moniker for a hard-core strand of white nationalists - as it was first associated to when it was popularised by the white nationalist Richard Spencer. See e.g. Nagle (“Goodbye Pepe”).
9 The most coherent and accessible archive of /pol/ is archive.4plebs.org. This archive forms the main source
complexities of “extracting meaning” from the imageboard. This process of methodological reflection is crucial because a blind trust in “mechanical objectivity” can create more epistemological problems than it solves (Rieder and Röhle 72-3), potentially forming “deterministic models [that] fail to recognize the contingency, creativity, and unpredictability of movement dynamics” (Uitermark 414). Further, most of the above studies focus on momentary snapshots instead of temporal (political) change, which is a necessary lens to understand the fluid dynamics in a space like /pol/.
To that end, this thesis forms an exploration into how the employment of quantitative techniques can aid in making sense of the political variability of /pol/. It presents an empirical inquiry into a data archive of /pol/, but the conceptual and methodological challenges in this research process are given prominent stage time. A conceptual challenge is present since 4chan’s anonymity eludes characterisation of emerging collectives as traditionally “individualistic” or “collectivist”. As Coleman notes, ontological characterisation of a phenomenon like Anonymous is complex because “Western philosophy, and in turn, much of Western culture more generally, has posited the self -- the individual -- as the site of epistemic inquiry” (50). Such individualistic characterisations fail to properly conceptualise the decentralised, leaderless (at least initially), anonymous collectives emerging from 4chan (Coleman, Hacker 50). This means the dynamics of such movements cannot be reduced to specific individual participants, but simultaneously, neither should such instable groups be elevated to the status of “hyperobject”, both because this mischaracterises movements that, despite their complexity, are still made up by the actions of human subjects (Coleman, Hacker 114), and because such mystifications can lead to troublesome misnomers like “trolls” and the “alt-right”, as highlighted above (Phillips et al.). As a theoretical handle for these complexities, I draw from the work of the 19th
Century sociologist Gabriel Tarde. His social theory is somewhat outdated (Katz), but is useful here because it offers a position in between the “micro” of individuals and the “macro” of collectives by focusing on relational “imitations” between the entities of what he generalised as a “society”. More importantly, this position can be extended to the methodological challenges. A data-driven study of /pol/’s concrete objects does not offer an automated, magical answer to the above conceptual issues, necessitating methodological pondering on how to interpret 4chan’s concrete objects. Tarde’s theory on imitation allows to consider these digital results not as static “objects” or infectious “memes”, but rather as part of subjective “beliefs and desires” (Schmidt 111). Resultingly, inferring “certain regularities or laws” (Marsden 1177) in this social data can identify what ideas or issues are important to its propagators, and possibly delineate what Tarde saw as a “public”. Considering political sentiments on /pol/ are in a constant flux, I use Tarde’s thought to propose that instead of thinking of 4chan’s and its collectives in a structural, metaphysical, top-down sense (“what is 4chan?” or “what is Anonymous?”), it is fruitful to empirically identify the ontic (“what is 4chan made of?”) and from there deploy “as many subjective perspectives as possible” (Venturini, “What Is”) to infer regularities or irregularities in the data, that in turn can potentially identify broader continuities and changes in the imageboard’s political current. However, this also requires practical discussion since 4chan lacks the
usual suspects in “natively digital” (Rogers, Digital Methods) objects one can repurpose to illustrate social, cultural or political patterns: neither stable objects of personal characterisation (e.g. names, profiles, friends), nor explicit objects of affect (likes, shares, up- and downvotes) are present in 4chan’s infrastructure. However, the simple images and texts on 4chan embed complex “webs of significance” (Geertz). This research attempts to “untangle” the webs of the latter through employing text mining techniques on a particular word, trump, by drawing from Latour et al.’s reconsideration of Tarde to “digitally navigate” the properties and attributes of such an object to test whether it can identify political shifts on /pol/.
Although this text knows a conventional structure, the chapters form more of an assemblage rather than a build-up to the case study in the third chapter, as to reserve enough room for the how. The first chapter poses the theoretical framing, discussing how Tarde’s anti-structuralist consideration that “the whole is never more than its parts” allows to de-generalise and untangle the complex social trends emerging from 4chan. Further, his focus on imitations allows to think of the imageboard as space that affords particularly “innovate” imitations, creating a lively “shared knowledge reservoir” (Rieder), which in turn can stimulate the rapid emergence of (issue) publics. In the second chapter, the Tardean approach is extended towards a methodology by exploring how this theory allows to trace an “imitated object” (like the word trump) to construct “different parts of a whole” and from these points of view speculate on grander claims. It discusses a set of text mining methods that map vocabulary change and identify topics animating an issue public (mapping cosine distances and extracting tf-idf terms), the semantic similarities of a word (word2vec with t-SNE), and ontological associations (word trees of “Trump is a…”). The third and last chapter presents a case study on the temporal change in the relational properties of the word trump. It uses a dataset of nearly all posts on /pol/ since November 2013 and “navigates” five particular weeks in five discursively differing clusters. The findings identify various shifts in word associations. Firstly, the faux-playful, ironic political activity, for instance by associating Trump with a “god-emperor”, rises during the election cycle but quickly evaporates afterwards due to a suspected falling-apart of “crowd-like” formations. Secondly, both Trump’s adversaries as well as the former businessman himself become more embedded in derogatory vernacular, illustrating the fractured, antagonistic content on the board. Thirdly, Trump is first associated with being a “meme candidate”, but changes to become a target for anti-Semitic hate speech. These findings are considered both proof and an experimental exercise in arguing one has to keep “safeguarding the record” (Phillips et al.) of a space like /pol/ before imposing structuralisms.
This introduction has illustrated the most relevant aspects of 4chan’s infrastructure, but a more comprehensive understanding can be helpful. To that end, appendix I offers a detailed overview of 4chan’s features.
Note on ethics and amplification
This study deals with an online space containing racist, sexist, extremist and occasionally violent discourse. Its main concern is textual data on /pol/, meaning the most offensive or gruesome images could be avoided. Still, some posts visualised or otherwise reused in this research contain extremist language. I am well aware that the inclusion of this discourse, “even if it’s done in the service of critical assessment” also “continues their circulation, and therefore may continue to normalize their antagonisms and marginalizations” (Milner 64). Still, I share Whitney Phillips’s statement in her research on Internet trolls that “a certain amount of offensive content is necessary to the coherence and in fact the accuracy of this study” (3). This is especially the case here, as antagonisms and extremisms are unfortunate cornerstones of /pol/ and hateful words form a crucial part of the case study in the third chapter. Ultimately, a case-by-case assessment was made whether the text significantly benefited from the inclusion of hateful content.
1. Theory: Tarde and Why 4chan Is Not One Person
Despite all its twists and turns, collective existence does have a sense (even if not straightforward, unique or simple). (Venturini, “Diving in Magma” 263) One of the main conceptual challenges for this research is to infer collective meaning from /pol/ without assuming or imposing problematic structuralisms. This chapter discusses how the French sociologist Gabriel Tarde’s idiosyncratic yet inspiring theory of imitation is useful vis-à-vis this problem, and how it offers conceptual alternatives. Adopting his anti-structuralism means neither 4chan nor the collectives emerging from the space should be elevated into the stratum of metaphysics. Simultaneously, as Tarde “refuses to take the individual human agent as the real stuff out of which society if made” (Latour, “Gabriel Tarde” 4), the individual neither forms the central point of Tardean analysis, sidestepping the individualistic epistemologies so problematic in conceptualising collectives like Anonymous (Coleman,
Hacker 50). Instead, the Tardean view focuses on his somewhat problematically inclusive concept of
“imitation”: the relational influence between various entities. Tarde predicted that analysing the “objects of imitation”, indicative of the “beliefs and desires” of human subjects, could explain how “social assimilation and political consensus [can] be brought about” (Leys 279-80). This bottom-up approach is fruitful here because it eludes generalising structuralisms through emphasising multitudinous influential relations. Considering digital communications networks have increased these interpersonal influences, Tarde is “far more relevant to the conditions that have created the Occupy movement” than to the sovereign or disciplinary powers of his time (Niezen 55-6). His thought is further useful because it can conceptualise how the imageboard is particularly effective in stimulating “imitation” and “innovation”, which in turn explains the rapid emergence of (albeit very loose) publics like Anonymous. These insights can help empirical research into 4chan move beyond mere metrics and think about what political ideologies or ideas its concrete objects demarcate. Because Tarde’s thought brings quite some conceptual luggage, I first briefly discuss his most relevant theories for this reserach (section 1.1) before relating them to 4chan (section 1.2).
1.1 Tarde’s imitations and anti-structuralism
In fin de siècle France, the sociologist Gabriel Tarde put forth his theory that imitation and invention are the elementary forces of social life. He posed that social research, instead of making grand, structural claims about societies, should scrutinise these minute imitations between humans to “infer certain regularities or laws that appeared to pattern the social world” (Marsden 1177). Tarde’s thought had profound impact on questions of interaction, crowds and public opinion (Kinnunen; Katz 264). However, his work was overshadowed by the structural macro-sociology of his contemporary peer, Émile Durkheim, leaving Tarde with the dubious label of “precursor” to subsequent thinkers (Latour, “Gabriel Tarde” 82). However, Tarde’s work elicited a reappreciation in the early 21st Century, most
notably through revitalisations by Bruno Latour (“Gabriel Tarde”; “Tarde’s Idea”), Andrew Barry and Nigel Thrift. The main reason for this is that Tarde’s century-old theories contain many foresights into contemporary issues on digital information “diffusion” (Barry and Thrift 510). Crucially, his wish to trace innumerable social imitations to distinguish larger social patterns has, to an extent, become attainable with the advent of Internet technologies and the social networks built on top of these wires and protocols (Latour, “Tarde’s Idea” 157-8).
Imitation requires explanation as it is perhaps not the most elegant term (Katz 264). Tarde uses
imitation as a general concept denoting relational influence between multiple actors. He sees imitation as the very core of societies, and human subjects only social insofar that they are “essentially imitative” (Tarde, Ld’I 11).10 This imitative principle does not just to the human subject, however: it forms the
“basic mechanical fact” of “the communication or any kind of modification of a movement determined by the action of one molecule or mass on another” (Tarde, “Les Deux” 64).11 As such, Tarde poses the
role of imitation between humans in a society is “analogous to that of heredity in organic life or that of vibration among organic bodies” (Tarde, Ld’I 11). Consequently, just like repetitious influences between atoms create “atomic societies” (Tarde, M&S 28), Tarde sees human societies not as ontologically distinct to these material constellations, but rather made up of more complex and innumerable imitative interactions, in their repetitions only imagined as a distinct structure. The single entity within these societies, be it a rock, a galactic constellation or a person, is not an isolated unit, but rather an always further reducible “point of intersection or interference between diverse lines of imitation” (Barry and Thrift 513).12
As atomic societies are rarely seen as having a metaphysical essence, Tarde argues the idea of a social society is essentially a construct, a label for a whole that is merely the sum of its parts. Such a stance could invite paralysing relativism, but for Tarde, small imitations offer crucial insights into how humans and societies “behave”. He provides the following example:
When a young farmer, facing the sun set, does not know if he should believe his school master asserting that the fall of the day is due to the movement of the earth and not of the sun, or if he should accept as witness his senses that tell him the opposite, in this case, there is one imitative ray, which, through his school master, ties him to Galileo. No matter what, it is enough for his hesitation, his internal strive, to find its origin in the social. (Tarde, LS 87-8) Instead of thinking of such influences as a “preestablished harmony” or structural “universal laws” (Tarde, LS 56), Tarde argues that the thing which makes the farmer hesitate, and ultimately believe that the earth spins, is an “imitative ray” produced and propagated by other actors – his school teacher,
10 I abbreviate Tarde’s works the following way: Ld’I: Les Lois d’Imitation, LS: Les Lois Sociales, M&D:
Monadologie et Sociologie, PE: Psychologie Economique.
11 Translated to English by Barry and Thrift (514).
12 As indicated by “diverse”, Tarde shies away from generalising the imitative production across various spheres,
as he recognises that differing “imitative bodies” generate differing “societies”, taking form in wildly different subjects (Barry and Thrift 512-3).
Galileo, and so on. In other words, that which produces a human “society” is the accumulation of diverse imitations between various subjects. As such, no one can act socially “without the collaboration of a great many other individuals, most of the time ignored” (Tarde, LS 66). It is this relationality between subjects, not the individual or societal structures, which takes central stage in Tarde’s thought (Barry and Thrift 514). Unsurprisingly, Tarde had little sympathy for considerations of society as a metaphysical entity or organic system (Ld’I 68).13
The word “imitation” is somewhat problematic since it implies “voluntaristic” behaviour (Katz 264) in the sense that humans are consciously mimicking others. However, Tarde’s imitations are broader in scope and can be “willed or not willed, passive or active” (Ld’I xiv), addressing humans as “social ‘somnambules’ who believe whatever they are told, but also as ‘natural’ beings who, to some degree at least, are able to resist those ‘contagious thoughts’ by means of a critical assessment of their
truth or falsity” (Schmidt 115). To clear up Tarde’s terminological inclusivity, Katz (266) provides a
handy typology (table 1) that better stresses the differences between voluntary or involuntary imitation.
A has influenced B
Is A aware that he or she has been influential? Is B aware that he or she has
been influenced? yes no
yes Persuasion / command Imitation
no Manipulation Contagion
Table 1: Forms of influence, derived from Katz (266)
Especially “contagion” is relevant in relation to anonymous online spaces, since not the “influencer”, but the content that takes centre stage, obscuring a direct awareness of its influential propagators. Yet “contagion” might sound overly biological, potentially downplaying the role of human agency. However, Tarde was “safe from the fatal memetic tendency to model cultural evolution too closely on genetic evolution” and “displac[ing] the ‘self’ by the meme” (Schmidt 111), because, as present in the quote on the young farmer, imitation is often met with “hesitation” (Tarde, PE). As such, Tarde eschews the model of the “memetic” information as “viral” or “selfish” objects (Schmidt 103), as e.g. theorised by Richard Dawkins and Susan Blackmore. Instead, Tarde’s imitations are bound to the human subject in both conscious and unconscious manners, but always indicative of subjectivity:
What is imitated is always an idea or a wish, a judgement or a plan, in which a certain amount of belief and desire are expressed, which is the entire soul of the words of a language, the prayers of a religion, the administrations of a government, the paragraphs of a code of law, the duties of a moral system, the work of an industry, the products of an art. (Tarde, Ld’I 157)
By arguing that the “ultimate ‘objects of imitation’ are our beliefs and desires” (Schmidt 111), these objects of imitation are not just “out there”, but also influence and indicate very human urges. These “beliefs and desires” can discerned through “objects of imitation”, rendering visible internal wishes that are ontologically subjective, only “exist[ing] insofar as they are experienced by human or animal subject” (e.g pain; Searle 15). Importantly, however, just as Tarde does not place the “meme” central, neither does he foreground the rational individual; if “beliefs and desires are ontologically subjective in the sense that there has to be someone who ‘has’” them, Tarde did not rigidly hold that these beliefs and desires were exclusively and intentionally “hers or his”, but rather that they are shared -- indeed, imitated -- between various actors (Schmidt 112). Therewith, Tarde did not focus on a separation between “the intentional subject and the intentional content” but rather emphasised “the relation between different intentional subjects” (112). The Tardean view thus means an analytical shift from denoting what objects of imitation “are about”, by whom they are made, or the individual’s intention in their propagation, and towards thinking about what social relations and patterns in “believes and desires” these objects demarcate (112) - even when they only indicate “ambivalence” (Phillips and Milner) or superficial personal commitments (Niezen 55)
Especially considering the generativity of 4chan, one could wonder: how does change, originality or creativity emerge when all of “society is imitation”? (Tarde, Ld’I 74) Again, Tarde’s “imitation” is such an inclusive concept that it also denotes “innovative” dynamics. For Tarde, innovations are not purely unique instances of new ideas, but rather the coming-together of different imitations that together create an original combination. As such, imitation and imitation are bound together: “even the most imitative of all men is innovative in some respect” (qtd. in Schmidt 110-1). However, innovations denote a more generative, creative combination of influences, more than mere repetitive imitation (and this distinction is used as such in this text). Because minute innovations can be further imitated and innovated upon, but (as stressed above) do not have to voluntarily adopt a norm, this innovative agency holds the potential for a domino effect leading to both progress and rebellion (Kinnunen, 434).
The fact that Tarde’s addressed difference and variance with innovations mean the association between him and “diffusion” might not be wholly apt since the latter term suggests an adoptive uniformity in information dissemination (Rieder “Refraction Chamber”). The concept of refraction might be more fitting, proposed by Rieder (“Refraction Chamber”) to denote “the space between identical reproduction and total heterogeneity”. This echoes Tarde, as he takes this “total heterogeneity” through innovations as a basis, but notes that cumulated imitations create homogeneity:
Heterogeneity, not homogeneity, is at the heart of things. […] Homogeneity is a likeness of parts and all likeness is the outcome of an assimilation which has been produced by the voluntary or non-voluntary repetition of what was in the beginning an individual innovation (Tarde, Ld’I 71-2).
Therewith, it is only when agglomerate entities (like humans) “imitate” and “innovate” enough that a “uniformity of […] repetitions makes an apparent homogeneity, in spite of the internal complexities of each group” (Tarde, Ld’I 18). In other words, imitations are chaotic and heterogeneous, causing not mere “diffusion” but also “refraction”, with a high enough density of communication causing general imitative patterns to emerge.
1.2 A Tardean view on 4chan
Why is Tarde useful for research into the fluidity of political sentiments on 4chan? Firstly, his anti-structuralism suggests an epistemological outlook that refuses to consider the total sum of activity on the imageboard as more than its parts. The subcultural obscurity and sheer breadth of 4chan’s content make it a difficult object of study, but as with any socially complex issue, it is not “inexorably chaotic and therefore impossible to interpret” (Venturini, “Diving in Magma” 263). Though it might house complex meaning, the imageboard’s infrastructure forms a stable network; not stable in the sense of a stable constellations of, and interactions between its users, but because (as noted in the introduction) its infrastructure and mechanics are straight-forward and have remained roughly identical over time.14
4chan’s ephemerality ensures the website is “trimmed” to a limited amount of nodes, making the platform never wholly unimaginable in terms of size. Of course, the posts and the innumerable webs of meaning within the nodes are complex, but even these are not “wholly random or simply chaotic” (Coleman, Hacker 17) because they form objects of, to some extent, traceable subjective production - not mythical symbols. It is the web of “imitative rays” influencing why and what an anon posts that forms the main complexity, since the intentionality of a post is often obscured by a lack of social cues.15
This does not ask for psychoanalytical analysis, however; from a Tardean approach, the resulting productions (or: “objects of imitation”) in images or texts are not a sole window into the soul of the author, nor distinct objects (i.e. “memes”), but rather demarcate “the relation between different intentional subjects” (Schmidt 112). This allows at least some generalisation into “patterns”, as the beliefs and desires expressed through these objects of imitation are shared between subjects. Since this can inform what shared political “sentiments”16 are present on /pol/, this outlook will be espoused in
the empirical sections of this study.
14 On-site affordances do change sometimes. For instance, Christopher “moot” Poole often went back-and-forth
with the settings on IDs, which denote a unique code per user, at times forcing them board-wide, at times removing them completely (see e.g. “Complete History of 4chan”).
15 As the Internet adage “Poe’s Law” states, on in the digital sphere, it is impossible to irrefutably know the
intention behind a statement without explicit social cues like smileys (see e.g. Aigin).
16 I use “sentiment” in relation to political views on 4chan because it can both describe overt political ideologies
1.2.1 4chan as an incubator for imitations and innovations
An unfortunate lack in Tarde’s theory is that he did not stress the co-constitutive nature of media, treating communication technologies like the newspaper “less as a mediator than as an intermediary, transmitting and distributing ideas that then can work as mediators” (Wiedemann, 315).17 However,
Tarde’s theories can fit alongside accepting the agency of material technology (Latour, “Gabriel Tarde”; Latour et al.; Parikka; Sampson; Thacker). Moreover, Tarde’s thought is not devoid of acknowledging non-human agency, since he argues human societies and e.g. ant societies share similar imitative dynamics (although the former is more complex than the latter), meaning a “nature and society divide is irrelevant for understanding the world of human interactions” (Latour, “Gabriel Tarde” 82). Partly for this reason, Bruno Latour labelled Tarde’s theory of imitation not as the precursor, but the successor to the more recent Actor-Network Theory (“Gabriel Tarde” 3). The similarities are indeed striking, because Tarde proposes the influential relation between actants “create” social production, behaviour and meaning. As Latour himself attests: “agency plus influence and imitation, is exactly what has been called, albeit with different words, an actor-network” (4).
For this research, the material reframing is necessary since 4chan’s mechanics actively stimulate imitation and innovation. As stressed above, its anonymity provides a platform for unconventional thoughts that could be considered “innovative”; not in a progressive sense, but because they can transgress the boundaries of regular norms without consequence. As such, repercussion-less anonymity means one can draw from a larger pool of usually (and often rightfully) unacceptable interpersonal imitations. At the same time, the cultural norm of anonymity stimulates repetitive imitation because “being a distinct flower in the field of anonymous daisies only leads to punishment”, and to avoid this, “users have to adhere to a very strict code of behavior and ritualized language, which means they are always monitoring their own discourse patterns” (Beyer 47). Similarly, 4chan’s ephemerality stimulates particularly repetitive forms of imitation. For instance, it enforces that discussion on a specific topic is often “renewed” by copy-pasting the same text in a new opening post (OP) after an old one gets purged (see e.g. Tuters et al.). This injection both “copy-pasted” and fresh content is the basic mode of operation on 4chan, since the only way for active participation is by posting text or images; OPs even require both text and an image. “Slacktivist” upvotes, likes, shares, retweets, etc. are absent. Combined with the sheer volume of posts (and other aspects I cannot discuss in detail here18), 4chan is an environment that is particularly effective in creating “a reservoir of shared ideas,
debates, stereotypes, facts, trivia, and so on” (Rieder, “Refraction”).
17 For Tarde, imitation was a precondition to information media rather than the other way around: “if people did
not talk, it would be futile to publish newspapers. […] They would exercise no durable or profound influence; they would be like a vibrating string without a sounding board” (qtd. in Clark 307).
18 For instance, the “gatekeeping” dynamics mentioned in the introduction also stimulate the creation of a
semi-secret code language. See Beyer for more on the dynamics between 4chan’s affordances and “innovative” political activity.
This interplay between imitation and innovation is akin to Deborah Tannen’s “fixity” and “novelty” in in language: the dynamics between established knowledge and original remixture that creates “new” shared meaning (49). This “balance between the new and the expected” forms the basis of the creation of shared imitated objects (“memes”) in vernacular spaces like /pol/ (Milner 88-9). This gives some credence to the close relation between “objects of imitation” and ontological subjectivity, since these objects can only become “objectified” when its propagators possess mentally shared knowledge. As a corollary, what are usually referred to as “memes” on /pol/ (I will try to refrain from the term) are not distinct units but propagators and indicators for relational knowledge -- and possibly “beliefs and desires” -- between subjects.
That these objects might “spread outwards” (Tarde, Ld’I 140) is of particular relevance to 4chan since it has historically formed a fringe hotbed for vernacular and imagery that later saw a mainstream adaptation.19 Just as it is often unclear how “a local dialect […] spoken by only certain families,
gradually becomes, through imitation, a national idiom” (Tarde, Ld’I, 17), many are unaware of the “cultural well” of 4chan, which led Phillips to describe the imageboard as “the most influential cultural force most people didn’t realize they were actually quite familiar with” (“Oxygen” 17). These “tiny obscure subcultural beginnings” have reached “mainstream public and political life”, not only as cat pictures, but in recent years increasingly as politicised messages (Nagle, Kill All Normies 9). One of the most famous examples is Donald Trump tweeting an image of Hillary Clinton in front of a wall of dollar bills, alongside a David Star labelled with “Most Corrupt Candidate Ever!” - an anti-Semitic image that allegedly sprouted from 8chan, another fringe imageboard similar to 4chan (Smith). However, the “success” of vernacular creativity should not only be measured by whether it reached outside attention; they also form “intracultural” objects that can evoke complex meaning with simple signs within the community. Exemplary for this is fig. 3, showing a minimalist version of the anti-Semitic “Happy Merchant” meme. Even if it would be completely ambiguous to an outsider, two black lines can make /pol/ anons recall the full image and the shared anti-Semitic, conspiratorial ideas they associate it with. As one anon reacts: “I’m actually impressed with how ingrained that imagine is in my head that I can identify it almost instantly even in this minimalist form” (Anon #3). Particularly because these objects can form extremist dog whistles, it necessitates awareness (but not exposure) of seemingly fringe online spaces.
19 Many famous Internet “memes” continue to originate or popularise on 4chan. Early “successful” Internet memes
include Rickrolling, LOLCats, advice animals and rage comics, while newer memes reaching mainstream audiences include Pepe the Frog and Feels Guy. For research on the cross-platform diffusion of memes, see Zannettou et al.
It is hard to identify whether such images are mainly a corollary of a broader political climate, or if (and if so, to what extent) their spread can significantly influence the beliefs of its recipients. Nonetheless, identifying and analysing their presence on the fringe might explain larger (political) events. This claim is Tardean, exemplified by Deleuze and Guattari in A Thousand Plateaus:
May 1968 in France was molecular, making what led up to it all the more imperceptible from the viewpoint of macropolitics. […] As Gabriel Tarde said, what one needs to know is which peasants, in which areas of the south of France, stopped greeting the local landowners. A very old, outdated landowner can in this case judge things better than a modernist. It was the same with May ’68: those who evaluated it in macropolitical terms understood nothing of the event because something unaccountable was escaping. (Deleuze and Guattari 216)
Drawing the parallels between the ’68 movements and the recent surge of the online right is apt when espousing Nagle’s argument that the same bottom-up, countercultural, transgressive style of the 60s has been adopted by the online right, partly manifested on /pol/. Unfortunately, while the above peasants might have had valid reasons to stop shaking the hands of the land owners, imitations on 4chan are often based on the accumulation of “bullshit” (Tuters et al.). Regardless of whether it is used for the production of bullshit like “race realism” or conspiracies, the shared discourse and in-group behaviours related to these social imitations can change from fringe ideas to accumulate into a “high level of group trust that lays the groundwork for political action” (Beyer 47-8). This observation was underlined when a single post about the “Pizzagate” conspiracy on /pol/ snowballed into a shooting at one of the parlours claimed to be part of the Clintons’ child pornography network (Fisher et al.).
1.2.2 4chan and the emergence of publics
Despite the importance of the material infrastructure, shared discourse, or common behaviours, perhaps the most crucial collectivising force on /pol/ is the presence of specific ideas. 4chan’s effectiveness in stimulating “objects of imitaion” can as well stimulate the propagation and development of mostly
Figure 3: A minimalist version of the anti-Semitic image “Le Happy Merchant”, posted on /pol/ on 4 January 2017. Derived
mental -- i.e. ontologically subjective -- (political) ideas. In turn, this can lead to the emergence of larger
publics. Tarde’s saw publics as “constellations of individuals brought into relationships through
networks of communication” (Niezen 53), connected by “the idea or the passion” (Tarde, “The Public” 288). Collectives like Anonymous and the “alt-right”, fractured as they might be, can be conceptualised as publics because they consist of individuals that share “ideas and passions” on specific issues -- be it anti-corporativism, “men’s rights”, or white supremacism -- and find this connection through online platforms. Of course, these constellations are always “comprised of a number of different perspectives and collectives, a cacophony of voices and interests constituting multiple publics” (Phillips and Milner 166) and participation in a “technical medium” like 4chan can form a “passion” in itself (Wiedemann 314). Still, shared belief, ideas and interests in a subject matter is arguably one of, if not the basic tenet holding the dispersed contributors on 4chan together. This is particularly the case for /pol/, since, despite remaining a cacophony of antagonistic voices, political views are homogenised by “polarization effects”; “as skeptical users opt out of these communities [like 4chan], they become echo chambers of like-minded believers without exposure to any differing views” (Marwick and Lewis 18). As noted above, /pol/ might not be a pure “echo chamber”, but also a “refraction chamber” (Rieder, “Refraction Chamber”), but its characterisation as a “containment board” and its infamy for allowing extremist content suggest a degree of like-mindedness, making /pol/ a “safe space for self-selecting misogynists and racists” (Phillips, “Oxygen” 23, emphasis added). Despite their close connection, publics should not be rigidly equated to the medium they connect through, since the former are not fixed to the latter. Anonymous, for instance, moved further away from its birthplace on 4chan as it “matured” and increasingly relied on IRC chats (Coleman, Hacker 48).20 As such, there is a subtle difference between
the network and the publics it enables and co-constitutes. This is relevant because the use of generalising misnomers has historically only strengthened in-group coherence, for instance when a CNN reporter famously referred to iCloud “hackers” with: “Who is this 4chan guy?” (Vincent).21 The idea of the
public is further useful here because it can “subjectify” the presence of reoccurring topics on /pol/, and anons often self-organise in threads dedicated to specific issues.22
20 Internet Relay Chat is a protocol that facilitates group messaging. It played a big role in the organisation of
Anonymous’s activist efforts (Coleman, Hacker 17).
21 4chan as a singular entity became a meme on multiple occasions. In the early Anonymous-era, Fox News
broadcasted a report on Anonymous, presenting the loose activist group as “hackers on steroids”, a “hacker gang” and “Internet Hate Machine”, giving 4chan a national platform as a dangerous, unanimous, well-coordinated group (Phillips, This Is Why 58-9). In 2014, 4chan’s presumed entitativity was presented as even more radical when after an iCloud hack leading to a leak of nude pictures of celebrities, a CNN reporter alluded to the culprit with the question “who is this 4chan guy?” - to the delight of many anons (Vincent). Many anons gladly joke about this entitativity, as do 4chan’s moderators, as the FAQ states: “Anonymous […] is a god amongst men. Anonymous invented the moon, assassinated former President David Palmer, and is also harder than the hardest metal known the man: diamond. His power level is rumored to be over nine thousand. He currently resides with his auntie and uncle in a town called Bel-Air (however, he is West Philadelphia born and raised). He does not forgive.” (4chan.org/faq)
Inherently, the public forms a simplification that is always further reducible into smaller “parts” (Phillips and Milner 166). However, the fact that anons also adopt this simplification, as I will illustrate here, raises an epistemological question: is the social “whole” still “never more than its parts” when its participants collectivise around a shared imagination of that whole? Humans, in contrast to atoms in “atomic societies” or ants in “ant societies”, can at least partially envision what they perceive as “the ‘whole’ in which they are said to reside” (Latour et al. 604-5). Just like a French person could formulate a public perception on “what it means to be French”, a 4chan or /pol/ visitor can be influenced by their ideas of what the broader culture, politics and customs of the imageboard entails. This imagination is not rare; many anons have found joy in roleplaying -- or actually believing23 -- that their collective effort
is more than the sum of its part. For instance, when 4chan was represented as an “Internet Hate Machine” by Fox News in 2007, the generalisations incentivised /b/ anons to remix the stigma into “larger and ever-more conspicuous structures” (Phillips, This Is Why 58-9). More recently, this structuralist belief even took satirical-religious forms when /pol/ anons claimed to have harnessed magical powers in memes that foretold future events - so-called “meme magic” (Phillips et al.). Despite these imaginations of structures, theorising them as “outside” the human would be folly since the human-perceived “wholes” are no sui generis super-organisms but rather the internalisation of a perception of a whole (Latour et al. 603-4). Still, such internalised imaginations of a wholes should not be dismissed since they form extremely interesting “objects of imitation” when researching political activity on 4chan, providing insight into shared beliefs, norms, wishes, ideological associations, and so on - to which I will return later. If these are common, it can possibly legitimise characterising it as a “public”.
◆
This chapter presented a Tardean perspective on 4chan. His non-structuralist view provides a frame to avoid generalising observations on the imageboard and its users. Further, it informs to think of 4chan as a network that is particularly effective in affording innovations, creating obscure subcultural “objects of imitation” indicating shared beliefs and knowledge, but that can also “spread outwards” into the mainstream. If Tarde proposed to study minute imitations before making grander claims about a “society”, a simple question remains: how to do so? More specifically, how can one “trace” the lines of the constantly shifting political sentiments on /pol/ by analysing what mental worlds are opened by looking at “objects of imitation”? What patterns in ontologically subjective “beliefs and desires” do these objects demarcate? These questions also captured Tarde’s attention, who dreamt of possessing enough data on flows of imitations (Kinnunen 434),24 but could not “turn his intuitions into data”
23 Burton illustrates how a “self-identified active member of the alt-right” told her: “I don’t believe in God. But I
say ‘Praise Kek’ more than I’ve ever said anything about God.”
24 Tarde and even criticised the French government for lacking “data on people’s values, religious activities,
(Latour, “Gabriel Tarde” 2). Now that the digital sphere presents an abundance of social data, Latour argues that “many of the argument[s] of Tarde can be turned into sound empirical use” (“Gabriel Tarde” 2). To that end, the next chapter will explore how such an empirical outlook can take form.
2. Method/ology: Tracing and Navigating Imitative Currents
Our problem is to learn why, given one hundred different innovations conceived of at the same time - innovations in the form of words, in mythical ideas, in industrial processes etc. - ten will spread abroad, while ninety will be forgotten. (Tarde, Ld’I 140) How can empirical analysis on the content of /pol/ denote what broader political sentiments coagulate, move, or disappear? Fortunately, because /pol/ is almost entirely archived, it allows for “digital traceability” that can form a “vindication” for both Tarde’s theory and methodology (Latour, “Tarde’s Idea” 157). As such, this chapter concerns the methodological pondering on how to combine approach,
data and methods for the purposes of tracing fluid political sentiments on /pol/, informed by the Tardean
insights of the prior chapter. In the first section, approach, I argue that fractured political mobilisations can be studied by analysing multiple “parts” of a whole. This follows the argument by Latour et al. that “aggregates” in data analysis do not offer an epistemologically distinct or fully objective insight, which is particularly the case here since it can further obscure the vernacular intertextuality of the objects of study. To remain some contact with the “dirt” of the data, one can take a particular data object and “[deploy] as many subjective perspectives as possible” (Venturini, “What Is”) to “digitally navigate” attributes indicating what the scrutinised object “is made of”, which in turn potentially offers “points of view” onto shared political sentiments or “beliefs”. In the second section, data, I discuss a database of all posts from /pol/ since November 2013 and how to sampled it in relation to a particular text object. In the third section, methods, I discuss five text mining techniques that can repurpose /pol/’s text data to indicate shifts in discourse and associations, each providing its own “point of view” into how the data object is imitated. While many of the reflections and practices in this chapter are generalisable, it also forms an outline for the case study in the next chapter, which takes the word trump as a particular object to longitudinally explore political sentiments on /pol/.
For all of the methods listed below, I wrote scripts in Python, extensively using the libraries
pandas, matplotlib, sqlite3, scikit-learn and gensim.25 The full scripts are available on GitHub.26
2.1 Approach: Circulating around differently conceived wholes
How can the Tardean view inform data-driven research into 4chan’s political sentiments? Latour et al. propose a methodological approach that espouses Tarde’s argument that “the whole is equal to no more than the sum of its parts” (Clark 17). They note that, in the social sciences, two epistemological levels form cornerstones for empirical research: the individuals and the aggregates. These are used as conceptual lenses, questioning how one can conceive of a “whole” through studying “individual”
25 See pandas.pydata.org, docs.python.org/2/library/sqlite3.html, scikit-learn.org and
radimrehurek.com/gensim/
elements by starting at the macro and, from there, dive into the micro, or vice versa (591). For instance, Latour et al. note how Farkas et al. understood the stadium wave (the macro) by characterising individual reactions as either excitable, active or passive (the micro). Though fruitful in this example, humans do not spend most of their time predictably adopting group behaviour in a football stadium -- they do not just “react” but also “refract” (Rieder, “Refraction Chamber”) -- meaning such an approach is usually “incapable of understanding more complex collective dynamics” (Latour et al. 598). This problem also forms a roadblock in researching 4chan. Case in point: Bernstein et al. attempt to categorise the types of posts on /b/ (52),27 which, however insightful, simplifies its complexity by
identifying a “100% whole” by labelling its parts with altogether structuralist generalisations like “themed”, “sharing content” or “discussion”. In another example, Rolling Stone “proved” that the use of “racist and fascist terms on [/pol/] has increased since the start of Donald Trump’s presidential campaign” by showing a bar graph with the temporal increase of several hate words (Reitman). I will not argue radicalisation is absent on /pol/, or fascism is no source of concern, but Reitman’s claim is problematically limited by inferring structural change in political sentiments (“the rise of white supremacy”) from simple metrics and aggregates (word trends). The blinding particularity of this case is made even clearer because adding normalised word counts percentages, omitted in Reitman’s text, show the hate words remained stagnant.28
To prevent the problem of inferring “individuals” from “wholes”, or the other way around, Latour et al. ask: “is there a way to define what is a longer lasting social order without making the assumption that there exist two levels?” (591). The reason for this questioning is informed by Tarde’s anti-structuralist premises that “the whole is never more than its parts” and thus not on a distinct “level”, and that neither the micro is an isolated unit, but rather a whole in itself. From this perspective, concepts like the “the individual” are but indicators for a “range located within a more complex social but not individual field, where the regions beneath and beyond the individual have their own domain” (Brighenti 298). As such, the micro and the macro are not ontologically distinct, but rather different epistemologies of the parts they are assumed to consist of. As an alternative, Latour et al. pose that these viewpoints are not the only options: by not inferring an essence through micros and macros, but
properties by employing various particular views. When defining e.g. “the individual”, but also
something like “Anonymous”, once swap this essence to asks what it is made of, noting that the parts
27 Bernstein et al. categorised the OPs in a sample of /b/-data as “themed” (28%), “sharing content” (19%)
“question, advice and recommendation” (10%), “sharing personal information” (9%), “discussion” (8%), “request for item” (8%), “request for action” (7%), “meta” (5%) and “other” (6%).
28 Reitman shows, for instance, the n-word and the anti-Semitic “kike”. Appendix IV contains the normalised
counts of these words, showing the former stayed constant, and the only latter slightly increased. Further, they do not correlate with the start of Trump’s presidency. Again, I am not stating that these numbers are not problematic, but rather that they behove more in-depth empirical scrutiny.
