Online Conspiracy Theories,Digital Platforms and Secondary Orality: Toward a Sociology of Online Monsters

State of the Art: The Deaf Spot of Online Conspiracism

In the last few years, a riptide of conspiracy narratives has flowed from the fringes of the internet to the center of the media system. While their threat has been swiftly acknowledged, none of the attempted remedies (from manual moderation to algorithmic suppression; from debunking to deriding) has proven effective. Part of this failure comes from our difficulty in taking these conspiratorial streams seriously and considering them as a credible cultural phenomenon, which is not only meaningful but also perfectly attuned to the attention regime of online media. As one of the reviewers of this paper aptly puts it, conspiracy narratives should be considered not only ‘as a problem – denying them, acting “bookish” or even normatively – but also as a solution. But to what problem?’ As this paper will argue, online conspirationism responds to incentives that are not only social or psychological but also specifically related to the acceleration of the media attention economy.

To explain how, we should start by questioning our own intellectual posture. Faced with stories about US Democrats trafficking children at pizza parlors, cabals of Satanic cannibals running the world, Bill Gates hiding microchips in vaccines, or the flatness of Earth, it is hard to shake off a feeling of incredulity. The flimsiness of their arguments, the incongruity of their reasoning, the amateurism of their presentation, all add to our puzzlement. From the columns of newspapers and scientific journals, commenters struggle to streamline the erratic ideas assembled by these narratives and to justify their success.

To deal with such incomprehension, research has taken two directions. Some researchers have refused to even question the sense of online misinformation and focused on its computational detection (Shu et al., 2017) and its topological spreading (Almaliki, 2019). Others have turned to the classic interpretations of conspiracism, reviving arguments about how pseudo-theories help cope with social complexity (Douglas et al., 2017, 2019; Marcus, 1999) and provide scapegoats for groups experiencing an erosion of their privileges – white middle-age upper and middle-class men in particular (Glick, 2005; Dafaure, 2020).

While insightful, both these approaches sidestep the pith of online conspiracism. One is a description without explanation that can be applied to all fads circulating online; the other is an explanation without description, germane to all forms of conspiratorial thinking, broadly defined as the mistaken view ‘that, whatever happens in society ... is the result of direct design by some powerful individuals and groups’ (Popper, 1945: 352). While excellent literature exists on conspiracy theories as a general social problem (cf., among others, Knight, 2000; Basham, 2006; Birchall, 2006b; Fenster, 2008), little research has yet focused on the features that single out online conspiracism and tune into its particular form of storytelling. The very existence of this blind spot or rather of this deaf spot (since, as I will discuss and as argued by Carmi, 2020, this is a problem of rhythm and thus of hearing rather than seeing) reveals something about its object. If online conspiracism seems hollow, it is because its core is far removed from the kind of analytic thinking that we use to question it.

This mismatch deserves attention for it points at a more general misunderstanding of internet vernaculars (Howard, 2008). At its highest with conspiracy narratives, this incomprehension concerns many other phenomena that bemuse and worry media research – from the alt-right to the manosphere, from LOLcats, to internet memes. Despite their huge popularity, these cultural expressions are still written off as youthful stupidities or extreme deviances, similarly to the way in which subcultures ‘are alternately dismissed, denounced and canonized; treated at different times as threats to public order and as harmless buffoons’ (Hebdige, 1979: 2). Yet if, following Hebdige’s lesson, subcultures should be taken seriously and understood by their opposition to mainstream hegemony, what is the nature of such an opposition in the case of online fringe cultures?

If scholars find it difficult to make sense of conspiracy narratives in particular and of online subcultures in general, I will argue, it is not because these cultural expressions are senseless, but because they appeal to a different system of sensemaking. Academic research and online conspiracism sit on opposite sides of a deep-seated cultural rift: the oral/literary divide. Our perplexity over online conspiracism thus parallels the contempt with which men of letters have long brushed off oral cultures as primitive and simple-minded (and conspiracists’ spite at established authorities reminds us of the mistrust of oral cultures towards literate technocrats; cf. Munn, 2019).

Once acknowledged, our unease in front of online conspiracism becomes a stepping-stone to understanding it as a form of digital folklore, though a degenerate one. Not to rehabilitate it, to be sure, but to fathom its success and counter it more effectively. To do so, I will summon the compelling yet somewhat forgotten notion of secondary orality.

Hypothesis: Online Vernacular Cultures as a Form of Secondary Orality

The notion of secondary orality was first introduced by Walter Ong in a book laying out the many ways in which preliterate cultures are different from (and misunderstood by) written societies. Walter Ong notices that ‘with telephone, radio, television and various kinds of sound tape, electronic technology has brought us into the striking age of “secondary orality”’ (1982: 133). Unlike primary orality, which precedes the introduction of alphabetic writing, secondary orality is a post-literate orality produced by the advent of electronic technologies and by the way in which these technologies reverse some of the cultural dynamics introduced by writing and printing.

The notion of secondary orality expands on the idea of global village by Marshall McLuhan (Ong’s PhD supervisor) and shifts its focus from the spatial to the temporal aspects of communication. While McLuhan is interested in how electronic media create a village-like proximity enabling ever more distant connections, Ong contends that the affinity between pre- and post-literate cultures is to be found in their temporal rhythms. According to Ong, the communication of electronic media is defined by the same quality that constitutes the essence of oral communication: its evanescence.

Sound exists only when it is going out of existence. It is not simply perishable but essentially evanescent, and it is sensed as evanescent. When I pronounce the word ‘permanence’, by the time I get to the ‘-nence’, the ‘perma-’ is gone, and has to be gone. (1982: 31–2)

This evanescence helps explain our difficulty in appreciating oral cultures, both primary and secondary. If the narratives of oral groups appear inconsistent to us, it is because their idea of consistency is different from ours. Because we live in a world where information and ideas are preserved as material inscriptions, our definition of consistency is based on the agreement with existing records. This agreement, however, is a nonproblem for oral cultures, whose main concern is rather to assure the durability of ideas in a world where all communication is evanescent. Because nothing survives in an oral culture if it is not remembered and repeated, successful oral narratives need to be, above all, memorable and repeatable. This need, according to cultural anthropologists, explains many features of oral cultures:

Formalised patterns of speech, recital under ritual conditions, the use of drums and other musical instruments, the employment of professional remembrancers – all such factors may shield at least part of the content of memory from the transmuting influence of the immediate pressures of the present. (Goody and Watt, 1963: 308)

Some of the features above resonate strongly with the experience of online communication. Hashtags, image templates, video challenges provide popular formalized patterns of speech in digital media; recitals under ritual conditions can be found in podcasts and video-streams; and professional remembrancers are employed in digital marketing in the form of paid commentators and social media influencers.

Nowhere, however, is the oral-like quality of digital communication more manifest than in conspiratorial narratives, which exhibit striking similarities with preliterate folklore both in style and in content. Using natural language processing techniques, researchers have shown that conspiracism and online misinformation are associated with a distinctively oral tone. Junk news, as a genre, is characterized by a style underusing typographical elements (e.g. quotes and punctuation) as well as technical and analytical words, while overusing repetition, lexical redundancy, all-caps (i.e. shouting in the web lingo) and proper names (Horne and Adali, 2017).

At the content level, conspiratorial narratives are characterized by a stunning capacity to absorb inconsistencies that reminds the homeostatic organization of preliterate societies. The lack of cogency that perplexes so many scholarly commentators is not a shortcoming for digital folklore but an essential oral-like quality that facilitates reproduction in an evanescent environment. Citing Goody and Watt’s seminal paper on orality and literacy:

In non-literate society ... the cultural tradition functions as a series of interlocking face-to-face conversations in which the very conditions of transmission operate to favor consistency between past and present, and to make criticism – the articulation of inconsistency – less likely to occur; and if it does, the inconsistency makes a less permanent impact, and is easily adjusted or forgotten. (Goody and Watt, 1963: 325)

These formal and substantial similarities suggest that online conspiracism may have its roots in a communication evanescence akin to the one that characterizes preliterate cultures. For some, examining online media through notions developed to investigate technologies such as the alphabet, print, radio, and television may seem like a stretch. However, as scholars have pointed out (see this thread in Culture Digitally: Pooley, 2015; Anderson, 2015; Sterne, 2015), it is a distinctive problem of media research that its objects evolve at an increasing speed. To cope with this changing landscape, media studies have proposed a multitude of ever-new concepts and theoretical constructs. This, however, comes with the risk of hiding long-term trends or, as this article argues, long-term swings. The attempt to explain online conspirationism through the old-fashioned notion of ‘secondary orality’ does not deny the novelty of this phenomenon and does not cover all its characteristics, but it has the advantage of reconnecting the current situation with established discussions in media studies.

Before unfolding this connection, however, it is necessary to explain how digital communication can be evanescent, despite being based on highly sophisticated technologies of archive and documentation. I will make this point first with a case study and then with a discussion of the happenings that have turned online communication into a form of secondary orality.

A Case Study: Baking Conspiracy Narratives in 4chan

4chan is an online forum created in 2003 as the English version of the Japanese imageboard 2chan. Initially dedicated to discussions about anime and manga, 4chan has quickly become the source of some of the weirdest but also most iconic internet memes (Shifman, 2013), including LOLcats (image macros of cats with misspelt captions); Rickrolling (the tricking of web users into old Rick Astley videos); and rage comics (crudely drawn cartoon faces expressing anger). 4Chan is also the home of several lively online subcultures, including bronies (the adult, and especially male, fandom of the My Little Pony franchise); Anonymous (the international anti-governmental hacktivist collective); and the alt-right (the far-right and white nationalist movement). Many popular online conspiracy narratives (notably Pizzagate and Qanon) have also originated on 4chan.

To understand how 4chan has become the Petri dish of online fringe folklore, it is crucial to consider two technical features of this platform. First, 4chan encourages its users to post anonymously (under the pseudonym ‘anonymous’, from which both the hacktivist collective and the Qanon cult took their name). Anonymity works as a liberating feature, allowing the publication of messages too outrageous for most other online and offline venues.

The second feature of 4chan is its evanescence by design (Hagen, 2018). The 74 boards of the platform are structured as lists of ‘threads’, ranked according to the most recent comment they received. Because of this ranking system, threads are immediately pushed down by new arrivals, and when falling below the 150th position are permanently deleted (after a three-day archive). To survive, threads need therefore to arouse the attention of 4chan users and spark a constant flow of comments to ‘bump them up’. The maximum number of comments per thread being limited to a few hundred (300 for most boards), however, even popular threads are rapidly closed, downgraded and deleted. In most 4chan boards, threads are extremely short-lived. Analyzing three years (July 2016 to October 2019) of data from 4chan/pol/ (the board dedicated to political discussions), Papasavva et al. (2020) found a daily average of more than 2800 threads competing for the 150 spots in the board. Bernstein et al. (2011) calculated that the median lifespan of threads in the ‘random board’ (4chan/b/) is below four minutes, with the longest-lived thread lasting a little more than six hours.

4chan is a perfect example of how, despite using communication formats (text and images) that could easily be preserved, online platforms can deliberately avoid such preservation. While there are external archives of 4chan’s discussions (e.g. 4plebs, yuki.la, 4chanarchives, the Internet Archive), their records are scarcely mobilized within the original platform. Instead, 4chan communities assure the perpetuation-without-preservation of their folklore through the classic oral technique of tireless repetition. In 4chan, the practice of summarizing and reposting threads about to reach the maximum number of comments (and thus being pruned) is called ‘baking the general thread’ (Bach et al., 2020). Investigating Pizzagate (a 4chan conspiracy story that gained traction during the 2016 US elections and alleged that senior Democratic Party officials used pizza restaurants to cover up a child sex ring), Tuters et al. (2018) describe how this practice allows not only the perpetuation but also the refining of 4chan folklore:

This process involves anons combing through previous discussion threads in order to create a new thread that compiles all the salient details on a given topic ... in addition to keeping a conversation alive after a thread has been purged, general threads are also crucial to the process of framing those discussions. (Tuters et al., 2018)

In a little more than 25 hours and through 19 rewrites, the storyline of Pizzagate was assembled, stitching together themes and characters from earlier conspiracies (see Figure 1). Through this refinement, the narrative did not become any more solid by the standards of analytical thinking but it did increase its memorability and transmissibility. In the words of 4chan’s founder:

I really don’t like permanent archives, I think that’s a pretty dramatic change to the old spirit of the site, being that it has no memory. The only things remembered are the things that become memes, things that are reposted, things that resonate with an audience and are re-posted over and over again, and endure the test of time. The only survivors are those things [that] swim upstream through the waterfall of content, just like salmon. (Christopher ‘moot’ Poole, Q&A sessions on retiring from the platform, 2015 [youtu.be/XYUKJBZuUig?t=5516, 01:35:14])

Processes of this kind may help to explain why memetic contents originating in 4chan enjoy such a large and rapid spread in other online platforms. As species grow super-resistant by breeding in an exceptionally harsh environment, 4chan’s subcultures survive by creating meaning in a platform that is engineered to be ephemeral, superficial, anonymous, and forgettable. No wonder the narratives thus created are twisted, but also no wonder that some of them spread like wildfire in online media. 4chaners call this vibrancy ‘meme magic’ and argue that its power was vindicated by the victory of their favorite candidate in the 2016 US election (Hine et al., 2017; Merrin, 2019).

OPEN IN VIEWER

Figure 1.

4chan post presenting Pizzagate mobilization as a form of meme magic (source archive.4plebs.org).

By embracing evanescence, 4chan has positioned itself as an extreme and monstrous example of the kind of secondary orality that, as discussed in the next section, increasingly characterizes all online communication – a positioning explicitly acknowledged in the ‘culture’ section of 4chan frequently asked questions.

A meme is basically an idea that is easily transferable from one mind to another. Think ‘catchphrases.’ Memes are created when a large group of users come to identify with a particular image or slogan. Their continued (mis)use will bring about the destruction of the universe. (www.4chan.org/faq)

Theoretical Framework: The Attention Economy of Online Communication Flows

As discussed above, 4chan provides a vivid example of how literacy techniques, such as listing, timestamping, and ranking, can be repurposed for objectives opposite to the ones for which they were originally introduced: not to solidify communication, but to accelerate its flow. The same media inversion can be observed, though less obviously, across all online platforms.

As noted by Ong, while primary orality comes from the lack of documentation techniques, secondary orality comes from the inversion of the same techniques and ‘is essentially a more deliberate and self-conscious orality, based permanently on the use of writing and print, which are essential for the manufacture and operation of the equipment and for its use as well’ (Ong, 1982: 133). Of course, Ong does not refer to online platforms, but to earlier broadcasting media. Since Raymond Williams’s seminal book on television flow (1974), scholars have noticed that audiovisual communication differs from written media (Newcomb and Hirsch, 1983; White, 2003). Watching television or listening to the radio encourages a floating attention and a constant flow of interruptions:

One night in Miami, still dazed from a week on an Atlantic liner, I began watching a film and at first had some difficulty in adjusting to a much greater frequency of commercial ‘breaks’. Yet this was a minor problem compared to what eventually happened. Two other films, which were due to be shown on the same channel on other nights, began to be inserted as trailers. A crime in San Francisco (the subject of the original film) began to operate in an extraordinary counterpoint not only with the deodorant and cereal commercials but with a romance in Paris and the eruption of a prehistoric monster who laid waste to New York. (Williams, 1974: 92)

If it is easy to appreciate the evanescence of radio and television, it can be difficult to recognize the same flow in the World Wide Web, a communication technology that was originally established as a literary medium. After all, the web communication protocol (the HyperText Transfer Protocol of Berners-Lee et al., 1992) was designed to implement the Memex’s dream of ‘a library of a million volumes ... into one end of a desk’ (Bush, 1945: 3). True to this inspiration, the early web was a network of texts, the ultimate achievement of a culture of literacy and archival sciences. Yet, this librarian utopia was soon unsettled by the same factor that generated the flow of radio and television: the choice of advertising as its main funding model. Contrarily to archiving, advertising generates value not from the storing of information but from the flow of attention (Lanham, 2006; Terranova, 2012; Crogan and Kinsley, 2012). By making information abundant, web technologies have diminished its value and spurred a market in which information is circulated at high speed and revenues extracted from attention flows. In the words of the first theorist of the attention economy:

In an information-rich world, the wealth of information means a dearth of something else: a scarcity of whatever it is that information consumes. What information consumes is rather obvious: it consumes the attention of its recipients. Hence a wealth of information creates a poverty of attention. (Simon, 1971: 40)

Turning the web into a market for advertising did not happen overnight and required two rounds of ‘economization’ efforts (Çalişkan and Callon, 2009). The first was the creation of a standard market for online attention with the launch of Google Ads in the early 2000s. Google Ads is characterized by a full automation allowed by two computational innovations (Lee, 2011). The first is the use of the PageRank algorithm (Rieder, 2012) for the thematic matching of the offer and demand for advertising. The second is the way in which advertising prices are ceaselessly established through an automatic system of asynchronous auctions (Mehta et al., 2007). This double automation allowed Google to handle micro-transactions unprofitable to traditional agencies, and to scale up its network to millions of buyers and sellers (thereby becoming the cornerstone of Google’s revenues).

Creating a standardized attention market accessible to all websites, Google inaugurated the so-called Web 2.0 (O’Reilly, 2007) and established advertising as its dominant business model. Still, Google Ads could not overcome the separation between the flow of advertising and the stillness of early web pages, nor control how websites organized contents and audiences consumed advertisements. Such control came with the rise of social media and their walled gardens (Berners-Lee, 2010; Helmond, 2015; Nechushtai, 2018). Structured as ‘platforms’ (Gillespie, 2010; Plantin et al., 2018), social media leverage their control on the production and consumption of digital contents to accelerate their circulation and thus increase their profits. Platforms do not function as bookstores organizing documents’ retrieval, nor as newspapers curating an orderly selection of information. Platforms serve their users a stream of news and entertainment that (unlike the programming of radio and television) is personalized and optimized through two sociotechnical affordances (Bucher and Helmond, 2017; Bucher, 2012): algorithms and interfaces.

While much has been written on how recommendation generates polarization in online communication (Sunstein, 2001; Pariser, 2011), less attention has been dedicated to the way in which algorithms accelerate attention. Recommendation systems do not just react to users’ preferences suggesting related contents, they also proactively push trending contents (Gillespie, 2016). Taking YouTube as an example, Joanna Zylinska (2020) observes that ‘instead of positioning the human against the machine, AI exponentially amplifies the knowledge shared by marketing experts with regard to our desires and fantasies, while being much quicker and much more efficient at actualising them’ (p. 71). As candidly admitted by a YouTube developer:

Many hours worth of videos are uploaded each second to YouTube. Recommending this recently uploaded (‘fresh’) content is extremely important for YouTube as a product. We consistently observe that users prefer fresh content, though not at the expense of relevance. In addition to the first-order effect of simply recommending new videos that users want to watch, there is a critical secondary phenomenon of boot-strapping and propagating viral content. (Covington et al., 2016: 193, emphasis added)

A similar push for trendiness comes from the interfaces of social media. These interfaces erode the separation between advertisement and content, not only by making them look alike through classic ‘native advertising’ techniques (Carlson, 2015), but also by providing all users with the same metrics available to commercial and political brands. By highlighting ‘vanity metrics’ such as views, likes, shares, upvotes (Rogers, 2018), online platforms encourage ‘micro-celebrity’ (Marwick and Boyd, 2011; Marwick, 2015) and ‘personal branding’ (Khamis et al., 2017). This incites users to adopt behaviors that mimic recommendation algorithms, finding and reposting trending content to perk up their followers’ attention.

Through their algorithms and interfaces, social media platforms extend the logic of flows beyond advertising, creating a ‘frictionless flow of behavioral surplus’ as Shoshana Zuboff calls it (2019: 101), enrolling contents and users in a relentless competition for an ephemeral visibility (Venturini, 2019). By turning the web from an archive of documents to a marketplace where attention is handled as a constantly renewable flow (Webster, 2014), online platforms create an evanescent communication environment not dissimilar to the one of oral cultures (Assman, 2016).

Findings: Two Oral-Like Features of Online Conspiracism

The argument unfolded above helps explain the success of conspiratorial folklore. Crafted by subcultures that embrace the evanescence of digital orality, conspiracy narratives are perfectly attuned to the attention flows of social media. Stories, such as Pizzagate or Qanon, thrive not despite but because of the qualities that make them unpalatable to literate commentators. Considered as positive rather than negative, their features are the same that assure the survival of ideas in preliterate societies. Reading from the list provided by Ong (1982: 36–56), they are additive rather than subordinative; aggregative rather than analytic; redundant or ‘copious’; conservative or traditionalist; close to the human lifeworld; agonistically toned; empathetic and participatory rather than objectively distanced; homeostatic; and situational rather than abstract. All these features equally apply to the myths of preliterate societies and to the conspiratorial cults (Campbell, 1972) of digital platforms. Two features of conspiratorial narratives – elasticity and flagrancy – deserve a special discussion as they can be directly mapped onto the two crucial qualities of oral folklore – repeatability and memorability.

The first feature – ‘elasticity’ or ‘homeostasis’ (as commonly called in cultural anthropology) – describes a process of sensemaking based on fluid adaptation rather than analytical accumulation. To appreciate the similarity between primary and secondary orality phenomena, just juxtapose the conclusion of Goody and Watt’s paper on The Consequences of Literacy (1963), with the introduction of Muirhead and Rosenblum’s book on online conspiracism (2019):

OPEN IN VIEWER

In oral societies the cultural tradition is transmitted almost entirely by face-to-face communication; and changes in its content are accompanied by the homeostatic process of forgetting or transforming those parts of the tradition that cease to be either necessary or relevant. Literate societies, on the other hand, cannot discard, absorb, or transmute the past in the same way. Instead, their members are faced with permanently recorded versions of the past and its beliefs; and because the past is thus set apart from the present, historical enquiry becomes possible. This in turn encourages scepticism; and scepticism, not only about the legendary past, but about received ideas about the universe as a whole. From here the next step is to see how to build up and to test alternative explanations: and out of this there arose the kind of logical, specialized, and cumulative intellectual tradition. (Goody and Watt, 1963: 344)

The new conspiracism is something different. There is no punctilious demand for proofs, no exhaustive amassing of evidence, no dots revealed to form a pattern, no close examination of the operators plotting in the shadows. The new conspiracism dispenses with the burden of explanation. ... This is conspiracy without the theory. What validates the new conspiracism is not evidence but repetition. When Trump tweeted the accusation that President Barack Obama had ordered the FBI to tap his phones in October before the 2016 election, no evidence of the charge was forthcoming. What mattered was not evidence but the number of retweets the president’s post would enjoy: the more retweets, the more credible the charge. Forwarding, reposting, retweeting, and ‘liking’: these are how doubts are instilled and accusations are validated in the new media. (Muirhead and Rosenblum, 2019: 3)

The fact that online conspiracism secures its durability through elasticity rather than consistency is the reason why this paper insists on calling its expressions ‘conspiracy narratives’ rather than ‘conspiracy theories’. In her monumental research on The Printing Press as an Agent of Change (1980), Elizabeth Eisenstein convincingly argued that ‘theory’, in the modern and scientific sense of the word, is only possible in a literate environment. Without writing and printing technologies assuring not only the conservation but also the reproduction of ideas in a fixed yet transferable form (as ‘immutable mobiles’, as Latour (1987) calls them), it is not possible to carry out the kind of analytical inspection and coherency check that constitutes the foundation of modern science:

A disorder previously concealed by oral presentation and piecemeal copying became more visible to copy-editors and indexers and more offensive to publishers who valued systematic routines. Classical criteria of unity, internal consistency and harmony were extended beyond orations, poems, and paintings to encompass the rearrangement of large compilations and of entire fields of study. (Eisenstein, 1980: 102)

With their call for transparency and their obsession for ‘connecting the dots’, classic conspiracy theories mimic (although in a twisted way) the analytical posture of sciences (Birchall, 2014). A posture that, as argued by Muirhead and Rosenblum (2019), cannot be further removed from the oral elasticity of online conspiracy narratives. Digital conspiratorial folklore might fulfill the same need of reassurance of classic conspiracy theories yet, unlike them, it does so not through over-rationalization but through the soothing power of repetition.

The second key feature of digital orality is its flagrancy, that is its capacity to command and retain attention by shocking its audience. In Ong’s (1982) list of oral qualities, this feature is captured by the idea of ‘agonistic tone’:

Many, if not all, oral or residually oral cultures strike literates as extraordinarily agonistic in their verbal performance and indeed in their lifestyle. ... Proverbs and riddles are not used simply to store knowledge but to engage others in verbal and intellectual combat. ... Bragging about one’s own prowess and/or verbal tongue-lashings of an opponent figure regularly in encounters between characters in narrative. (p. 43)

Enthusiastic description of physical violence often marks oral narrative. In the Iliad, for example, Books VIII and X would at least rival the most sensational television and cinema shows today in outright violence and far surpass them in exquisitely gory detail. (p. 44)

Once more, this quality can be attributed to the evanescence of oral communication. Agonism and violence are prevalent in oral communication because their flagrancy is hard to ignore or disengage from – a virtue that has not been lost on oral performers of all times, including the digital ones. Common in online settings, this communication technique is known as ‘trolling’ (Bishop, 2014; Schwartz, 2008). Online trolls hijack attention by asking silly questions; voicing outraging ideas; insulting others; violating community codes; and, in general, pushing others into controversies and ‘flame wars’ (Lee, 2005; Schachaf and Hara, 2010). Online trolling bears striking similarities to the practice of name-calling so ‘standard in oral societies across the world [that] it has been fitted with a specific name in linguistics: flyting’ (Ong, 1982: 43).

Online conspiracism is, in many ways, a form of trolling, as explicitly professed by many of its proponents, among which the users of 4chan/pol, who proudly refer to themselves as ‘shitposters’ (Phillips, 2015). Compared to traditional conspiracists, online shitposters are less interested in persuading their audience than in provoking them, in a sort of communicational ‘live action role-playing’ (Tuters, 2019). The provocative and conflict-ridden nature of conspiratorial folklore is so marked that computer scientists working with sentiment analysis have suggested that these qualities could be effectively used to identify junk news, items of which ‘tend to be shorter in length, convey less clout (expertise or confidence), appear more negative in tone (greater anxiety, sadness, or hostility), and denote lesser analytical thinking (more informal, personal, here-and-now, and narrative thinking)’ (Singh et al., 2017). Recent research has also suggested that fabricated stories tend to inspire surprise and disgust and that these high-arousal feelings can explain why these stories spread ‘farther, faster, deeper, and more broadly’ in online media (Vosoughi et al., 2018; see also Berger and Milkman, 2012; Guadagno et al., 2013).

The tendency to use outrage to retain attention is thus crucial to understanding the success of online conspiracism and to reconciling it with its ‘monstrosity’. Conspiratorial narratives thrive not because they succeed in widening the range of topics acceptable to the public but, on the contrary, because they continuously position themselves outside this ‘Overton window’ (Daniels, 2018; Peeters, 2020). By their extreme political incorrectness, conspiratorial narratives constitute forms of ‘excitable speech’ with a ‘power to injure’ (Butler, 1997), which make it difficult to ignore or forget them, even in the short-memory environment of online media. As the alt-right slogan goes: ‘conflict is attention and attention is influence’ (Marantz, 2019). Recognizing that conspiracy narratives create outrage to make themselves more memorable suggests that just ignoring them may be more effective than denouncing or debunking them, thus providing the ‘oxygen of amplification’ they need to survive online (Phillips, 2016).

Conclusion: Studying Conspiracism through Time Series and Quali-Quantitative Methods

While this paper has shed some new light on the nature of online conspiracism, it has certainly not exhausted its inquiry. The landscape of digital orality is so uncharted that its mapping could only start with a rough conceptual outline. Hopefully, this outline will encourage further empirical explorations for, as in most collective phenomena, the devil is in the details.

For example, while every platform relies on some form of attention queue management, the precise functioning of these systems is extremely varied. In 4chan, attention management is strictly chronological. In Reddit, visibility is regulated by a voting system; in Twitter, by a system of keywords (hashtags); in YouTube, by a recommendation algorithm; in Facebook by a mix of the previous. Considering the nuts and bolts of these ‘engines of order’ (Rieder, 2020) is crucial, for each of them may have major communication impacts. For instance, when in 2012 YouTube readjusted its recommendation algorithm to maximize the total watch time rather than the number of views (Meyerson, 2012), the change had profound repercussions on the popularity (and on the monetization) of videos and thus on the platform’s practices. Almost overnight, catchy short snippets became less interesting than longer and more captivating videos (Van Es, 2020).

And visibility algorithms are but one of the factors shaping online attention dynamics and conspiratorial cultures. As all forms of folklore, conspiratorial narratives cannot be separated from the situations in which they emerge and from the factors that facilitate or hinder their translation to other situations. Each instance of conspiratorial folklore should then be investigated by asking at least five research questions roughly corresponding to the components of oral performances:

The last one is a meta-question essential to recognize that the study of online conspiracism cannot be unaffected by the oral nature of its object. Sidestepping our literate biases, as we did in this paper, is necessary but not sufficient. Inquiring into digital folklore also requires finding methodological tools capable of seizing the fluid nature of online communication. With some notable exceptions (e.g. Leskovec et al., 2009; Boydstun et al., 2014), media studies have not yet adjusted to the inversion of online media we discussed and still rely on methods developed for the study of document archives rather than attention streams. Taking conspiratorial folklore seriously requires not only a conceptual reorientation but also a twofold methodological shift: towards research techniques that are (1) temporal rather than spatial and (2) immersive rather than data oriented.

Crucial in the 1970s and 1980s to investigate a media system ruled by the programming of a few broadcasters (Downs, 1972; Hilgartner and Bosk, 1988), temporal methods lost steam with the advent of digital networks and the extension they brought with them. Because of this extension, the question of the occupation of public debate began to be formulated in spatial rather than in temporal terms. Armed with methods derived from scientometrics and network analysis, media scholars (Venturini, 2012; Venturini et al., 2017) started inquiring into the clustering (Adamic and Glance, 2005; Barberá et al., 2015) and hierarchies (Barabási, 2002; Hindman, 2009) of online networks, focusing on where and by whom issues were discussed, rather than when and for how long.

Insightful as it is, this type of research is ill adapted to the temporal dynamics of secondary orality. The elements characterizing the early web (hyperlinks, keywords, and search indexes) and the methods developed for their study (Rogers, 2013, 2019) have lost some of their pregnancy and need to be combined with techniques derived from the analysis of time series and self-exciting processes (Crane and Sornette, 2008; Lorenz-Spreen et al., 2019; see also Castaldo et al., 2021 for a first installment of this research). My argument here echoes McCombs’s plea (2005) to revive the ‘agenda setting’ research (McCombs and Shaw, 1972; McCombs, 2004), but invert the direction of such research – investigating not how stories trickle down from the center to the periphery of the media system, but how they trickle up from the fringes to the mainstream of the web.

The second methodological shift circles back to the initial difficulty considered in this document: if online conspiratorial folklore is an oral phenomenon, then its nature contrasts not only with the analytical thinking typical of literate science, but also with the kind of data-oriented methods characteristic of computational social sciences (Lazer et al., 2009) and digital sociology (Marres, 2017). Leveraging these methods in this research program therefore requires using them in a way that does not deny the nature of online folklore – which raises problems similar to the ones connected to the use of audio recording in the study of preliterate culture (Goody, 2010). To make sense of a set of narratives that are as evanescent as preliterate folklore and yet based on the most sophisticated technologies of digital tracing, future research will have to rely on data-intensive methods but also combine them with a fieldwork sensitivity inspired by ethnography, in an extreme attempt of quali-quantitative research (Venturini and Latour, 2010; Venturini et al., 2015). This paper, no doubt about it, has not said the last word on online folklore.