Influence Without Metrics: Analyzing the Impact of Far-Right Users in an Online Discussion Forum

Abstract

The study presented in this article explores the processes through which influence takes shape in eclectic online forums with few vanity metrics. Using a dataset of 7.5 million posts in the large Swedish online discussion forum Flashback, it explores who becomes influential, their strategies for appealing to the community, and others’ support of them. While it has been known that Flashback hosts far-right users and content, the current study shows that these sentiments are not fringe or obscure, but instead seemingly widely supported and influential in the forum. It illustrates that the influential users—those who are supported and acknowledged by others as important—exclusively and continuously expressed far-right ideas and displayed an embeddedness within the far-right, as well as in the forum’s culture. The study finds that despite few visible markers, many users learned to recognize influential users and their far-right content as worthy of support. In the absence of built-in functions, some users engaged in manual “liking” and “sharing” of influential users’ content via their replies, acknowledging it as a way to legitimize them. At the same time, the analysis showcased how a lack of vanity metrics countered potential echo chamber effects in the forum as disliked users—advocating progressive gender and immigration ideas—were unintentionally amplified by those who attempted to silence them. The article also discusses the role of Flashback as a platform in the proliferation of hate.

Introduction

Online discussion forums like Reddit and 4chan are eclectic—accommodating a variety of different users, interests, and ideas. Sites like these can enable openness and identity exploration (Bernstein et al., 2011; van der Nagel & van der Frith, 2015), but can also be powerful in facilitating far-right discourse as they provide anonymous, hidden-away spaces with limited degrees of accountability and moderation (Costello & Hawdon, 2018; Gaudette et al., 2020; Massanari, 2017). The fact that different types of users co-exist on sites like these means that far-right discourse created in these milieus can easily get transported to mainstream settings and audiences (Åkerlund, 2021; Daniels, 2018) or work to gradually radicalize users toward more extremist contexts and ideas (see also Burris et al., 2000; Munn, 2019). Research which focuses on the actions of influential users in such settings has the potential to expand our understandings of how these communities function and, importantly, how they may contribute both to the mainstreaming of far-right discourse and to the radicalization of users on these sites (Colley & Moore, 2020).

When it comes to exploring individuals engaged in far-right communication online, some recent efforts have recognized the role of “micro-celebrities” and “influencers” in mainstreaming far-right discourse, on conventional social media platforms like YouTube (Lewis, 2018, 2020; Maly, 2020; Winter, 2019). Less research, however, has explored more subtle forms of influence in online forums—that which functions without apparently large followings (Kilgo et al., 2016) and with a limited number of “vanity metrics” to showcase social status (cf. Rogers, 2018).

This article addresses the need for research that explores the processes through which influence takes shape in online forums. Empirically, the study focuses on the large, anonymous, and seemingly non-hierarchical Swedish discussion forum Flashback. Flashback is an important and easily accessible online setting in Sweden, thought to be used by nearly a third of Sweden’s population (The Swedish Internet Foundation, 2019). It hosts a variety of topics and users—having received attention for its users’ efforts at grassroots journalism (Fernquist et al., 2019), as well as its accommodation of hateful discourse (Blomberg & Stier, 2019; Malmqvist, 2015; Törnberg & Törnberg, 2016a, 2016b). Using a combination of network analysis and a critical discourse perspective, this article aims to understand how influence takes shape in eclectic online forums where few vanity metrics are displayed. To do this, it explores Flashback, analyzing who becomes influential, their strategies for appealing to the community, and others’ support of them through interaction and uses of the forum’s technological features.

Online Forums and the Amplification of Hate

While content production online can bypass many traditional editorial filters and gatekeepers, it is not without mediation (Thompson, 2020). Business models, technological features, and moderation practices are some of the things that impact how and what can be communicated in a given online setting (Gillespie, 2015; van Dijck & Poell, 2013).

The enablement of far-right discourse in many online forums is made by the sociotechnical features of these sites, and often under the pretense of free-speech (Marwick & Lewis, 2017). In particular, the fact that online forums are often anonymous and impose few restrictions and little content moderation is considered important for enabling hate (Costello & Hawdon, 2018; Tuters & Hagen, 2020). In the context of Reddit specifically, the voting algorithm has been found to work as an amplifier, aggregating hateful content and users (Gaudette et al., 2020; Massanari, 2017).

Such mechanisms for quantifying popularity afford some content and users disproportionate visibility and reach (Lewis, 2018; Marwick & Lewis, 2017). Technological features like Reddit’s voting mechanism, or for instance “follower,” “like,” or “share” functions on other sites, have been referred to as vanity metrics because of how they show off users’ social status within an online setting (Rogers, 2018). However, it is not the existence of certain technological features alone that contribute to the proliferation of far-right discourse in online settings. Users themselves must also actively deploy features and engage in content creation as a means of creating and spreading hate (Donovan et al., 2019; Gaudette et al., 2020). This combination of platform functions and user practices in the proliferation of hate has elsewhere been referred to as expressions of “toxic technoculture” (Massanari, 2017).

Yet, despite the importance of user practices in enabling hate, attention has seldom been paid to the users themselves in these settings, instead analyses have often focused on content at group or topic levels (Colley & Moore, 2020). Online forums may often appear unstructured and leaderless (Massanari, 2017), especially those in which it is hard to showcase status, as a result of anonymity or the absence of metrics (e.g., Bernstein et al., 2011; Kilgo et al., 2016). Exploring influence in these settings could provide knowledge not only about how these sites and their users function but also about processes of online mainstreaming and radicalization more generally (see also Colley & Moore, 2020).

Performative Influence

The so-called influencers are notably the most prominent form of influential users. They are—unlike “elites” (Mills, 1956) or “agenda setters” (McCombs & Shaw, 1972) who control public discourse through their official positions of status in, for instance, politics or media—“regular” users who have grown increasingly important and famous in the eyes of others online (Lewis, 2020). Previous research has shown how far-right influencers coordinate movement activities (Donovan et al., 2019) and, importantly, that they enable the mainstreaming of far-right discourse through their “bridging” roles between varying levels of extremist settings, users and ideas, and the mainstream (Laaksonen et al., 2020; Lewis, 2020; Maly, 2020; Marwick & Lewis, 2017; Winter, 2019). They become successful due to their knowledge and skillful use of conventional social media platforms’ technological features and vanity metrics, boosting their statuses and increasing their exposure to broader audiences (Maly, 2020; Marwick & Lewis, 2017). Studies have, for instance, shown how this is done through search engine optimization (Lewis, 2020), monetization of content (van der Vegt et al., 2020), cross-sharing content between users and channels (Laaksonen et al., 2020), and by staging content for it to appear relatable, authentic, and intimate (Lewis, 2018, 2020; Maly, 2020).

Aside from these very public influencers, there are users who have important, mediating roles in social networks without gaining celebrity status. The subtle influence that they exert can be thought to resemble that of the long-standing notion of “opinion leaders”—those who function as mediators of information to less active sections of the public through their personal relationships (Lazarsfeld et al., 1944). Such covert ways of exercising influence can potentially be of particular importance in the context of far-right discourse, due to its socially stigmatized and potentially even illegal nature (see also Crosset et al., 2019; Futrell et al., 2018). Due to its subtlety, this form of influential user could have the potential power to spread hateful ideas widely while at the same time avoid any public scrutiny and accountability (Åkerlund, 2020).

Due to the performative nature of influence and its co-creation through the support of others (Bakardjieva et al., 2018; Papacharissi & de Fatima Oliveira, 2012; see also Weber, 1922/1978), the key to gaining and sustaining both more and less public forms of influence is understanding the given social setting, its user practices, and culture (Bernstein et al., 2011; Maly, 2020; Nissenbaum & Shifman, 2017). While previous research on far-right settings has shown that influential users tend to be constitute very small share of users (Berger & Strathearn, 2013) who are more active (Kleinberg et al., 2020) and senior (Åkerlund, 2020; Scrivens, 2017), there are also things that differ between settings. Whereas influential users on far-right sites like Stormfront have been found to use overtly racist, extremist language (Kleinberg et al., 2020; Scrivens, 2017), influential users on conventional social media sites must conform to stricter platform regulations while at the same time appeal to their intended user bases (Åkerlund, 2020; Laaksonen et al., 2020; van der Vegt et al., 2020). In online forums with few vanity metrics and diverse types of users and interests, it is difficult to anticipate who becomes influential, how influence takes shape, and in turn what that says about such sites more broadly.

Data and Research Design

Case Study

Flashback was launched in May 2000 with the aim of creating a space for “divergent opinion. . . where everyone was welcome and where there was complete freedom of speech” (Flashback, 2020). The forum has around 100 moderators who monitor the approximately 15,000 new posts published each day. In total, more than 63 million posts, primarily in Swedish, have been published in the forum.

Technologically, Flashback has a layout similar to that of an early Bulletin Board System, accommodating only text-based posts. On Flashback, as stated, there is a very limited amount of publicly visible vanity metrics that could indicate users’ statuses. There are no functions for users to share or like others’ content, and besides the date of registration and the number of published posts, there are no metrics that, for instance, signal user’s “karma,” like on Reddit (cf. Kilgo et al., 2016). There are also vast possibilities for user anonymity; forum regulations even state that it is forbidden to disclose other users’ identities (Flashback, 2020). Users can engage in discussions with others without the need for pre-existing or offline social ties, simply by posting and replying to others in a given discussion thread. And posts are publicly visible and accessible also to non-member visitors of the forum.

Sampling

Flashback consists of a number of forum categories, each with hundreds of thousands up to millions of posts in the threads of their subforums. The entire Flashback forum category Politics with its 20 subforums ¹ was collected using Flashbackscraper (Kullenberg, 2018). These 20 subforums contained a total of 124,175 threads and 7,488,468 posts, at the time of data collection, between 1 and 8 January 2019. The data span from 22 May 2000, when the forum was first launched, until 8 January 2019 ² and include posts from 113,550 user accounts. To identify influential users, this study uses a two-step approach that first identifies those who are most central among all user accounts and then explores a sample of these users in depth.

Starting with the identification of central users, network measures were calculated based on user interactions through the “quote” (Swedish: “citera”) function, which is one of two ways in which Flashback users can communicate publicly. The other is through posting in the threaded comment discussions of the forum. The quote function was used to sample central users both because it is a widely utilized interaction tool—over 90% of users in this dataset had either responded or been responded to directly through it—and because it is a more deliberate form of interacting with a specific user than simply adding to the ongoing discussion threads (see also González-Bailón et al., 2013).

Centrality is defined in the article through the commonly used measure betweenness. Betweenness centrality measures the centrality of nodes in terms of how often they are on the shortest paths, the so-called geodesics, between other nodes. In theory, those with high betweenness centrality—much like opinion leaders—are thought to act as information brokers or mediators in a network, holding significant positions in the flow and control of information (Freeman, 1977; Wasserman & Faust, 1994). Because anyone can access any discussion on Flashback, users cannot be considered brokers in the sense that they have power to withhold or control the flow of information. They can, however, be thought of as mediators or bridges in the communication, through quoting, between heterogeneous, disparate users or groups of users in the forum (see also Diani, 2003; Valle & Bravo, 2018). This ability to engage various kinds of users in quoting practices is used to indicate an ability to appeal to different audiences, and in a sense an ability to bring the community together.

Betweenness centrality was calculated in the Politics forum category at large, throughout the entire period studied. With this approach, results are likely to be skewed toward more active and senior users. These users have been able to engage others over time, and not just at isolated points in time. This sampling method also counters that so-called “shitposters” who attempt to derail or troll in the forum threads would end up among the most central. Flashback has strict rules forbidding “nonsense and off-topic” (rule 0.03), “cross-posting” content between threads (rule 0.04), and “systematically publishing low-quality posts or insufficient material for debate” (rule 2.02). Content that breaches these regulations is removed, and such users risk being suspended. Therefore, it is unlikely that a shitposter would be able to post (and be quoted) over an extended period of time.

To explore a reasonably sized group of users with high levels of betweenness centrality, the sample of central users was limited to those with the 100 highest betweenness centralities overall. The qualitative analysis of influential users began with an initial coding of responses (using the quote function) to these users’ posts. Specifically, 1,200 quoting responses to the 100 users’ 10 most quoted posts (12 posts per user ³ ) were systematically read and coded into four categories: negative, neutral, positive, or those that were of a more debating nature (see Figure 1). Users who received supportive replies in at least half the sampled instances were defined as being influential (rather than just central) based on conceptualizing the influence of a user as them producing content that resonates with others (Bakardjieva et al., 2018; Papacharissi & de Fatima Oliveira, 2012; see also Weber, 1922/1978). This rendered a subsample of 22 out of the 100 central users (see highlighted area in Figure 1). It should be noted, however, that other users could of course also create content which becomes influential in the forum, and that not all content by influential users is necessarily influential. To explore the subsample of the 22 influential users, as well as the quoting responses to these users, texts were analyzed using a critical discourse perspective.

Figure 1.

Coding of sampled replies to the 100 users with highest betweenness centralities.

What and who are valued in settings with few vanity metrics is constantly performed and negotiated among users (e.g., Nissenbaum & Shifman, 2017). The critical discourse perspective focuses on scrutinizing such struggles over the power to define discursive meaning (van Dijk, 2009) and will be helpful in this paper to gain insight into who are influential, how they appeal to the Flashback community, and how their influence is negotiated in interaction with other users. More specifically, this part of the analysis entailed close readings of the 10 most replied to posts for each of the 22 influential users and all replies associated with these 220 posts, 1,241 in total. This involved analyzing how other texts and voices were incorporated (intertextuality), ideas about common perspective or knowledge (assumptions), how actors self-identified, and how they positioned others in the texts (styles and representation of social actors) (Fairclough, 2003).

Analysis

Who Is Influential and Who Is Not?

This section explores the 100 central users more closely to explain how only 22 out of these could actually be thought of as influential. Expectedly, due to the sampling method, these users are both exceedingly active and have tenure in terms of longevity in the forum. While the 100 users are periodically in central positions in terms of betweenness, it does not appear, however, that they are able to gain and then retain their positions steadily over time, as seen in Figure 2 which illustrates centrality over time for the central users. In the figure, each line corresponds to one of the 100 users, and the fluctuations indicate shifts in their levels of relative centrality—their centrality in relation to other users—calculated separately for each month in the sample period.

Figure 2.

Relative betweenness centralities over time for the top 100 users.

Furthermore, a closer look at these fluctuations in relation to their posting activity shows that there are periods where central users, despite being among the most active in the Politics forum section at large, are not able to make others quote them to any great extent. As identified elsewhere, this finding highlights the performative nature of influence in digital settings (cf. Bakardjieva et al., 2018). Because no content is algorithmically favored on Flashback (cf. van Dijck & Poell, 2013), but instead appears in a continuous stream of ongoing discussion, users must continuously produce content which others find appealing if they are to stay relevant. Even when the 100 users generate quotes, as displayed by peaks of centrality in Figure 2, this is not in itself the same thing as them exerting influence.

Moving on to the close reading of posts, most of the analyzed 100 users sometimes receive either criticism or praise, and sometimes generate debate. However, some are also more consistently supported. Close readings of posts show that all 22 influential users consistently, throughout the sample period, reiterate well-established far-right ideas, including anti-establishment, anti-feminist, and anti-immigrant discourse–notions which have been explored in detail elsewhere (e.g., Åkerlund, 2021; Krämer, 2017; Merrill, 2020; Merrill & Åkerlund, 2018; Mral et al., 2013; Wodak, 2015). While these far-right ideas also receive criticism by those quoting, this occurs seldom and seemingly only comes from those who self-identify as belonging to the targeted groups, that is, women or immigrants.

In contrast, close readings show that some of the other 78 users, while holding the best positions in terms of network centrality, seem to systematically be the target of criticism and even personal attacks. Interestingly, these criticized users all advocate progressive values—speaking in favor of women rights, immigrants and anti-racism, or socialism ideas—indicating that such ideas might be widely disapproved by various users in the forum.

For instance, one of these users, who points out that maybe some immigrants could bring positive things to Sweden after all, gets the reply that “we hoped that you had quit, you tasteless troll.” The use of “we” illustrates an assumed common ground in which the quoter seemingly purports to speak for the entire forum. Another heavily criticized user who questions “the conservative views of racists” is consequently labeled by the responder as belonging with the “muzzie huggers” who “always twist facts.” “Muslim” is used by the respondent as an interchangeable term for “immigrant” (Merrill & Åkerlund, 2018), thus placing Muslims more generally as aliens, independently of whether they have immigrated or not (Mudde, 2019). The “muzzie hugger” in this statement represents the inner enemy—someone who is not an immigrant themselves but part of the alleged political establishment that favors and protects them at the cost of the imagined true and good, native people (Sakki & Pettersson, 2016).

Another user, who raises issues of men’s violence against women, gender inequality, and prostitution, is constantly questioned and receives mocking and degrading responses like “get help,” “that you are not capable of understanding this is sad, but not surprising,” and “how do you feel about never adding anything meaningful to any conversation?” When the user criticizes men for not taking enough responsibility at home, someone points out women’s lack of intelligence and adds “go play and let the men talk,” signaling the quoter’s misogynist perspective and their self-identification with an imagined in-group of likeminded (men) in the forum. Another criticized user prompting men to take more responsibility for contraceptives is met with a conspiracy theory about how women trick random men in bars to get them pregnant and that really “women should compensate the innocent men.” While showcasing a seemingly widespread misogynist perspective among the responders, which might be partially explained by Flashback’s male-dominated user base (The Swedish Internet Foundation, 2019), it is also indicative of the often toxic masculinity of far-right discourse (Blee, 2020; Dignam & Rohlinger, 2019), and more generally of widespread gender-based hate practices online (Banet-Weiser & Miltner, 2016).

The quoters’ disapproval showcases an agreed-upon political perspective held among many users in the Politics forum category on Flashback, and with this, the “freedom of speech” promoted by the forum seems only applicable to certain political perspectives. While those not fitting into this ideal are not censored by the Flashback forum itself, they become heavily targeted by the community. This echoes previous research which shows how those who do not conform or adhere to the more or less informal rules in many online forums are reprimanded by the community (Bernstein et al., 2011; Colley & Moore, 2020; Gaudette et al., 2020; Nissenbaum & Shifman, 2017). More than that, the analysis demonstrates how hateful comments—not only toward ideas or perspectives but also toward specific users—are being passively endorsed by the forum, as moderators allow them to persist. This, as noted by Massanari (2017, p. 339), is a choice in itself by the platform that “valorizes the rights of the majority while often trampling over the rights of others.

Still, as Flashback displays no vanity metrics, it is hard for users to get an overview of how much attention is actually paid to these disliked users. Because of this, the users whom others are in a sense trying to silence through their hateful responses are in reality being magnified. Had others been aware of how their quotes put these users in privileged network positions, they might not have kept engaging with them. As such, the lack of vanity measures seem to also help amplify dissimilar content and expose users to ideas they do not necessarily agree with, therein countering potential echo chamber effects (cf. Hirvonen, 2013).

How Influential Users Appeal to the Community

Elsewhere, it has been noted that for someone to gain influence, they must share the “traditions of a given culture” (Willner & Willner, 1965, p. 84; see also Bernstein et al., 2011; Nissenbaum & Shifman, 2017). Influential users showcase their knowledge of and embeddedness in Flashback’s culture in several ways. Prominently, close readings show an extensive use of jargon and coded language. A fluency in the culture of an online forum is something that can separate those who belong from those who do not, or are too new to be familiar with its informal ways (Bernstein et al., 2011).

The jargon used by influential users is related to forum communication as such, for instance, “OT” (off topic) and “TS” (thread starter) but they also use jargon which positions them as embedded within the far-right. They, for instance, use abbreviations like “MSM” (mainstream media) and “JÅ” (far-right Sweden Democrat leader Jimmie Åkesson) and describe the seven governmental parties (besides the Sweden Democrats) as “7k” (seven-leaf clover), which is a derogatory expression to signal that these parties are all the same. All users might, however, not be familiar with these and other similar expressions. The jargon then functions as a set of code words which pass over the heads of those who do not share the same far-right affiliations while being easily decoded by other far-right sympathizers (Åkerlund, 2021; Bhat & Klein, 2020; Bull & Simon-Vandenbergen, 2014).

Relatedly, the influential users name-call—for instance, referring to Social Democrat politician Mona Sahlin as “Mona Muslim,” the liberal-right political cooperation Alliansen (the Alliance) as “Allah-iansen,” and ironically calling The Swedish National Council for Crime Prevention (BRÅ) as “the truth ministry.” Using nicknames, influential users make clear where these actors’ allegiances are considered to lie and illustrate a distancing from and “othering” of an alleged political establishment. Thus, using these malicious nicknames, influential users not only signal their knowledge of far-right lingo, but they also enable community building—something which has been identified as important for influential far-right users elsewhere (Åkerlund, 2020; Scrivens, 2017). Like previous research of other settings have shown, the far-right in-group is seldom sharply delineated (Bull & Simon-Vandenbergen, 2014; Törnberg & Wahlström, 2018). Still, the distancing from established politics and governmental agencies consequently works as an inward-oriented activity to construct a dichotomous us versus them, wherein the Flashback community is positioned as ideologically opposite “the establishment” (see also Wodak & KhosraviNik, 2013).

Influential users’ knowledge of the forum’s ways can also be seen more quantitatively, in how they provide material for debate. The analysis shows that influential users’ posts are longer than other users’—their original posts are on average 97 words, compared with 76 words for other users. They also rely more heavily on intertextual references than other users—28% of their original posts, compared with 16% of posts by other users, contain URLs. This elaboration of their ideas and use of references show familiarity with the ways of the forum, wherein users are quickly reprimanded if they attempt to make unsubstantiated claims. Closer examination of some of these external links show that they often reference both mainstream and alternative news media outlets. However, it is not with the purpose of bringing different perspectives into the debate, but rather a decontextualized and selective use as a means to further their own far-right agendas (Merrill, 2020; Merrill & Åkerlund, 2018).

The influential users vary somewhat in their levels of far-right radicalism. However, many still often present their arguments in (pseudo-)rational manners to justify certain action or make causal connections (Angouri & Wodak, 2014). For instance, with reference to a far-right news source, one influential user questions why Black people, if they are equally intelligent as white people, so rarely have won any science awards throughout history. In another example with reference to a mainstream news item, an influential user questions whether men in general really are “the issue” when it comes to sexual assault, seeing as other countries (with stricter immigration policies) have fewer reported rape cases. Sometimes these othering representations make dissociated comparisons, placing groups or problems in opposition to each other. For instance, arguing that “if X immoral Muslim tradition is allowed in Sweden, then why not, with the same logic, allow sexual assault?”

Influential users might not have to “code” their far-right utterances to avoid sanctions by the public (cf. Wodak, 2015) or platform regulations (cf. Åkerlund, 2020) considering the widespread support of far-right ideas in the forum, and the fact that forum regulations state that “[d]erogatory statements against one or more designated groups of people are allowed, provided the criticism is factual and justified” (Flashback, n.d.). Still, a (pseudo-)rational approach is a strategy to distance oneself from the most blatantly expressed hate while still showing a clear commitment to far-right ideas (Bull & Simon-Vandenbergen, 2014; see also Klein & Muis, 2019). The resonance with different levels of radical users is also shown in the quoted replies, where some quoters use little to no slurs, engaging in somewhat well-mannered complaining about the state of the country, whereas others’ utterances are more explicitly hateful or racist descriptions of things like “primitive immigrants” and “the Jewish conspiracy.”

Quoting Users’ Co-Construction of Influence

Often, those who quote influential users express a familiarity with them. Likely, this is in part due to influential users’ active posting behaviors and tenure in the forum. Other users even seem to know the types of opinion they have expressed previously, for instance, pointing out when they have changed their minds or positions. Quoters further show their acquaintance with influential users by mentioning them by name or even through nicknames, indicating a sense of community among the members (Bowman-Grieve, 2009). It is also not uncommon that influential users quote each other. In fact, 13 of the 22 influential users are among the 100 most prolific quoters, and all but one are among the top 400. While less formal, this can have some of the same marketing advantages as the cross-appearances identified by Laaksonen et al. (2020) among far-right influencers on YouTube.

There are also instances where the influential users are listened to and publicly acknowledged as important, through legitimizing responses—for instance, when others comment “this is so good and important” and “many have waited for you to comment, what do you think?” or “this is EXACTLY how I also think.” It becomes especially personal in those instances where the influential users are named, for instance, “I don’t know if I’ve told you before [username], but I really like you” or “Thanks for being you, [username].” Sometimes, quoting users make it clear that influential users put into words what they themselves cannot, thus positioning them as experts: “thanks to this post there are certain things I realised that I did not think about before” and “you should be in front of a large crowd sharing this! It would make people actually start to think about things.”

Despite Flashback not having any technological features for “liking” or “voting,” quoters circumvent this by commenting their likes. They also seem to realize that this promotes the account in question, as one quoter notes: “This deserves attention, could be the best post on Flashback ever.” Moreover, with a degree of ephemerality in the forum as posts get lost in the stream of new content, and with no opportunities for users to pin or bookmark specific posts, users must themselves remember and save content that they like (Bernstein et al., 2011). There are examples when quoting users have put forth posts as good examples, manually “sharing” them together with messages like “many here remember the unforgettable text you posted, fantastic.” There is even an example of a user quoting something posted several years ago by an influential user, which is quoted as a “powerful observation” for a current discussion.

However, similarly to how Cocker and Cronin (2017, p. 462) see YouTube influencers as a “refraction point for collective self-admiration” for others, and how Gustafsson and Weinryb (2020, p. 6) consider social media users more generally to be the “subject/follower of their own actions,” those quoting also focus on self-representation. The quote function lets responding users interconnect their content with that of the influential users (see also Angouri & Wodak, 2014). Recognitions of influential users and their content through quoting can act as ways for the quoting users to signal their knowledge of the forum and its important actors, and therein their own belonging (Huffaker, 2010). Having in mind that users sometimes develop relationships online for status (Li, 2011), potentially these are efforts for quoting users to themselves gain influence within the forum.

Furthermore, while quoting users acknowledge those who are influential and sometimes indicate seemingly genuine affection and admiration for them and their ideas, quoters also focus on discussing the issues at hand. It might be so that in a setting with few vanity metrics, where influence is not clearly visible, gaining status might be less of a focus for some, in favor of the discussions themselves.

Discussion

The study presented in this article explored processes through which influence takes shape in eclectic online forums with few vanity metrics, using the large Swedish online discussion forum Flashback as a case study.

The study demonstrated that some users were largely uninfluential even though they were quoted by a wide range of others. This could perhaps be understood in light of the findings of Buttliere and Buder (2017), who, when studying online discussions, found that people more often replied to content they disagreed with compared to that which they agreed with. Correspondingly, some users in this dataset acquired the best positions in terms of network centrality not because they were influential, but rather because of how other users disagreed with them and their content. Even though Flashback is not a dedicated far-right forum, but instead an eclectic site used by nearly a third of Swedes, the analysis showed that these widely disliked users all advocated progressive gender and immigration ideas. Relatedly, the influential users, those who were supported and acknowledged by others as important, exclusively and continuously expressed common far-right views, including anti-establishment, anti-feminist, and anti-immigrant discourse.

Influential users can inform about community culture more broadly and about processes of far-right mainstreaming and radicalization (see also Colley & Moore, 2020). While it has been known that Flashback hosts more or less extremist users and content, this article shows that these sentiments are not fringe or obscure, but instead seemingly widely supported and influential in the forum. This is notable because an online community can have impact on individual users’ norms and values (Bowman-Grieve, 2009). This culture of anonymous unaccountability and relatively unmoderated freedom of speech promotes hate, if that is what the community wants (Massanari, 2017). With previous research identifying how far-right discourse on Flashback also seems to reach those not interested in such content (Åkerlund, 2021), the prominence of far-right users’ ideas within the forum risks affecting less extremist users, enabling their radicalization, as well as the mainstreaming of far-right ideas.

Flashback, contrary to platforms like Facebook, Twitter, or even Reddit, has no functions for users to demonstrate their own social statuses within the site, nor to know that of other users. This absence of vanity metrics means that the importance of users on Flashback becomes based on reputation. As such, influence on Flashback naturally works differently than on sites where users have “followers” or “channels” that others can subscribe to out of interest for that particular user (cf. Cocker & Cronin, 2017). First, interactions in the forum are content-driven and users connect publicly through discussion threads, where no content or user is algorithmically favored (cf. van Dijck & Poell, 2013). Instead, one post simply appears after the previous in the discussion threads. This fluidity of forum discussions means that influence becomes unstable, and users must continuously produce content that gets supported by others to stay relevant. Second, in the forum, groups of users form because of their joint engagement in discussion threads (see also Bruns et al., 2016; Rambukkana, 2015). Thus, it is perhaps unsurprising that users who quoted others often focused on discussing the issues at hand.

Nevertheless, the analysis showed that despite few visible markers, many users had learned to know and recognize influential users and their far-right content as worthy of support. Without any built-in functions, some users instead engaged in manual “liking” and “sharing” of influential users’ content via their replies, acknowledging it as a way to promote and legitimize them. However, with self-centered motives often in play online, this acknowledgment could potentially be quoting users’ ways to increase their own statuses within the forum (see also Gustafsson & Weinryb, 2020; Huffaker, 2010; Li, 2011). With influence being of such an informal and anonymous character, it requires much in terms of familiarity with the community to recognize, and thus, makes it very difficult to detect as an outsider (see also Colley & Moore, 2020; Kilgo et al., 2016).

Hate is both actively and passively endorsed by the forum. Having the capacity to make certain users, views, and features more or less visible, Flashback’s role in enabling the spread of far-right discourse should not be understated. Although the analysis showed how a lack of vanity metrics countered potential echo chamber effects in the forum—as disliked users were amplified beyond what would have been likely had users been able to downvote content, like on Reddit (Gaudette et al., 2020; see also Massanari, 2017)—providing a limited number of vanity metrics is convenient for Flashback and other forums like it because it obscures what types of content and users are most highly appreciated.

While Flashback has around 100 moderators, possibilities for users to produce hateful content are enabled by the extensive, if not impossible, task for these moderators to monitor approximately 15,000 new posts published each day. In addition, Flashback users cannot delete their own accounts, posts, or threads, and with this, the likelihood that hateful content persists in the forum is high. Similarly to Reddit (Massanari, 2017) and contrary to niched online forums, Flashback addresses a range of interests and is easily accessible and usable. While 113,550 users had posted in the Politics section in the Flashback forum during the sampling period, considerably more were likely the lurkers and non-registered visitors to the site over the same period of time. This ease of use, accessibility, and widespread adoption allow for hateful ideas expressed in the forum to easily reach outside of fringe far-right circles.

By extension, Flashback could work both as a funnel or gateway toward more radical settings (Åkerlund, 2021; see also Burris et al., 2000) and as an amplifier of hateful content, helping to expand and mainstream far-right discourse in Sweden (Blomberg & Stier, 2019; see also Merrill & Åkerlund, 2018). Its sociotechnical features make it a toxic technoculture (Massanari, 2017) that reprimand users who do not conform to hateful ideas (see also Malmqvist, 2015), its regulations promoting anonymity and freedom of speech become a veil which disguises hateful content and users (see also Daniels, 2009; Gerstenfeld et al., 2003; Kilgo et al., 2018), and results in difficulties in detecting and scrutinizing far-right influence.

Limitations and Suggestions for Future Research

This study is subject to limitations which could be addressed in future studies. It should be noted that this article does not explore the resonance of users with low network centrality, other centrality measures than betweenness, or shifting views of users over time. Furthermore, there is only a limited amount of information about how messages are received and possibly internalized by quoting users, as well as the intentionality and strategy involved in influential users’ posting behaviors. Additional insight could be gained by interacting directly with influential users and quoters to better understand their thoughts and actions. Future research should also continue exploring user dynamics and subtle user hierarchies in other far-right settings.