If enough people push it, it is very hard for the Arbitration Committee to go the other way.
—A former member of the Arbitration CommitteeIf ten people are in dispute with one person, the easiest way to soothe the dispute is to censure the one person rather than the ten…. The Arbitration Committee wanted to end the dispute, and wanted the fight to be not happening. It was easier to restrict one person rather than one hundred.
—An editor who was sanctioned by the Arbitration CommitteeINTRODUCTION
On its twenty-second birthday, Wikipedia is by all counts the most ambitious and successful attempt at gathering knowledge since the Encyclopedia of the Enlightenment. English Wikipedia, the largest and oldest of Wikipedia’s language editions, prides itself on counting more than six million pages, four million registered users, and a billion edits.Footnote 1 Wikipedia’s enormous success is even more remarkable given that it relies entirely on the collaborative efforts of individuals who work as volunteers to create the “free encyclopedia.” The world’s foremost encyclopedia is written by whoever wants to contribute to it, thus giving credence to Wikipedia’s slogan: “anyone can edit.” These individuals, who call themselves “editors,” form a rather small group, with about ten thousand of them contributing regular “edits,” of whom one or two thousand make significant contributions to the encyclopedia. Wikipedia editors share a strong egalitarian and collaborative ethos (Reagle Reference Reagle2010, 125). They believe that the accretion of edits through amendments and corrections brings about the best possible outcome in terms of the encyclopedia’s content, and that each voice counts in this process. In contrast to other encyclopedias that select world-famous specialists who take full responsibility for one or more articles, Wikipedia relies on nonspecialist volunteer editors who draft the articles collectively. Due to this collaborative work, Wikipedia is the modern and electronic version of a medieval palimpsest, one that comes close to Encyclopedia Britannica in terms of accuracy (Giles Reference Giles2005).
However, the capacity of editors to leave their mark on Wikipedia depends on their ability to reach a consensus on a specific topic. Consensus has been described as the “preferred method of making decisions at Wikipedia” (Reagle Reference Reagle2010, 110). If they do not manage to build consensus during the editing process, there is a risk that the editors will “repeatedly override each other’s contributions,” thus leading to what they call “edit warring.”Footnote 2 Faithful to its collaborative ideal, Wikipedia’s philosophy is to resolve these disputes organically through discussion and compromise resulting in consensus. Disputes are aired in specific spaces, such as “Talk Pages” or “Dispute Resolution Requests/Noticeboards,” where editors try to work through disagreements by means of intensive discussions (Sinders, Poore, and Earley Reference Sinders, Poore and Earley2007). In this discussion process, disagreements about content often lead to grievances about conduct (Peake Reference Peake2015). Wikipedia has developed a complex ecosystem of rules of conduct, including strong norms of civility that forbid editors from expressing anger, leveling accusations against each other, or behaving in a biased way.Footnote 3 A category of editors called “administrators,” who are elected by the community of editors to exercise supervisory duties, can intervene by blocking or unblocking editors when they breach these norms. Despite these safeguards, edit warring became so intense that, in 2003, Wikipedia’s cofounder, Jimmy Wales, decided to create an Arbitration Committee (commonly called “Arb Com”), tasked “with the difficult and painful and regrettable task of banning someone from editing.”Footnote 4 The Arbitration Committee, sometimes dubbed “Wikipedia’s Supreme Court,” is a controversial institution on English Wikipedia (as well as in other national Wikipedia projects). Some editors resent it as a hierarchical element in an otherwise egalitarian universe that strongly favors a model of “self-managing organization” (Konieczny Reference Konieczny2010, 264). Others fear its significant powers, which range from merely reprimanding editors to indefinitely banning them. The fifteen members of the Arbitration Committee are elected from among all Wikipedia editors by tranche every year in December. Their term ranges between one and three years. The main duties and responsibilities of the Arbitration Committee are defined as follows: “to act as a final binding decision-maker primarily for serious conduct disputes the community has been unable to resolve ….”Footnote 5 If an editor is dissatisfied with an administrator’s decision or if the discussions among editors fail to lead to a satisfactory outcome, another editor can bring a case before the Arbitration Committee. The work of the Arbitration Committee is subject to an important limitation insofar as it is not allowed to review the substance of the edits (which is left to the community to discuss), but only the conduct of the editors (whom it can sanction). The Arbitration Committee can also overturn an administrator’s decision, thus issuing a sort of de facto appeal. Although its decisions resemble court judgments in their structure (with a template setting out the “case information,” “preliminary statements,” “preliminary decisions,” and “final decision”), the functioning of the Arbitration Committee is rather informal and flexible. Parties and third parties are allowed to state their position in writing within a defined time frame and word limit, but the Arbitration Committee is not bound by strict procedural or evidentiary rules. It rarely, if ever, refers to due process.Footnote 6 Prior to examining the merits of the dispute, however, the Arbitration Committee examines whether the parties have exhausted their other options for dispute resolution in the dedicated spaces on Wikipedia. Typical cases drag on for months or years before reaching the Arbitration Committee. Examples include a dispute concerning the behavior of editors over the renaming of Chelsea (then Bradley) Manning’s Wikipedia page after she announced her intention to transition (five editors were banned from editing any page relating to transgender rights, and one was “admonished” and had their use of administrator’s tools restricted),Footnote 7 and a dispute concerning the editing of the Church of Scientology’s Wikipedia page (which resulted in the blocking of all edits from Church of Scientology IP addresses).Footnote 8
Based on empirical data drawn from a series of interviews and an exhaustive analysis of its cases between 2004 and 2020, this article examines the ways in which the Arbitration Committee handles conduct-related disputes between editors, and the impact of social capital, summarily defined as the “capital captured through social relations” (Lin Reference Lin2001, 19), on this process. My empirical material indicates that the Arbitration Committee not only examines the merits of the claims made by the disputants, but also and more crucially facilitates a social contest by considering the position of each disputant within the community of editors. Ultimately, the Arbitration Committee seeks to “cancel” these disputes in order to preserve Wikipedia’s social fabric.
On a more general level, the material presented in this article casts new light on a facet of social capital that is often overlooked in the literature. Indeed, with some rare exceptions (Bourdieu Reference Bourdieu1980; Bourdieu Reference Bourdieu and Richardson1986; Portes Reference Portes1998, 15–18; Portes Reference Portes2014), most scholars adopt a celebratory vision of social capital (Putnam Reference Putnam2000; Fernandez, Castilla, and Moore Reference Fernandez, Castilla and Moore2000; Paxton Reference Paxton2002; Migheli Reference Migheli2012; Son Reference Son2020, 98–123). For instance, scholars argue that social capital facilitates compliance with norms (Coleman Reference Coleman1988a; Piskorski and Gorbatâi Reference Piskorski and Gorbatâi2017; Arnold and Kay Reference Arnold and Kay1995). By contrast, the data presented in this article suggests that well-connected members of a group use social capital to obfuscate their noncompliance with norms, thus leading to what I call “dispute cancellation.” This observation brings light on how dispute settlers maximize the legitimacy of their final decision by prioritizing the individuals who stand at the core of society. It also illustrates how power relations emerge in decentralized and horizontal communities (Freeman Reference Freeman2013) by specifically examining the ways in which these power relations affect the resolution of disputes. In doing so, the article draws a line between the top-down, centralized model of “commercial content moderation” and the horizontal, decentralized model of “community moderation” (Grimmelmann Reference Grimmelmann2015; Seering Reference Seering2020; Cullen and Kairam Reference Cullen and Kairam2022). It shows how the moderation of a horizontal space by the Wikipedia community relies on implicit, but no less real, social hierarchies. In particular, the consensus that Wikipedia boasts of as representing a mark of egalitarian and collaborative governance leads to the domination of an “aristocracy” composed of white Western men (Peake Reference Peake2015), who prevail in disputes brought to the Arbitration Committee by mobilizing their powerful networks.
The article proceeds in four steps. It starts with preliminary observations building on a review of the literature on social capital and drawing a hypothesis on this basis. It then proceeds with a presentation of the empirical data and methods, before setting out the results and discussing the findings.
PRELIMINARY OBSERVATIONS
This section presents competing narratives of social capital and explains how one of these narratives has left a particularly deep imprint on scholarly analysis of compliance with norms. The main argument of this dominant narrative is that high levels of social capital facilitate compliance with norms. When applied to the study of Wikipedia, however, this argument may face empirical and interpretative challenges that reveal the limitations of the dominant narrative.
The Various Shades of Social Capital: Bourdieu vs. Coleman
Social capital is a complex notion that has been used in different senses in socioeconomic scholarship (Farr Reference Farr2004). For instance, Marx uses the term to designate the cumulative amount of financial capital held by individuals in a given society (Marx [1867] Reference Marx and Fowkes1976, 777). However, the first significant attempt to delineate the notion of social capital was made by Pierre Bourdieu, who defined it as the “resources, actual or virtual, that accrue to an individual or a group by virtue of possessing a durable network of more or less institutionalized relationships of mutual acquaintance and recognition” (Bourdieu Reference Bourdieu1980, 2; Bourdieu and Wacquant Reference Bourdieu and Wacquant1992, 119; Bourdieu Reference Bourdieu and Richardson1986, 21). Bourdieu’s definition has two distinguishing features: (i) it highlights that social capital matters not only at the group level but also and more importantly at the individual level; and (ii) it insists that dominant individuals can strategically use social capital as a “resource” within a group. Beyond this attempt at defining social capital, there exist many debates about its effects. Bourdieu himself adopts a rather critical stance in these debates, arguing that social capital is a tool that individuals use to maintain or even expand their dominance within society (Bourdieu Reference Bourdieu2000, 363–64). Social capital, according to Bourdieu, helps individuals define and maintain their position of dominance within a given field or across fields (Bourdieu Reference Bourdieu1984, 3). For instance, physicians benefit from the human and symbolic capital associated with their title, but also draw on the vast social networks to which medical associations, congresses, or alumni groups give them access (Bourdieu Reference Bourdieu1984, 8). Individuals typically mobilize the networks to which they belong in order to reinforce their position in society, potentially to the detriment of social justice and equality. In Bourdieu’s view, social capital is “another way of maintaining and reproducing the dominant class” (Lin Reference Lin2001, 24).
Bourdieu’s rather critical account of social capital stands in contrast with another view according to which social capital is largely beneficial to society (Siisiäinen Reference Siisiäinen2000). In this view, social capital is deemed to enhance “the outcomes of actions” (Lin Reference Lin2001, 19–20). An important voice in these debates is that of James Coleman, “a seminal theorist of social capital” (Son Reference Son2020, 12), who defines social capital in terms of its function, namely the fact that it is composed of entities that “facilitate certain actions of actors – whether persons or corporate actors – within the structure” (Coleman Reference Coleman1988a, S98; Coleman Reference Coleman1990, 302). The overall success of Coleman’s theory has largely overshadowed Bourdieu’s earlier, less optimistic analysis of social capital (Siisiäinen Reference Siisiäinen2000; Sobel Reference Sobel2002, 145–46), as Bourdieu himself acknowledged (Bourdieu Reference Bourdieu, Bourdieu and James1991, 377–84). I argue below that Coleman’s definition of social capital differs from Bourdieu’s definition in two main respects.
First, Coleman’s definition emphasizes the positive effects of social capital, most importantly, within the present context, with respect to compliance with norms. The gist of Coleman’s argument is that social capital “facilitates action” (Coleman Reference Coleman1988a, S98, S100) through various conduits. Social norms, which are presumed to operate effectively in closed social networks (Coleman Reference Coleman1988a, S104–08), provide one such conduit. According to Coleman, a “network with closure” (or a “closed network”) is one in which all members are interconnected and develop a consensus about what is proper or just (Coleman Reference Coleman1988b, 383). Coleman argues that the structural features of these groups “lead to the development of norms” (Coleman Reference Coleman1988b, 383) and “sustain effective norms” (Coleman Reference Coleman1988a, S117). This occurs because individuals have better knowledge of these social norms (what I call the “normative knowledge” argument) but also because they are more likely to face retributions from the many people with whom they have social ties when a breach occurs (what I call the “deterrence of breach” argument). In other words, the wealth of social ties multiplies the potential forms of social control, and norms are more effective in groups with high levels of social capital (Coleman Reference Coleman1988a, S107). Another implication is that, because norms are internalized and self-enforced, disputes are less likely to arise in these closed networks in the first place (leading to a nonadversarial compliance regime). Coleman’s analysis has received some empirical support from the scholarship, for instance in the study of a Canadian disciplinary body for lawyers (Arnold and Kay Reference Arnold and Kay1995). This positive view of social capital contrasts with the more pessimistic (some would say, realistic) view defended by thinkers such as Bourdieu, who insists on the ways in which individuals can mobilize their social networks to gain the upper hand in social struggles.
Second, Bourdieu’s and Coleman’s analyses of social capital focus on different levels. Bourdieu emphasizes inequalities of social capital at the individual level, whereas Coleman adopts a macro view of social capital as it exists “within the [social] structure.” As we have seen, Coleman pays particular attention to a type of social structure characterized by “closure,” in which each group member is related to all other group members. He seems to consider that, in this type of social structure, social capital is evenly distributed among actors. However, the conditions of “closure” are so restrictive that they are unlikely to be met in practice, other than in very small, tight-knit, groups. In fact, most societies seem to be characterized by the uneven distribution of social capital. The challenge is then to examine the impact that unequal concentrations of social capital may have on the effectiveness of norms. It is noteworthy in this regard that the two mechanisms outlined above (i.e., “normative knowledge” and “deterrence of breach”) are less likely to operate when social capital is distributed unevenly.
The Role of Social Capital in the Enforcement of Wikipedia Norms
The theoretical groundwork sketched above informs the ways in which scholars turned to Wikipedia as a case study to test Coleman’s idea that social capital facilitates compliance with social norms (Piskorski and Gorbatâi Reference Piskorski and Gorbatâi2017). Piskorski and Gorbatâi’s hypothesis is that network density and solidarity encourage insiders to comply with social norms and discourage them from violating these norms. In order to test their hypothesis, they use quantitative data drawn from “undos” and “reverts” carried out by Wikipedia editors (Piskorski and Gorbatâi Reference Piskorski and Gorbatâi2017, 1205). A bit of context is needed to understand the significance of their research for the study of Wikipedia. The “undo” function allows a Wikipedia editor to restore original content after an edit has been made by another editor, and the “revert” function allows an editor to reinstate an edit by restoring content. Piskorski and Gorbatâi observe, inter alia, that editors belonging to a dense network are less likely to experience an undo and that, if they do experience an undo, there is more likely to be a subsequent revert by a third-party contributor (Piskorski and Gorbatâi Reference Piskorski and Gorbatâi2017, 1205). Based on these findings, Piskorski and Gorbatâi reach the following conclusion: “(i) contributors in dense networks are less likely to violate norms or experience norm violations, (ii) they are more likely to punish others who violate or experience such punishment, and (iii) they are more likely to reward others who punish norm violators or witness such a reward” (Piskorski and Gorbatâi Reference Piskorski and Gorbatâi2017, 1209). According to Piskorski and Gorbatâi, these conclusions show the merit of Coleman’s theory of social capital as applied to the enforcement of norms. According to them, the fact that contributors with multiple connections experience fewer undos and more reverts indicates that these individuals are more compliant with social norms (the “normative knowledge” argument) and that they benefit from the potential retribution exacted by contributors belonging to their network (the “deterrence of breach” argument). Their analysis does, however, raise some potential issues concerning the empirical context in which their data is situated and the different ways in which this data can be interpreted.
Empirical and Interpretative Issues
The first issue with Piskorski and Gorbatâi’s analysis is that they fail to consider the full scope of normative enforcement and dispute resolution process on Wikipedia. A key problem is the discrepancy between Coleman’s compliance regime and the empirical reality of this dispute resolution process. In Coleman’s enforcement regime, disputes are unlikely to arise due to the combined effect of “normative knowledge” and “deterrence of breach.” However, Wikipedia does not entirely fit with this model of a nonadversarial compliance regime. In fact, as was pointed out in the introduction, disputes do commonly arise on Wikipedia.Footnote 9
Second, Piskorski and Gorbatâi’s finding that editors with high social capital experience fewer undos and more reverts than others can be interpreted in two diametrically opposed ways. The first interpretation is consistent with Coleman’s analysis and is, in fact, the one they adopt. According to this interpretation, editors belonging to dense social networks experience fewer undos and more reverts because they are more likely to comply with norms than editors with less social capital. However, another interpretation of Piskorski and Gorbatâi’s finding is that editors with high social capital experience fewer undos and more reverts not because of their compliance with norms, but because they are able to mobilize their social capital in order to support their own version of the truth. For instance, well-connected editors might experience fewer undos as a result of their prominent position within the social network, which discourages others from acting against them. Similarly, they might experience more reverts because their numerous allies come to their rescue when other editors dare to undo their edits. Based on an empirical study of the cases brought before the Arbitration Committee, I propose to apply this second interpretation to the resolution of disputes relating to conduct on Wikipedia. My main hypothesis is that, just as social capital can be used to push a certain version of the truth, it can also be exploited by well-connected editors in order to prevail in conduct-related disputes. While this hypothesis is not incompatible with the argument that social capital considered at the group level might foster normative compliance, it examines how strategic actors use their individual capital in order to prevent their own normative breaches from being sanctioned. This hypothesis finds some resonance in the existing literature. For instance, Hoffman and Mehra argue that dispute resolution on Wikipedia “functions not so much to resolve disputes and make peace between conflicting users, but to weed out problematic users while weeding potentially productive users back in to participate” (Hoffman and Mehra Reference Hoffman and Mehra2009, 151). This article complements Hoffman and Mehra’s analysis of the ways in which the Arbitration Committee regulates the conduct of editors by examining more specifically how the Arbitration Committee considers the social position of these editors when sanctioning their conduct. Before discussing my results below, I will present my methods and data.
METHODS AND DATA
I use a mixed method in order to test this hypothesis. This mixed method has the advantage of contextualizing quantitative findings (Silverman Reference Silverman2020, 405), while helping to understand “the mechanisms underlying the quantitative patterns” (Kritzer Reference Kritzer, Halliday and Schmidt2009, 272). I first collected quantitative data by reviewing the online archives of the Arbitration Committee between 2004 (the date of its first case) and 2020, and assessed the relationship between the harshness of the remedies pronounced by the Arbitration Committee, on the one hand, and the levels of social capital displayed by the losing parties, on the other hand. The second step was to further examine this relationship through qualitative interviews.
The first step in the data collection consisted in reviewing 524 cases submitted to the Arbitration Committee between February 2004 and July 2020. These 524 cases represent the entire population of cases decided by the Arbitration Committee during this time period. The files of these cases are available in Wikipedia’s online archivesFootnote 10 and contain details that were listed in a table, including: the date of the request for arbitration, the date of the final decision, the name of the arbitrators, the name of claimant(s), the name of respondent(s), the decision of the Arbitration Committee, and the remedies ordered by the Arbitration Committee. The remedies ordered by the Arbitration Committee are particularly important for the assessment of my hypothesis, and I have ranked them by level of stringency, assuming that the Arbitration Committee is more likely to order a stringent remedy if the corresponding breach is stronger (an assumption that I will question further below). I then assessed the social capital of each party by identifying their number of edits in specific subject areas. Reagle describes “edit counts” as the “coin of the realm” on Wikipedia and a “rough approximation of one’s involvement and commitment to the project” (Reagle Reference Reagle2010, 157). These edit counts usually relate to “normal” Wikipedia pages, but they can also concern discussions regarding the governance of Wikipedia itself (McDowell and Vetter Reference McDowell and Vetter2021, 3). I focus here on specific edit counts called “Wikipedia” and “Wikipedia Talk,” which provide an estimate for the number of edits made by a given editor on topics associated with the Wikipedia project itself, specifically its norms, policies, and governance (these pages are identified by the abbreviation “WP:” or “Wikipedia:”—e.g., “WP:NPOV” for a discussion of the “neutral point of view”). More specifically, the “Wikipedia” count concerns the number of edits made to these pages, while the “Wikipedia Talk” count relates to the number of edits made in the “talk section” of these pages (which concerns discussions surrounding these pages). As in many other societies, social prominence on Wikipedia is associated with active participation in debates concerning policy and governance. Editors who frequently edit “Wikipedia” or “Wikipedia Talk” pages are prominent members with extensive networks and significant social capital on Wikipedia (a point confirmed by my interviewees). These indicators are available on XTools, a collection of statistical data on Wikipedia.Footnote 11
By way of example, Figure 1 shows the breakdown of the “edit contributions” of Wikipedia user “Theresa Knott” on XTools. It is possible to identify the various “namespaces” where Theresa Knott contributes to Wikipedia, as well as the number of “Wikipedia” and “Wikipedia Talk” edits that she has made (highlighted respectively in dark and light blue).
FIGURE 1. Edit Contributions by User Theresa Knott (XTools).
A comparison between these two proxies on the one hand and the remedies ordered by the Arbitration Committee on the other hand allows us to examine the relationship between the levels of social capital and the strength of the remedies ordered by the Arbitration Committee.
The second step in my data collection consisted of a series of interviews with a wide range of Wikipedia editors who have been involved in various capacities with the Arbitration Committee. I interviewed twenty-eight individuals one or more times over a nine-month period online, by phone, or by email (depending on the interviewees’ concern for privacyFootnote 12 ). These individuals belong to four broad categories: current or former members of the Arbitration Committee; editors who have been involved as parties before the Arbitration Committee; editors who have commented on cases as third parties before the Arbitration Committee; and editors who have knowledge of the dispute resolution processes due to their long-standing involvement with Wikipedia. I selected editors whose involvement covered the entire time period of my study (2004–2020). Of the twenty-eight editors whom I have interviewed, eight formally identified themselves as female, which represents a slightly higher proportion (28.6 percent) than the overall proportion of female editors on Wikipedia (ranging between 13 percent and 20 percent depending on the surveysFootnote 13 ). The majority of interviews were semistructured, with a list of questions providing a flexible framework for the discussions, some of which were open-ended. In order to promote frank and transparent discussions with my interviewees, I have chosen to keep their names confidential (often at their own request).
RESULTS
The quantitative data I collected provides insights into the relationship between the remedies ordered by the Arbitration Committee, on the one hand, and the social capital displayed by the losing parties, on the other hand (measured by two proxies, “Wikipedia” count and “Wikipedia Talk” count). Figure 2 represents the median “Wikipedia” and “Wikipedia Talk” counts for the parties condemned by the Arbitration Committee and breaks down these results by categories of remedies: administrator rights removed, warning, admonishment, interaction ban, usage restrictions, topic ban, and Wikipedia ban.
FIGURE 2. Social Capital and the Arbitration Committee’s Remedies.
The right side of Figure 2 shows that the median “Wikipedia” count for all parties involved in cases of the Arbitration Committee is 1,170, and that the median “Wikipedia Talk” count for the same parties is 218. These figures provide a useful benchmark for the analysis of the remedies presented on the left side of the figure. As shown in Figure 2, the median “Wikipedia” count for parties condemned to have their “administrator rights removed” is 2,738 and the median “Wikipedia Talk” count for these same parties is 487.5. Both figures are significantly higher than the median counts of all parties who appeared before the Arbitration Committee (by a factor of more than two). However, this result is not significant for the purposes of this study, as the “administrator rights removed” remedy is applied to parties who by definition possess high levels of social capital. Indeed, this remedy exclusively targets “administrators” who, due to their very functions, have high “Wikipedia” and “Wikipedia Talk” counts.
The remaining remedies—warning, admonishment, interaction ban, usage restrictions, topic ban, and Wikipedia ban—are ranked in ascending order of severity. If one excludes the “warning” remedy (which is associated with low “Wikipedia” and “Wikipedia Talk” countsFootnote 14 ), a negative correlation appears between the “Wikipedia” and “Wikipedia Talk” counts, on the one hand, and the strength of the remedies applied by the Arbitration Committee, on the other. This negative correlation appears from the part of Figure 2 delimited by a rectangle: editors with high levels of social capital are more likely to face lighter sanctions than harsher ones. For instance, these editors are much more likely to be given an admonishment, a relatively light sanction, than a ban, the harshest sanction on Wikipedia. Another way of analyzing the same data is to focus on the two categories on the far right of the rectangle (topic ban and Wikipedia ban). It appears that editors with fairly low levels of social capital are much more likely to face these harsh sanctions than editors with high levels of social capital.
However, these simple descriptive statistics do not say much as the negative correlation presented in Figure 2 can be interpreted in diametrically opposed ways. This correlation could indicate that editors with high social capital are unlikely to face harsh sanctions because they are more compliant with Wikipedia norms—an explanation that is entirely consistent with Coleman’s theory. Alternatively, this correlation could mean that the Arbitration Committee is reluctant to sanction well-established editors or, on a stronger version of the same argument, that well-established editors successfully weigh on the dispute resolution process by mobilizing their social networks—a hypothesis that is closer to Bourdieu’s theory of social capital. The interviews that I conducted provide guidance on how to assess the strength of each position. Several themes that directly inform my main hypothesis have emerged from the interviews. These themes fall under two broad categories: one concerns the behavior of parties before the Arbitration Committee; another relates to the dynamics of the Arbitration Committee itself.
Disputing Parties, Factions, and Preliminary Statements
My interviewees unanimously report what some call “factionalism” on Wikipedia. Disputing parties actively seek the support of allies and friends, in more or less obvious ways, in relation to their case before the Arbitration Committee (Jemielniak Reference Jemielniak2014, 50–55). These parties turn toward long-standing friends or allies when gathering their “factions” and “cliques.” For instance, a Wikipedia insider told me how a disputant prepared her “faction” for months before bringing a case before the Arbitration Committee (which she ended up winning). These efforts are usually made covertly, as Wikipedia norms prohibit what is called “canvassing” (defined as a “notification done with the intention of influencing the outcome of a discussion in a particular way”).Footnote 15 For instance, a group of editors specialized in the history and politics of Eastern Europe decided to coordinate their activities on a secret mailing list, a breach of Wikipedia’s anticanvassing policy, which led to a case before the Arbitration Committee.Footnote 16 Notwithstanding this anticanvassing policy, individuals with high levels of social capital typically know how to mobilize their networks without making it too obvious. A long-standing editor who was described as a member of Wikipedia’s “aristocracy” told me: “we are a tight clique of very long-standing editors and none of our words find their way onto the site” (which was meant to capture their propensity to “canvass” without getting detected). A member of the Arbitration Committee told me that he was close to this small “aristocratic” group (although he did not consider himself as being part of it), which he compared to a “cabal.” One editor told me how a “Wiki aristocracy” made of “200 or 300 people” can create “consensus” by generating “flash mobs” in support of their position. According to this editor, members of the “aristocracy” deem themselves to be part of the “same class of editors who hold the Wikipedia project together.” Because their views potentially conflict with Wikipedia’s egalitarian philosophy, these individuals coordinate their action “offsite.” An exchange between Wikipedia editors to which I was made privy further illustrates these dynamics:
X <X@gmail.com> Mon, Jun 28, 6:52 PM to Y
There’s such a HUGE power advantage that experienced editors have who are familiar with the ins & outs and “unwritten rules” that affect who wins an edit war
Y <Y@gmail.com> Mon, Jun 28, 6:52 PM to X
Yeah, it’s unfortunate. I think it’s better than a lot of communities online but it’s far from perfect. And experienced editors can just grind you down over wars of attrition. It’s so deflating.
Y <Y@gmail.com> Mon, Jun 28, 6:53 PM to X
Yes, exactly. I have said that so many times, including this morning.
X <X@gmail.com> Mon, Jun 28, 6:53 PM to Y
yeah the grinding-down is the worst part
X <X@gmail.com> Mon, Jun 28, 6:53 PM to Y
they will wear you out and exhaust you into submission
Y <Y@gmail.com> Mon, Jun 28, 6:53 PM to X
Exactly.
…
X <X@gmail.com> Mon, Jun 28, 6:54 PM to Y
they know how to game the system and make themselves look like the victim and make you look like the perpetrator.
Experienced editors know how to navigate the procedural subtleties of the dispute resolution process, for instance by mobilizing their allies in subtle ways without lapsing into “canvassing.” There are, however, two moments when these networks display their strength more openly in the dispute resolution process, namely an early phase of the arbitration process when parties and third parties make “preliminary statements” that find their way into the decision of the Arbitration Committee and a “workshop page” where the parties, the community, and arbitrators offer suggestions and obtain feedback about final decision proposals. At both times, factions and cliques are called in to make their contributions, which are often numerous and detailed. For instance, in a case concerning the pricing of drugs and medicines, no fewer than thirty editors made preliminary statements.Footnote 17 These “preliminary statements” can turn the process before the Arbitration Committee into a shouting contest. A member of the Arbitration Committee described this process as follows:
A long-retired arbitrator once told me that the only real benefit of the workshop page of an ArbCom page was to keep the disputants shouting at each other in one place, so they wouldn’t be messing up the rest of the project. While there’s some cynical truth to that, the reality is that it’s almost impossible for arbitrators to hear each other’s voices on workshop pages, many of the initial case requests, or any time there is a discussion about anything ArbCom-related onwiki: there’s too much noise, and too many side discussions, with too many people complaining that their own small aspect of the greater issue has not been addressed. It’s like having a philosophical discussion in a middle of a riot.Footnote 18
For instance, in a case brought against an editor called Magioladitis (who came under scrutiny for having set up a “bot,” i.e., an automated editing system), eighteen editors who were not otherwise involved in the case made statements for or against him. Many of them explicitly referred to their prior knowledge of Magioladitis and his actions, whether in positive or negative terms: “Magioladitis is one of those few highly skilled, dedicated, experienced, respectable, etc., etc., editors who mass-apply minute ‘fixes’ to a vast number of articles at a high rate”; “Perhaps it is easy to forget the sheer amount of work that Magioladitis has done, both through his coding work on AWB, his bot edits, through Yobot, and through manually editing himself”; “Magioladitis has done sterling service, for little or no reward, and has always been civil with those who question what he does.”Footnote 19 Each group of editors also takes a side vis-à-vis the other group, thus transforming the “preliminary statements” into a form of battleground: “I don’t recall interacting with you directly, but I had much respect to your work; if you care about this respect, choose your words more carefully and don’t twist mine”; “@Beyond My Ken: If you disagree with what the policy actually says, ArbCom isn’t the place to try to change it. Misrepresenting your own opinion as policy is also no helpful”; “OID, I haven’t made any claims about the civility of the friction caused by a vague policy in my statement above.”Footnote 20 It remains to be seen how the Arbitration Committee reacts in the face of these contests. I argue below that the Arbitration Committee pays particular attention to the balance of social power when seeking to resolve these contests.
Social Capital on Trial
Some arbitrators try to remain indifferent toward this social pressure. A member of the Arbitration Committee stated that he did not “care about the politics; I am generally less inclined towards the unblockables and vested contributors,” while admitting that “we [the Arbitration Committee] give longtime contributors more leeway.” Other arbitrators seem more amenable to cut well-connected editors some slack. One arbitrator called this process the “net positive/net negative questions” and confided that “you might be more predisposed to assume good if they [the parties] have a long history of positive contributions.” According to the same arbitrator, one “accumulates social capital by being a good contributive editor.” This leads to a situation deplored by multiple interviewees, in which well-established editors—sometimes referred to as “unblockables”—are so deeply embedded in Wikipedia’s social networks that “you have to put up with their bullshit” (in the words of an interviewee). One editor who was banned by the Arbitration Committee put it bluntly: “Social capital is the basic currency for getting things done on Wikipedia.” When using these words, this editor signals the fact that a “positive” contribution on Wikipedia crucially depends on one’s ability to gather troops in support of a version of the truth (which ultimately reflects one’s level of social capital). A seasoned Wikipedian confirmed that “there were some extremely powerful editors with some very close friends” and that “it was almost impossible to penalize them.” Because of their long-standing involvement with Wikipedia, editors with dense networks are less likely to be sanctioned by the Arbitration Committee. A Wikipedia editor told me that he was “afraid of Arb Com” because it prioritizes larger groups over smaller ones when deciding disputes. The same interviewee, whom the Arbitration Committee stripped of his administrator’s rights, explained how groups of people were “piling on against him” during the dispute resolution process and that “Arb Com wanted to end the dispute, and wanted the fight to be not happening.” When asked why he did not use his own networks to respond to the other side’s display of force, this individual said: “I could have tried to mobilize people, but I had reached my limits with Wiki. In addition, there are strong guidelines against canvassing on Wiki and I don’t play the politics. A few editors would probably stand up for me, but most didn’t want to be involved in this train wreck.” When asked about the reasons for his having been sanctioned by the Arbitration Committee, another interviewee told me: “I just irritated people who have a lot of pull. You want to keep your head down when the aristocracy gets involved.” Another editor who was involved in several cases before the Arbitration Committee confirmed that “cliques” are a fundamental aspect of the disputes and that “you will lose the case unless you have some powerful administrators to protect you.” One could argue that, when preserving insiders, the Arbitration Committee is trying to improve the accuracy of Wikipedia. This argument assumes, however, that insiders promote truthfulness, which is not necessarily the case on Wikipedia. In fact, Wikipedians sometimes refer to the fictional figure of “Randy in Boise” as the “archetypal uninformed but relentless Wikipedia editor” who successfully imposes their version of the truth over, say, an established expert in the field. Social capital is no synonym for expertise.
DISCUSSION
The key theme emerging from the data presented above is the role played by social capital in what I call the “cancellation of disputes.” This notion of “cancellation” draws directly from the words used by my interviewees (“The Arbitration Committee wanted to end the dispute, and wanted the fight to be not happening,” “Arb Com’s explicit commitment to break the back of the dispute structurally biases it towards censuring the dissenter(s) rather than determining who is right or wrong”) (emphasis added). I will further analyze this notion of “cancellation” below and explain how it can be used to complement existing theories of social capital and dispute resolution.
The study of Wikipedia’s Arbitration Committee highlights the role played by social capital in compliance with norms and its contribution to what I call the “cancellation of disputes.” My data shows that it is groups of individuals acting in concert, rather than individuals operating in isolation from one another, that actively participate in the dispute resolution process before the Arbitration Committee. These social groups seek to influence the dispute resolution process by expressing their views and making a show of their strength by means of “preliminary statements.” Editors who make preliminary statements often mention their knowledge of the accused in these statements. One editor who participated in several cases before the Arbitration Committee described the submission of preliminary statements as a “power play,” with some editors submitting their comments right before the deadline set out by the Arbitration Committee in order to prevent other editors from responding, while others multiply statements in order to prevent the accused from meaningfully responding to all the allegations. When faced with a large number of statements, an accused editor may indeed be unable to answer most of the allegations.
This “power play” seems to have an impact on the work of the Arbitration Committee. As indicated above, my quantitative data indicates a negative correlation between the level of social capital displayed by the losing parties and the severity of the sanctions imposed by the Arbitration Committee: the more social capital is displayed by particular editors, the less likely it is that they will face harsh sanctions from the Arbitration Committee. During interviews, members of the Arbitration Committee confirmed that social capital plays a role in the outcome of the disputes and that they are more reluctant to sanction editors who display considerable social capital. Several editors who have been involved in cases before the Arbitration Committee provide an explanation for this apparent bias, claiming that the Arbitration Committee’s primary objective is not to resolve disputes, but to “break the back” of these disputes. The Wikipedia Guide to arbitration (which was written by prominent arbitrators) refers to this objective in plain words: “Arbitration aims to ‘break the back’ of the dispute.”Footnote 21 In other words, the disputes run so deeply into Wikipedia’s networks that it is impossible to fully resolve them (in ways that are reminiscent, for instance, of multiplex societies where conflict is a “response to contradictory pressures” with neither side being fully guilty or innocent (Eckert and Newmark Reference Eckert and Newmark1980, 193)). It is possible, however, for the Arbitration Committee to maximize the level of concord that is necessary for the project to remain effective, by ensuring that a maximum number of editors are satisfied with the outcome of the dispute. This leads the Arbitration Committee to engage in a kind of “social balancing” exercise, which consists of weighing up the social strength of each side before issuing its decision. An individual who was involved in two cases before the Arbitration Committee (one in which he prevailed, another in which he lost) put it bluntly: the Arbitration Committee is “more sensitive to claims made by prominent editors.” This individual won the first case, on his own account, because his opponent “was not prominent.” He lost the second case against another individual, however, because his opponent was “more prominent” than himself. Of course, the reality of cases is more complex and contested than what can be glimpsed from the perception of a single party (especially a party that has lost its case). For instance, the Arbitration Committee tries to keep the “social balancing” exercise as discreet as possible, by making sure that its decisions are supported by applicable “principles” (although an insider might be able to track the strength of the social contest in the “Preliminary Statements” section and “Workshop” page of the decision), thus showing the controversial character of its decision-making process.
There is another important element that makes it possible to add nuance to the picture that I have drawn thus far of the Arbitration Committee’s work, namely the sanctions that the Arbitration Committee imposes on administrators. Most administrators are well connected due to their election by the community (usually with majorities of well above 65 percent) and the special powers that they hold (they can, for instance, block or unblock user accounts).Footnote 22 Despite the administrators’ social status, the Arbitration Committee does not hesitate to pronounce harsh sanctions against them when they are accused of misconduct, precisely because they hold significant powers. “We don’t want bad cops,” explains a former member of the Arbitration Committee. However, the “harshness” of the Committee’s sanctions is only relative: it often strips administrators of their powers, but rarely touches their editing rights (what Wikipedians call the “Super Mario” effect: when “Super Mario” loses his special powers, he can still carry on as “Mario”).Footnote 23 This “Super Mario” effect, combined with the fact that most administrators have “some very close friends” on the Arbitration Committee, has led to the perception that administrators enjoy “a certain immunity” according to an interviewee, who added that “Arb Com gradually got the courage to take some actions [against them].” Another dimension of the problem is that administrators often become unpopular: “It’s like being a cop,” says a former administrator, adding that he “alienate[d] a lot of people” and “buil[t] enemies” as a result of blocking editors. This unpopularity can affect the administrators’ capacity to mobilize networks when a dispute arises. Overall, the weight of my evidence indicates that the Arbitration Committee is aware of its own tendency to favor powerful individuals, which it tries to mitigate with varying degrees of success.
The “arbitrariness” of the Arbitration Committee’s decisions is another important theme that transpires from my data. An editor compared the Arbitration Committee to “riot cops” (who arguably target the weaker parties when trying to end a battle) to underline its focus on the “cancellation” rather than the “resolution” of disputes. These riot cops can be compared to the “repressive peacemakers” described by Black and Baumgartner: they are indifferent to the substantive issues, since their goal is merely to end the dispute as quickly as possible (Black and Baumgartner Reference Black, Baumgartner, Keith and Mather2015, 109–11). What ultimately matters for the Arbitration Committee is guaranteeing the level of social peace that is necessary for the Wikipedia project to continue to unfold, even to the detriment of fairness. A prominent member of the Arbitration Committee criticized the view according to which the Arbitration Committee is a “supreme court,” noting that it is “nothing like that at all.” The same interviewee told me: “Arb Com is more like a field court martial with a huge amount of discretion; it operates on the basis of what is best for the encyclopedia and it does not consider legal principles.” Because the Arbitration Committee is so attentive to the social life of the network when considering cases, its decisions are sometimes deemed arbitrary or unpredictable (to the extent that it is hard to predict how the social contest will develop, which partly depends on the willingness and capacity of (third) parties to mobilize their networks). One editor told me: “I never advise people to go to Arb Com because its decisions are unpredictable. People who bring the case end up being sanctioned.” A former member of the Arbitration Committee confirmed this allegation:
Before I joined arb com, when someone asked me if they should appeal there, my advice was to stay away from these processes, because almost nobody came out the better for it. Now that I have some considerable experience at Arb Com, I feel this way all the more strongly.Footnote 24
The Arbitration Committee does not arbitrate disputes by striking compromises between the parties. Nor does it decide these disputes in accordance with legal principles (although it does refer to applicable “principles” in its final decision). The work of the Arbitration Committee seems to be guided by a single overarching goal: to do what is best for the encyclopedia, even to the detriment of fairness or predictability. The Arbitration Committee therefore tends to prioritize individuals who make greater contributions to the encyclopedia, as evidenced by their ability to mobilize friends and allies in support of their position in the early phases of the proceedings. As mentioned above, this balancing of social strength is not explicit (although an insider might be able to track the social contest at play in the “Preliminary Statements” section of the Arbitration Committee’s decision or on its workshop page). The Arbitration Committee makes factual findings and refers to applicable “principles” (which it often restates), seemingly indicating an awareness of the need to legitimize its controversial decisions.
The interrelationship between dispute settlers and society, as well as the ways in which social actors mobilize their social capital in order to prevail in disputes, find some illustrations in the literature. In three lectures given between 1971 and 1984, Michel Foucault provides a potentially relevant analysis of this process, based on a study of Greek and feudal law. One of Foucault’s key contentions is that, in Ancient Greece, as in the feudal societies of Western Europe, dispute resolution processes provide a regulated avenue for social fights which they extend and control (Foucault Reference Foucault2012, 37; Foucault Reference Foucault2015, 115; Foucault Reference Foucault2001, 1440–41). In Foucault’s words: “The rule and the fight, the rule in the fight, is what defines the legal” (Foucault Reference Foucault2015, 115). Foucault gives several examples that are reminiscent of what can be observed in the case of the Arbitration Committee. He mentions, for instance, a passage from the Iliad where Hephaestus engraves a dispute between two individuals on the shield of Achilles (Foucault Reference Foucault2012, 36). In this engraving, each individual is surrounded by a group of people that is facing off against the opposing group. Foucault extends this analysis to the resolution of disputes in the feudal societies of Western Europe. In these societies, Foucault argues, the key is to declare “peace or victory,” not to find the “truth” (Foucault Reference Foucault2015, 129). Foucault further refers to a “social ordeal” aimed at “proving not the truth, but the strength, the weight and the importance of who is talking” (Foucault Reference Foucault2001, 1442). In this feudal system, a person accused of murder is deemed innocent if they are able to gather twelve witnesses who swear to their innocence (Foucault Reference Foucault2001, 1442). The reliability of these witnesses is assessed in light of their kinship with the accused, not their knowledge of the criminal events that they might not even have been present at (Foucault Reference Foucault2001, 1442). In other words, this “ordeal” assesses the social importance of the accused, irrespective of their actions. The same “social ordeal” has been observed in other societies such as the Berbers (Rosen Reference Rosen2018, 24) or the Eskimos (Hoebel Reference Hoebel1954, 92–99; Eckert and Newmark Reference Eckert and Newmark1980). In these societies, disputants solve disputes in “song duels,” with the participation of the members of their respective households, who act as a chorus enhancing the power of the song (Hoebel Reference Hoebel1954, 96). These processes are all aimed at ensuring concord and peace, but they achieve this goal in different ways: some processes (e.g., Wikipedia or the examples mentioned by Foucault) are aimed at minimizing the social consequences of the sanctions by punishing the weakest party; others (e.g., the song duels in Berber or Eskimo societies) are aimed at minimizing the social consequences of the sanctions by “reintegrate[ing] both parties in a conflict into the normal functioning of the community” (Eckert and Newmark Reference Eckert and Newmark1980, 200).
These observations cast additional light on the relationship between social capital and compliance with social norms. Does social capital improve normative compliance? Or can it be used to obfuscate normative breaches? The data that I have gathered with regard to Wikipedia suggests a middle way in these debates. It is clear that editors with significant experience and dense networks on Wikipedia have better knowledge of the network’s norms and how to comply with them. This situation is exacerbated by the fact that Wikipedia’s norms are made of “minutiae upon minutiae” (according to a member of the Arbitration Committee). Experienced editors master these specificities, which they can use to put pressure on less experienced editors. According to another member of the Arbitration Committee whom I interviewed, Wikipedia is “horrible to newcomers” because there are “so many norms about reliability, copyright.” But the same interviewee added: “Once you learn the way of talking, people listen.” Seasoned editors know, for instance, how and when to sign their posts—an apparently innocuous process that is, in reality, regulated by byzantine norms—and draw prestige from their knowledge of and ability to comply with these norms.Footnote 25 My empirical findings are consistent with the argument that, despite its rhetoric of inclusiveness (“anyone can edit”), Wikipedia is an “unwelcoming and exclusive environment” for newcomers (McDowell and Vetter Reference McDowell and Vetter2021, 74), which tends to reinforce the “hegemony” of a consensus that is mostly shaped and controlled by white Western men (Peake Reference Peake2015). It therefore appears that, due to their knowledge of norms and the socialization process on Wikipedia, editors with significant social capital are more likely to comply with Wikipedia’s social norms, thus supporting Coleman’s theory. But what also emerges from the evidence I have collected, and is perhaps more conclusive, is that experienced editors with dense networks are well positioned to avoid the consequences of their own breaches and to use their power to prevail in disputes against weaker parties. This is because these editors know how to navigate the normative subtleties of Wikipedia in order to end up on the winning side. In particular, they know how to signal their social weight to the Arbitration Committee, as well as the potential difficulties that the encyclopedia would face if they were excluded from the community. The Arbitration Committee’s sensitivity to this social contest can be explained by its focus on the long-term well-being of the encyclopedia. A former party to a dispute whom I interviewed very astutely described this state of affairs: “If ten people are in dispute with one person, the easiest way to soothe the dispute is to censure the one person rather than the ten.” A member of the Arbitration Committee confirmed that: “If enough people push it, it is very hard for the Arbitration Committee to go the other way.” This observation is entirely consistent with Bourdieu’s focus on the strategic use of social capital by individuals belonging to the dominating class, and how social capital helps maintain its position. It also shows the importance of understanding social capital not only at the group level but also and more importantly at the individual level. Disputes disappear on Wikipedia not because of the mechanisms of nonadversarial compliance described by Coleman, but because of the repressive peacemaking that derives from an unequal distribution of social capital.
CONCLUSION
This article highlights the processes through which a dispute resolution body, the Arbitration Committee, handles the disputes relating to conduct that frequently arise among Wikipedia editors. Because it is embedded in a social space where collaboration is of the utmost importance, the Arbitration Committee resolves disputes by canceling them. Rather than deciding which editor is right or wrong or arbitrating disputes, the Arbitration Committee identifies and favors the parties who hold high levels of social capital due to their long-term interactions as editors of the encyclopedia. The data presented in this article expands on the existing scholarship, showing how horizontal communities generate hierarchies (Freeman Reference Freeman2013) and, in my example, how hierarchies influence the ways in which disputes are resolved in these communities. In the case of Wikipedia, these hierarchies benefit a well-established group of white, Western men who shape the “rough consensus” that governs Wikipedia (Peake Reference Peake2015). The present article shows how these individuals leverage their dominant position within Wikipedia’s society by first transforming content disputes into conduct disputes (in which they enjoy a comparative advantage due to their familiarity with the rules), and then transforming conduct disputes into a social contest in which they are even more likely to prevail. Consensus among the many leads to the domination of the few. This domination does not go unnoticed or unchallenged: most Wikipedians are aware of these issues, and many try to address them. The fact remains that social capital ultimately offers incentives and sanctions that can foster compliance with norms, but it also provides the means for well-connected individuals to avoid the consequences of their actions, resulting in the cancellation of disputes and the obfuscation of violations.
Open access
