1. Introduction
[1.1] When scholars and fans discuss the politics of transformative works fandom, they tend to assume that it is progressive (Stanfill 2020). Even if the statement "fandom is progressive" were unarguably true, however, it would not guarantee racial inclusion. It is quite possible to be progressive in some ways and nevertheless racist, as scholarly and social movement critiques of white supremacy have long demonstrated (Combahee River Collective 1981; Reddy 2011). This presumption of progressiveness has suppressed attention to structural inequality within fandom, particularly racism (Pande 2016, 2018a; Stanfill 2019, 2020). This article looks at the racial structures underpinning one of the fandom institutions most widely seen as progressive, the fan fiction platform Archive of Our Own (hereafter AO3 or the Archive), and the nonprofit Organization for Transformative Works (hereafter OTW) that maintains it. Arguing that some form of action toward the unmaking of structural racism within AO3 is urgent and necessary, we examine some possibilities for what that action could be, working through the contradictions and implications each possibility entails.
[1.2] In the summer of 2020, when the language of racial reckoning entered US and transnational public spheres following the murder of George Floyd and the widespread uprisings protesting his addition to the litany of Black people killed by white supremacy, the contradictions of fandom's claims to progressive politics became sharply visible. On June 10, 2020, the OTW released its usual weekly collection of fandom-related news—including events relevant to the racial moment such as Star Wars actor John Boyega's impassioned speech at a London Black Lives Matter Rally (Suliman 2020) and fans of Korean pop music electronically shouting down a white supremacist hashtag (Lee 2020)—but made no particular statement of solidarity on its own behalf (Jess H. 2020a). Fans of color, who saw this as the latest in a long line of instances where the OTW and AO3 had not been responsive to their needs, expressed anger. A 2020 open letter with specific demands asking AO3 to address the issue of racist content in the archive began to circulate, eventually garnering 1,663 signatures (including the authors'); OTW responded on June 24 with a promise to address the concerns ("Open Letter to the OTW on Racism in Fandom" 2020; Jess H. 2020b). This article takes up the long-standing issues raised by the open letter, unpacking and contextualizing how racist content on AO3 might be addressed. What is particular about racism in transformative works fandom, and how might it best be contested? How could AO3's users, networks, and systems best participate in the work of dismantling fannish racism?
[1.3] AO3's creators have sought maximal inclusivity since the Archive's foundation at a moment when platforms had imposed restrictions on fan fiction. In particular, cofounder Francesca Coppa (2014, 78) writes that AO3 demonstrates "what information science looks like in a feminist universe." The problem is that this universe has particular, and particularly racialized, limits. The principles of free expression and radical copyright on which AO3 is built assume a shared set of values and the power of good faith—but these are insufficient to reckon with the structuring reality of whiteness in fandom. We argue that AO3 falls short on its promise of inclusion as a consequence of the progressive, feminist, predominantly white community whose norms were encoded into the archive. This whiteness becomes visible when the Archive is challenged to take users' experiences of race and racism into account—a task that was long treated as outside the scope of its mission.
[1.4] In the remainder of this essay, we offer a brief history of critiques of fannish racism, before turning to the specifics of AO3, the political commitments embedded in its systems, and the ways that attention to racial justice could transform them. Drawing on fan fiction genres, we offer three potential models for thinking through these possibilities: a fix-it that would extend the Archive's existing metadata structures in new directions; a canon divergence that would alter the makeup of the content on AO3; and an alternate universe that draws from abolitionist organizing to imagine the broadest structural changes of all. In each case, we consider the potential benefits and harms the approach would likely entail. While we offer speculative and exploratory analysis rather than implementable solutions, our work is motivated by this observation and commitment: racial justice is not currently part of fan fiction's digital infrastructure, but it can and should be.
2. Fandom is racist: Fan of color critics' ongoing interventions
[2.1] Fundamentally, media fandom's claims to feminist and queer politics share the limitations of other feminist and queer communities and movements dominated by white participants—they fail to incorporate an analysis of race. As Pande (2018b) has catalogued, to claim that fandom provides empowerment for women often goes along with viewing mentions of race as spoiling the fun, with critics of fannish racism cast in the role Ahmed (2017) identifies as the killjoy to white feminist pleasure. If fandom is often thought of as a "safe space," it certainly isn't so for everyone (Pande 2016). Benjamin Woo (2017, 250) puts a finer point on it, asking "whether fan communities are spaces for people with racialized bodies and identities." We are speaking here of critique by fans in fan spaces, although the academic field of fan studies certainly shares this problem of white feminist and queer critique at the expense of serious reckoning with race (Wanzo 2015; Pande 2018a; 2016; Stanfill 2018). As Mel Stanfill (2020, 126) asks, "To what extent was the narrative that fandom was open and inclusive because the people who set the stage were precisely the educated, middle-class, liberal white people fandom was open to?"
[2.2] Against this dominant trend, nuanced analyses of racism in fandom have nevertheless been developed over a long period. We call this body of thought "fan of color critique" in acknowledgment of its shared commitments with women of color feminist thought and queer of color critique—which have called white feminist movements and white queer studies to account, demanding theoretical framings cognizant of the complex intersections among sexuality, race, class, and gender (Muñoz 1999; Ferguson 2004; Cohen 1997). We use this term to highlight the vital theoretical and activist work being done to name, challenge, and overturn the whiteness of dominant fan cultures and communities and to move racialized fannish spaces and practices from margin to center. While fan of color critique is not necessarily an expression of fans' individual racialized identities, fan critics of color have been particularly subjected to racist silencing and harassment when they have spoken up in both physical and online spaces. In recent years, social media has raised the profile of fan of color critique, resulting both in a certain amount of establishment soul-searching and in increased abuse and violence directed toward marginalized people who raise their voices online (note 1).
[2.3] In order to highlight the longevity of fan of color critique (which seems repeatedly to come as a surprise to white fandom and fan studies) without bringing uncalled-for attention to posts intended for specific communities at specific moments in time, this article cites work that is already in academic circulation, particularly two roundtable interviews with fan of color critics edited by the authors and published in 2009 and 2019 respectively (TWC Editor 2009; Stanfill 2019). While this offers, inevitably, a selection of racial critique in fandom, we believe it is a representative one. We chose to draw primarily from these collective texts because, upon revisiting them for this piece, we were struck by the profound similarities in fan of color critics' analyses of fannish racism across a ten-year period; even as cultures and spaces of transformative works fandom changed radically, their racial structures did not.
[2.4] Fan of color critique tells us that fandom is structurally white. This is not to say that the population of fans is entirely white but rather that white positionality comes to be the default because both dominant Western culture in general and media in particular default to whiteness. As coffeeandink puts it, such groups have "a way of affirming the public space is white, as a space where white concerns are paramount" (quoted in TWC Editor 2009, ¶ 6.3). This default whiteness can also be seen in treating the desires of white fans as universal: "I see a lot of folks in fandom try and say that having a canon white gay couple is a win for the entire community" (Leo quoted in Stanfill 2019, ¶ 7.4). The harms to and concerns of white fans are similarly prioritized: "Fandom, as encompassing women, immediately understands the power of the phrase character rape, while it does not so swiftly regard the words colonial or Oriental with equal visceral horror, even though the effects of rape and colonialism have been equally scarring on women of color" (Deepa D. quoted in TWC Editor 2009, ¶ 5.1).
[2.5] Fan of color critique also demonstrates that fandom has been a space in which racism is actively fomented and perpetuated. As Rebecca Wanzo (2015, ¶ 1.4) notes, it is important to recognize "attachments to whiteness and xenophobic or racist affect as frequently central to fan practices." This has not been sufficiently understood because, when fans are assumed to be the underdog in relation to the media, the way some fans exercise power over others is suppressed. As Eri put it, "the biggest problem is people refusing to acknowledge they are being racist and they should change'' (quoted in Stanfill 2019, ¶ 4.6). Rather than seen for what it is, such behavior is subsumed under antifandom, or generalized dislike of a text or character, which, Emma Jane (2019, 43) argues, "is increasingly being used as a sort of metaphorical fig leaf for preexisting prejudice and bigotry."
[2.6] Especially relevant to fan fiction is fan of color critique's assertion that pleasure in texts is not innocent but rather participates in racism and can reinforce it in sometimes subtle ways. Prior to AO3's formation, fan of color critics had already highlighted the problematic racial dynamics of fan fiction within the community that created and would come to dominate the Archive. Slash fan fiction had long been dominated by stories about white men, with characters of color stereotyped or relegated to the margins in a variety of ways, while stories of racialized trauma became backdrops to white gay romance (Lowe 2019; Pande 2018b). Pande (2018a, 323) calls on us to "pay attention to patterns of fan prioritization so as to identify which characters are consistently valued over others within fandom spaces, even when issues of social justice are highlighted." As Stanfill (2018, 310) describes the defensive response of structural whiteness to this argument, "People write out of pleasure and men of color (who apparently do not bring pleasure) would have to be artificially added out of antiracist duty."
[2.7] Efforts to shift or challenge these dynamics, or simply to tell stories that do not center whiteness, frequently elicit claims to innocence and defensive reactions (Wanzo 2015; Johnson 2019). When fans of color raise these issues, often the framing from white fans is that they are the problem: "It's really common to have our concerns dismissed as us not being grateful enough or being too picky" (Leo quoted in Stanfill 2019, ¶ 7.2). Pande (2016, 213–14) identifies the response to these discussions as "policing, both in terms of content (responses like 'this is not an appropriate topic of discussion here') and in terms of tone (rules like 'please keep this list friendly and supportive')." Discussion of whether dealing with hard issues is disruptive to fan enjoyment assumes people are comfortable and included to begin with—far more true for white fans than fans of color. Fans of color may be subject to "a demand to stay silent in spaces which are dismissive or outright hostile to discussion of these issues" (Pande 2018a, 329). Meanwhile, white fans may respond by recentering themselves. As one fan memorably put it in 2009, this is crying "white women's tears," meaning that after a conflict, "suddenly everyone is comforting the white woman instead of continuing to discuss race" (Sparkymonster quoted in TWC Editor 2009, ¶ 4.2). Such a response assumes the only damage in this process is to white fans asked to consider race, not to fans of color subjected to racism. Overall, it's essential to heed calls to consider not only how fandom is liberatory, but how it is often marginalizing at the same time.
3. AO3 and the politics of platforms
[3.1] AO3 is a quintessential example of transformative works fandom's capacity to simultaneously liberate and marginalize. Users and scholars alike tend to think of AO3 less as a social media platform and more as a community space designed, coded, and managed by the users it serves (De Kosnik 2016; Fiesler, Morrison, and Bruckman 2016). Indeed, AO3 was formed in explicit opposition to social media's emergent economic logics. Its umbrella organization, the OTW—whose other projects include Transformative Works and Cultures—was created in 2007 after fans became outraged at a startup aiming to make a profit from fan fiction. This was also shortly after many fan fiction writers had been ousted from LiveJournal for allegedly obscene sexual expression. One of the founders, Speranza, offered a rallying cry in 2008 that has shaped the Archive's ongoing positionality: "I want us to own the goddamned servers, ok? Because I want a place where…no one can turn the lights off or try to dictate to us what kind of stories we can tell each other" (quoted in Lothian 2013, 547). AO3 was created to serve as a "grassroots archive" that would host the cultural memory of a self-defined community—imagined as a space in which future formations of fan fiction writing could freely emerge without the pressures of monetization or censorship (Lothian 2013, 547). By the end of 2020, with more than seven million works posted by more than three million users in more than thirty-eight thousand fandoms, AO3 had grown beyond the bounds of any single community (Wright 2020). Yet its identity as a "fan-created, fan-run, nonprofit, noncommercial archive for transformative fanworks'' remains strong (https://archiveofourown.org/). This is particularly apparent when contrasted with other large fan fiction hosting sites like for-profit WattPad, which describes itself as a "social storytelling platform" with ninety million users (https://www.wattpad.com/). Nevertheless, AO3's scale means that its current existence cannot be analyzed solely through its origin story or through the narratives it tells about itself—like it or not, AO3 is also a platform. And the question of who is or is not included in the "us" who tell the stories and own the servers remains a vexed one.
[3.2] In addition to its community framing, AO3 is often discussed as a feminist and queer endeavor. De Kosnik (2016, 134) notes that "fan archiving is driven by a political longing: a longing to protect and sustain female and queer communities and cultures." AO3's founders largely belonged to fan communities that understood their practices through this lens. Many fan fiction writers and readers who were AO3's creators, beta-testers, and first users were part of a genealogy that self-consciously traces its origins to the Kirk/Spock slash zines gleefully described by Joanna Russ in 1985 as "pornography by women for women, with love"; participants in a 2007 roundtable called slash fandom a "queer female space" (Lothian, Busse, and Reid 2007). Building on these origins, Fiesler, Morrison, and Bruckman (2016, 2574–75) argue, AO3 has an "underlying commitment to many core feminist values such as agency, inclusivity, diversity, and empowerment." As the participants in their interviews often emphasized, as AO3 was being designed and built, "There was also a broader mission towards making everyone feel welcome in using the archive" (Fiesler, Morrison, and Bruckman 2016, 2579).
[3.3] Such ideals of community and inclusion don't sustain themselves, however. As Tarleton Gillespie (2018, 16) argues, "Communities need care […] to address the challenges of harm and offense, and to develop forms of governance […] and embodied democratic procedures that match their values and the values of their users." In any group, there is the potential for conflict and therefore for exclusion, which must be managed in some way. The values of the people who made AO3 also matter in another way—the design of the platform, and its content moderation, are instantiations of values, "build[ing] a set of possibilities" and limitations into its structure (Stanfill 2015, 1060). Gillespie (2018, 47) argues that community guidelines "constitute a gesture: to users, that the platform will honor and protect online speech and at the same time shield them from offense and abuse." The role of norms in design is often unconscious, but at AO3's inception the question was quite explicit: "Because the controversy that sparked its existence was surrounding a disconnect with the community's value system, baking these values into the design of the site was a priority," meaning the aforementioned values identified as feminist: "agency, inclusivity, diversity, and empowerment" (Fiesler, Morrison, and Bruckman 2016, 2574).
[3.4] This value-conscious design raises two further questions. First, struggles over race and racism in fandom and at AO3 indicate that inclusiveness is not as total as its proponents claim: not everyone feels welcome. As André Brock (2011, 1088) points out, "Racial formation is an undeniable cornerstone of technology: articulated in belief, enacted through certain practices, and embodied in material design." In the absence of active efforts otherwise, AO3 participates in "the norming of technology as a White/human discursive enterprise, where efforts by non-whites to stake out space within the realm are unwelcome" (Brock 2011, 1096). It might be argued that, given the "close alignment of design with the existing values and norms of the fan fiction community" (Fiesler, Morrison, and Bruckman 2016, 2575), AO3 only reflects problems in the community and should not be held accountable for reproducing its whiteness. However, there is precedent for fan archives shaping and not just reflecting values. For example, AO3's design around the issue of copyright and permission "got a little bit ahead of the broader community's internal values'' as "a deliberate concerted decision" to argue that fans should be able to remix one another's work just as they remixed corporate materials (Fiesler, Morrison, and Bruckman 2016, 2583). If similar deliberate, concerted decisions are not made regarding the Archive's racial structures, it will continue to reproduce the unmarked white defaults that allow racism to flow unchecked.
4. A fix-it for the infrastructure: Warning for racism?
[4.1] Though there is a narrative that "anything goes" on the Archive, there are in fact some limitations on what can be posted, and how. In addition to metadata specifying the gender of a story's romantic pairings (a structure that shows the influence of AO3's originating communities), two kinds of labels are required: the ratings tag and the archive warning. Both allow readers to be aware of the nature of a story before they open it—an essential point. The ratings tag marks a story as for general audiences (family friendly), teen and up (not for children under age thirteen), mature (adult themes or language), explicit (sex or violence), or unrated (could be anything); built on Motion Picture Association of America ratings and similar national systems by which film and other media are evaluated, this structure seems designed to evade complaints or litigation from parents angry that their children have been exposed to inappropriate content. The archive warning, the final required element, is a complex and contested item that has been a flashpoint in discussions of racism on AO3.
[4.2] Archive warnings were created in response to community needs and demands that were never unconflicted. As Lothian (2016, 746) writes, "AO3's list of warnings, from which any user posting to the archive must select, stand as a trace of many rounds of fierce and deeply felt debate between some fans who felt that warning for key plot events would ruin their stories' effects and others unwilling to read without finely grained preparation for what lay ahead." Warnings (also called content notes) name issues in a story that are likely to provoke enough of a reaction that some might seek to avoid them—though the structure equally allows readers to seek out a warned-for feature. Those who post a new fic must check a box to state whether it contains "graphic depictions of violence," "major character death," "rape/non-con," or "underage"; they may also state that no archive warnings apply, or mark that they "choose not to use archive warnings." The warnings mark what the archive's creators, and the community for which they were designing, considered to be significant affective (and, potentially, legal) flashpoints. With the exception of the warning for death of a beloved character, these also map onto some of the issues that are most likely to provoke a trigger warning for potentially traumatizing material in nonfannish contexts.
[4.3] The warning most immediately connected to AO3's feminism is the warning for "rape/non-con" (nonconsensual). Feminist critics and activists have long challenged the commonplace appearance of rape onscreen, and the normalization of sexual and gendered violence it signals (Projansky 2001). The #metoo movement on social media made that normalization visible and brought feminist critiques into the mainstream (Jackson, Bailey, and Welles 2019). If AO3's structure takes as self-evident that one should never be asked to encounter rape without warning, that undoubtedly points to a feminist design element, one that embeds a reminder for the Archive's users that the fiction they post can contribute to violent, hierarchical structures that cause real harm. That "non-con," a fannish term used for fiction that derives its erotic charge from play with nonconsensual sexual activity, joins with the starker "rape" here demonstrates an understanding that the effect of such depictions exists regardless of the intentions of creators. (A scene of sexual violence in a gritty, feminist story of trauma and recovery would be expected to elicit the same warning, and have a comparable potential to do harm, as a nonrealistic sexual fantasy in which a character's "no" stands in for a writer and an audience's "yes.") The warning is, or at least is designed to be, nonjudgmental; fictions and fantasies of sexual violence and the violation of consent are written for many reasons, after all, and are frequently created by survivors reckoning with histories of trauma (Busse 2017). Mandating that potentially harmful fic be appropriately labeled (or explicitly marked as unlabeled) allows AO3 to offer readers a way to avoid the harm such fic can cause while maintaining its commitment to full freedom of sexual and other expression.
[4.4] The "rape/non-con" warning is an example of an AO3 structure enabling users to create the distinctions they need to navigate content according to their own priorities, capacities, and risks of harm. But it is limited in the harm it can accommodate. That sexual violence appears self-evidently necessary to warn for is a victory rooted in long and ongoing histories of feminist struggle. But it is also marked by the tendency within those histories to presume that sex- and gender-based violence can be understood in isolation from other social systems—in particular, race. As Ashwini Tambe (2018) writes in her critical race feminist analysis of the "me too" movement and moment, to mark sexual coercion as a form of abuse different from all others presumes a relationship between sex and subjectivity that is not only gendered but classed and raced. Fan fiction writers' long (albeit never fully settled) history of navigating the gendered relationships between consent and coercion, and fantasy and violence, does not easily transfer to other intersecting modalities of power, abuse, and representation.
[4.5] The 2020 open letter demanded a reckoning with racism within AO3's content warning structure. Its authors called for "changes to the Archive's technological structure and abuse policies" in order to "address racist content with the same seriousness they currently accord to violence, sexual assault, and underage content" ("Open Letter to the OTW on Racism in Fandom" 2020). This demand references AO3's existing structural investments in mitigating harm while also repeating calls for an organizational response to racism on the Archive that had been going on for years. As fan of color critique has repeatedly and laboriously demonstrated, racism in fan fiction causes harm. AO3's design allows for structural mitigation of some forms of harm. How could existing tools be mobilized to address this one?
[4.6] If a warning for racist content were to be useful, it would have to take account of the particular ways that race is likely to operate in fan fiction. One thing AO3's current warning policy gets right, in our opinion, is that it allows for the operation of fiction in general, and fan fiction in particular, as the articulation of fantasy. Fantasies, sexual or otherwise, are never apolitical, but neither are they transparent expressions of politics. The expression of a fantasy with problematic politics illustrates the pervasiveness of dominant structures but does not necessarily reinforce them in a straightforward way. The damage comes when others encounter the fantasy's expression. Warning for content mitigates potential harm by giving readers who are navigating a landscape of other people's fantasies a map that they can calibrate to their own needs and desires. However, some content is easier to demarcate. While there are plenty of ambiguous and liminal cases when it comes to fantasies like nonconsensual sex, violence, and intergenerational sex, determination of the presence of an act is likely to be far easier to identify than is a structure like racism. It is aphoristically true that the politics of fantasy are most present when they are least consciously expressed; for racial politics, the structures of white racial innocence mean the most problematic and egregious examples are likely to be produced by those who are the least aware of any problem (Bonilla-Silva 2003).
[4.7] Let us imagine the kind of story to which a racism warning could most obviously be applied. Imagine a fic in which racial trauma forms a backdrop to the fannish objects' romance: perhaps someone who holds the power of white supremacy is idealized problematically by a racialized other. (Many stories on AO3 meet this description; we avoid naming any particular one in order to keep our arguments speculative, focusing on patterns rather than specific instances.) Such a story might find itself the subject of intense critique as a sympathetic representation of a perpetrator of atrocities; its author might seek to defend themself by arguing that they were exploring social and political issues, or focusing on romantic and erotic tropes, or reckoning with histories of violence. Remaining hypothetical allows us to avoid discussing whether such defenses would be justified, focusing instead on the fact that such a story might cause harm irrespective of authorial intent. How could that harm be mitigated through AO3's infrastructure if a warning flag for racist content were in place?
[4.8] The goal of such a warning system would be for the fic not to appear to those for whom it is harmful—since even seeing such a story appearing in search results could feel violent to those harmed by white supremacy. The warning system on AO3 was created to allow users to avoid violence and trauma, and the site's search feature was updated in 2018 to allow explicit exclusion of content on the basis of this and other metadata ("Upcoming Changes to the Search & Filter Functionality" 2018). However, what if the story were, as many are, tagged "Choose Not to Use Archive Warnings," meaning the author has opted out of the warning system for that story? It is worth noting that as of May 2021, almost a third of the works posted to the Archive selected "Choose Not to Use Archive Warnings," meaning that a significant proportion of Archive users are actively refusing the warning system in its current iteration (note 2). This immediately highlights one difficulty—if a voluntary warning system is not widely adopted, it is of little use. In searching for fan fiction to read, users may filter out particular warnings, but in order to have truly reliable results, they would have to exclude a significant proportion of works posted to the Archive. Alterations in the mechanics of "choose not to warn" could open up other possibilities, such as asking writers to actively affirm that certain topics are not present in their fic, or having multiple sets of warnings that would allow writers to more finely hone the areas for which one can "choose not to warn." It might also be possible to make the option of hiding fics that do not use the warning system more visible to new or unfamiliar users, through either tutorials or the default search options.
[4.9] What else might a warning system for racism require? How could such a system serve the needs of diverse readers, writers, and Archive administrators who would come to it from a wide range of entry points? In first thinking through these questions, we wondered whether it would make sense to incorporate some way to highlight intentional engagement with questions of racial justice, enabling writers to mark when they are consciously touching upon sensitive areas. However, we can expect neither that intent precludes harm nor that there will be consensus over the effectiveness of an intentionally critical engagement. Better, then, to draw upon the language of the existing warnings and focus on what is concretely depicted, removing the question of intent. Choices would have to be made about how specific that would be: "Story depicts racist violence"; or "Story depicts prejudice and/or discrimination"; or "Characters use racial/ethnic slurs." More specificity would make readerly choice easier, while making Archive use more unwieldy, given the size of the list that would be required to account for every eventuality and the potential harm in demanding that writers confront detailed lists of prejudice and violence each time they post. Broad tags allow for previously unimagined possibilities while also being so open to interpretation as to risk incomprehension or misuse.
[4.10] Decisions about the structure of warnings would be complex and contested, since to some unavoidable extent they involve the impossible task of deciding upon the commensurability of systems of oppression. We do not map out what a racial warning infrastructure might look like here, as decisions and designs should be informed primarily by those who are most affected. However, we would say that the first step must be to look at how fans do this work already, informally (paralleling the process by which the existing AO3 warnings were made infrastructural). Fans are already curating their own experience by focusing on trusted authors, sharing recommendations, and building community through comments and social media. For some, this might be enough, and even preferable to a systemic structure. For others, as the repeated calls for change have made clear, it will not be. Any effort to implement a warning policy for racist content would need to look in depth at how current warnings are operating in practice. How often are users' warnings challenged in complaints/reports and how often are such complaints/reports upheld? Who is doing this labor and is it sustainable? This is especially necessary because any change to the system is going to elicit enormous pushback to which AO3 and its users will have to respond.
5. Content moderation as canon divergence: Can racist content be banned?
[5.1] It should be clear from the above analysis that while a warning policy to deal with racism and other forms of structural oppression in fan fiction might be necessary, it would not be sufficient. What else might be needed, and what else might be possible? In its response to the 2020 Open Letter, the OTW opened with a claim that "AO3 was designed specifically with maximum inclusivity of content in mind, and we remain committed to that principle" (Jess H. 2020b). The OTW was to a large extent founded in order to improve the situation for fans whose creative and sexual expression was excluded by hosts who found it legally risky; hence, its policy "to never practice censorship or deletion" (De Kosnik 2016, 93). This tracks closely with what legal scholar Mary Anne Franks (2019) calls First Amendment fundamentalism, an approach to freedom of speech that forefronts more speech as the essential value—even, she points out, speech that chills other speech. However, unlike other sites known for hands-off approaches to moderation, "those behind AO3 are sensitive to the fact that their inclusive policy means that there is content on the site that many people may not want to see" (Fiesler, Morrison, and Bruckman 2016, 2582–83), as shown by the warning system. Research on hate speech is useful here, as it explains that the costs of speech are unevenly distributed, borne not by society at large but by the targets. Mari Matsuda (1993, 18) calls this "a psychic tax imposed on those least able to pay." Charles Lawrence (1993, 80) notes that in framing hate speech through free speech "we ask Blacks and other subordinate groups to bear a burden for the good of society—to pay the price for the societal benefit of creating more room for speech. And we assign this burden to them without seeking their advice or consent. This amounts to white domination, pure and simple." While AO3 content rarely (though not never) constitutes hate speech, the uneven burden is similar.
[5.2] A potential solution, which some commenters on AO3's response demanded, would be to ban harmful content—thereby making the Archive safer for those at risk, who would not be reliant on the vagaries of a warning policy. Under this model, the hypothetical story we discussed in the previous section might be taken down from the Archive. This is ultimately a question of content moderation. Content moderation is a topic most commonly considered in the context of social media like Facebook, Twitter, and so on, but AO3, as a site with content that is provided by its users, faces the same challenges:
[5.3] Social media platforms need rules that can be followed, that make sense to users, that give their policy team a reasonably clear guide for deciding what to remove, that leave enough breathing room for questionable content they might want to retain, that can change over time, and that will provide a satisfactory justification for removals if they're disputed, whether by users themselves or in the glare of public scrutiny. (Gillespie 2018, 45)
[5.4] As the tension between permissiveness and harm already begins to suggest, content moderation is no easy task. Even for social media corporations with billions in revenue, "This is an exhausting and unwinnable game to play for those who moderate these platforms, as every rule immediately appears restrictive to some and lax to others" (Gillespie 2018, 73). In addition to the impossibility of pleasing everyone with a single set of rules, "Moderation is hard because it is resource intensive and relentless; because it requires making difficult and often untenable distinctions; because it is wholly unclear what the standards should be; and because one failure can incur enough public outrage to overshadow a million quiet successes"(Gillespie 2018, 9). A content moderation model for AO3 focuses on what happens when racist fic is posted: readers report or moderators identify it, and it is taken down or, if it is particularly egregious, the user is banned. This is already how AO3 deals with violations of its terms of service (https://archiveofourown.org/tos_faq).
[5.5] However, the costs of content moderation for the moderators, and the unevenness of those costs along familiar lines of racialized and gendered hierarchy, are well documented. Roberts (2019) describes the damaging, traumatizing work conditions of those charged with protecting commercial platforms' brands from the violence of the unfiltered internet. Connecting (poorly) paid with unpaid moderating labor, Nakamura (2015, 106) describes the "hidden and often-stigmatised and dangerous labour performed by women of colour, queer and trans people, and racial minorities who call out, educate, protest, and design around toxic social environments in digital media" as "the engine that powers the internet." She focuses on the ways that women of color, from early social media communities through Twitter, have worked against the sea of racist, sexist content that floods digital spaces, even as their efforts are in turn "unwanted, punished, and viewed as censorship" or "accused of 'policing' social media, of lacking a sense of humour, and of imposing 'pc' values on other users by protesting misogyny, racism, and homophobia when they see it" (Nakamura 2015, 111). The labor Nakamura describes, and the abuse received in response, describes the experiences of many of the fan of color critics we have been citing throughout.
[5.6] Combining the informal work of resisting racism with formal content moderation presents intense and conflicted questions of labor and abuse. Who, after all, would determine which fics were harmful and which themes would be banned? It would have to be fans with insider knowledge: there is a difference between a Black fan writing fan fiction to reckon with and reimagine a stereotypical and racist portrayal and an unthinking reproduction of that same portrayal, which could prove very difficult to separate for an AO3 volunteer armed only with a basic understanding of what the stereotype is. In other words, any effort to ban all harmful content would likely demand a vast investment of labor from the very fans who are likely to be most harmed by that content—at least in our current conjuncture where few white people are deeply educated about race and racialization. Nor, as even the shallowest experience within scholarly and activist spaces will show, are individuals who share the same marginalized identity and the same broad politics especially likely to share the same interpretation of any given text. Moreover, arguments about racist content in stories frequently home in on the author's identity, asking whether certain stories can be told by those who aren't part of the affected group, but even if this logic were unassailable and universally applicable, it would not solve the problem. After all, it is not always straightforward to determine whether an author writing (as fan fiction authors almost always do) under a pseudonym is part of an in-group or not, especially in a context where naming oneself as part of an often-targeted group may be a fraught decision. And regardless, as queer of color scholars have discussed, racist, sexist, and homophobic representations are frequently consumed and even produced by those targeted by those structures who are working on, through, and against them through fantasy that does not always look like good representation (Cruz 2016; Raymundo 2019; Rodríguez 2014).
[5.7] Furthermore, this all assumes that the system would be used entirely in good faith and not weaponized for abuse and harassment—an impossibility, given that all such systems, including AO3's current reporting system, are misused in this way (often with racist intent and outcome). In the comments to the open letter, someone said "incest, pedo, and racist fics should be banned"; taboo sexual expression and reinforcement of white supremacy are already being lumped together in ways that any effort to challenge racist representation would have to tease apart, given the likely targeting of women, queer, and trans people of color by respectability policing. Thus, a content moderation solution would likely diminish the amount of racist content, but at a steep cost.
6. Imagining an alternative universe: What if fannish racism were abolished?
[6.1] In working through the significant challenges of both warning and banning, we don't mean to imply that the problem of fannish racism is insurmountable. However, the challenges show the limits of available solutions, highlighting the need for a radical reimagining. Nakamura (2015, 112) argues for the recognition of moderator labor as part of the structure of the internet; but more than that, she suggests that women of color's "critical counter speech" "outside of capital" may open the possibility for a "vital and resurgent space for new styles of community" that would form an alternative to commercial social media. We find this an exciting and generative prospect for the Archive. What models of moderation might operate (and perhaps already are operating) if we lean in to the "community" side of fandom while also leaning away from its whiteness?
[6.2] The experiences of "Melinda," one of the moderators Roberts (2019) interviewed, are illustrative. While Melinda's experiences in commercial content moderation during the late 2000s were as violent as one might expect, she speaks warmly of the moderation culture in a smaller, explicitly antiracist online community at the same time. In that space, "where people self-selected and where the conversations were very tightly moderated and very clearly bounded by guidelines and rules adhered to by all participants," "if somebody was given a warning we had resources that they could go look at to understand why they fucked up and how not to do it again in the future." The moderation worked because shared values were explicit: "people were there because they chose to be in a community that was actively about White people unlearning their racism," and "there was a great deal of personal investment" in that ongoing work (Roberts 2019, 165–67). In this community space, posts or statements that supported or reproduced structural racism were not acceptable, but when they were made, they were not deleted out of hand; the poster was encouraged and invited to participate in self-education to more fully understand the antiracist norms to which they had explicitly committed themselves when joining the community.
[6.3] Melinda's community and others like it were active at the same time AO3 was founded, and fan of color critics saw connections between the work of education they carried out and the transformative works of media fandom. In 2009, when science fiction and fantasy fandom was experiencing an explosion of conflict about its pervasive racism, Deepa D remarked that transformative works fandom had a strong set of tools for confronting racism because of its community moderation practices:
[6.4] While fic is archived and vids are shown publicly, in fandom, there is also always the idea that it's possible to rewrite things. Fans who are involved with the production of transformative works are used to discussions where a community can constantly reinvent itself, where it can self-correct if things are going wrong. You can edit something that you've posted online…It's ironic and ridiculous in some ways, but the fact that we know we can change things makes us more responsible about changing things. We recognize that things are permanent when they're on the public record, that anyone can take a screencap and save it, but because things are constantly changing, it's our duty to document that change. We recognize that we self-correct, that sometimes that's for good and sometimes for bad, and we try to acknowledge that power and create a structure that takes account of it. (quoted in TWC Editor 2009, ¶ 5.7–5.8)
[6.5] Deepa D's sense of transformative works and their digital archives as rife with the possibility of structural change is one that we think is worth holding onto.
[6.6] In Melinda's recollection and Deepa D's assertion, we see the possibility of a digitally manifested social and interpersonal infrastructure in which racism might not only be held at bay by shutting out its most egregious examples (a necessary precondition) but where it could be actively and constantly contested through everyday practices of accountability and education. In thinking through the ultimate goals a fan of color feminist digital infrastructure might work toward, concepts from abolitionist social movement thinkers may offer a useful alternative to the free speech frameworks on which AO3 has so far relied. Police and prison abolitionists, whose ideas entered mainstream conversation in unprecedented ways with calls to "defund the police" in the summer of 2020, argue that "prisons, policing, and the criminal punishment system in general are racist, oppressive, and ineffective" (Kaba 2020). Abolitionist visions of racial, gender, and disability justice demand the unmaking of state systems that, while promising protection that rarely arrives, perpetuate murderous violence against marginalized people. AO3 is not a state, but the specter of policing is often raised in discussions of its lack of racial justice. Advocates of unrestricted free speech feel that calls to moderate racist content police their expression and unfairly punish transgressors, while fan of color critics find themselves simultaneously cast as the police and policed in turn for the tone or content of their criticisms. An abolitionist vision for AO3 might suggest a new framework in which to see these conversations.
[6.7] Abolitionist adrienne maree brown (2020) is among many who have argued that police, prisons, and the carceral state are internal as well as external structures that get reproduced in community and interpersonal interactions even among those who are committed to their dismantling in the world. She states that the "toxic substance" of intersecting systems of racial, gender, ableist, and settler colonial supremacy produces and maintains a "systemic practice of harm" that is both institutional and interpersonal and that cannot be ended "by isolating and picking off individuals" but that must be acknowledged, confronted, and reimagined as a continual practice (sec. 87). Her visionary hope is for the creation of "small pockets of movement so irresistibly accountable that people who don't even know what a movement is come running towards us, expecting that they will be welcomed, flawed and whole, by a community committed to growth; knowing that there is a place in this violent, punitive world that is already committed to, and practicing, a healing and transformative iteration of justice…[with] a low bar for entry and a high standard for conduct" (sec. 111). What if AO3 tried to be that kind of place?
[6.8] brown's (2011) argument relies on a "we" that names active social justice movements centering Black, Indigenous, and queer and trans knowledge and life. Fandom is currently far from meeting that description, as we have shown; it remains a predominantly liberal, predominantly white space in which even condemnations of explicit racism can be controversial. But neither is fandom's structural whiteness something that is irrevocable. To imagine a broad fandom community aligned with the movement from and to which brown speaks is to speculate, but not without some grounding in reality. Within fandom, and in particular within what we have named fan of color critique, are networks and spaces whose values and practices are indeed actively contesting structural supremacy. What if they displaced the norms that have shaped fandom infrastructures thus far? Fandom has been a pedagogical space for many—cisgender women have found the capacity to express their sexuality freely, queer and trans fans have found community, and amid a larger racist sea, fans of color and white fans have become educated about race and racialization in the contexts of fan of color critique. AO3 contains the capacity for self-correction Deepa D identified. If we gave up the narrative that fandom is somehow inherently progressive and instead asked what it would take for the myriad diverse fandoms to develop a shared accountability for their role in reproducing and their capacity to reimagine systems of supremacy and structural harm, what might become possible?
[6.9] The open letter against racism on AO3 called for paid expertise to be deployed. If that expertise interfaced with the global reach and diversity of AO3, what questions might it raise, and what transformations might it suggest, about race and colonization, antiblackness and the carceral, across fannish contexts? Rather than finding and excluding racist content, the project of racial justice in AO3 could become one of community accountability and ongoing pedagogy, teaching and learning how creators unthinkingly reproduce white supremacy and how they may avoid doing so.
7. Conclusion: What it takes to reimagine
[7.1] Our discussions of warning labels and content moderation practices aim to offer ways into some concrete possibilities that may be stopgaps in a movement to structurally shift the political assumptions embedded in AO3's structure. Only one thing could solve AO3's racism problem, and that would be if fandom truly lived up to its progressive self-image and adopted the practice of ongoing antiracist and decolonial struggle. On a global scale, this is not much more likely than the overturning of white supremacy writ large—but one thing it is vital to learn from those with a lifetime in that struggle is that the dismissal of freedom dreams as unrealistic is a weapon in the arsenal of the status quo. We should see the ongoing struggles of fan of color critique and the fact that changes, if slow and sometimes disappointing, do result, as signals of what could become possible if these voices were no longer outliers and if the full implications of their critiques (which are not all the same and not all in agreement) were listened to seriously and acted upon collectively. Some such changes to fan communities resulting from fan of color critique have already begun, as scholars such as De Kosnik (2016) and carrington (2016) have documented. However, our speculations on radical transformation should absolutely not be understood as an abandonment of the urgency for action within AO3's current structures of content moderation and warning labels.
[7.2] If either content moderation or community accountability models were to be successful, we would end up with a set of fictional expressions that would no longer be acceptable on AO3; the "maximal inclusivity" to which it has long committed would encounter explicit limitations. This may be a tough pill to swallow for some, but we hope we have shown some ways that "maximal inclusivity" already operates to implicitly exclude when it is not accompanied by a commitment to minimal harm. In limiting search capacities to allow Archive users to hide their works from users who are not logged in, AO3 already structurally acknowledges that not everything needs to be in public. Moving stories perceived as racist out of the Archive or limiting their searchability does run many risks. In addition to the potential abuse of any system, it could lead to self-perpetuating enclaves that can breed the white supremacist politics of grievance. Nevertheless, we think that some version of such an effort must be tried. And we think that, along with the risks, possibilities would be opened by it even for those whose works get challenged. Perhaps, for example, making someone realize that they would have to go to such an enclave to have their fiction presented in the way that they prefer would be of pedagogical value—if not for the writer, for the reader who must contemplate what they are seeking out and why, rather than allowing damaging ideas to pass unnoticed.
[7.3] The 2020 open letter was spread with great rapidity. Response from a large and distributed organization is unavoidably slow and hesitant in comparison. Demands are easy and promises are hard, especially when we are talking about implementation of significant changes to a large, complex system run on volunteer labor in the middle of a global pandemic. But it is easy to get stuck on the "how"—a problem with any argument that requires us to change the fundamental terms of a system's operation. That feeling of stuckness often comes from defensiveness on the part of those who are asked to change. It plays into a fear of being critiqued that is part of how structural white supremacy gets reproduced even by well-meaning, well-educated people who wish to support racial justice. The way to become unstuck is not to suddenly imagine perfect creative solutions but to acknowledge that such solutions are not ever likely to be available. Discomfort and harm are not avoidable, but the directions they take and the likelihood of who will experience them most can change. And their direction can be changed through attention to the ways that those who have avoided discomfort and harm so far may be perpetuating or even perpetrating them on others.
[7.4] The political commitment to abolition operates as a means of imagining otherwise while acting in the everyday—a horizon against which to measure every potential change. We suggest that even a fan fiction archive can work toward this horizon and can develop other radical horizons toward which it wishes to move. The OTW was once, after all, calling for an imaginary radical change when its founders demanded that creators of fan fiction own the servers on which their works were published. Now it does and must hold the responsibility that comes with that power, no longer able to claim the position of the marginalized and accountable to the marginalization perpetuated by its own structures. The question of what AO3 should do is not one that we seek to answer; plans for concrete action should come from a collective process led by Black fans, Indigenous fans, fans of color, and nonwhite fans with diverse and transnational perspectives. But everybody with a stake in the Archive has transformative work to do.