“Obscene,pornographic,or otherwise objectionable”: Biased definitions of sexual content in video game live streaming

Live streaming, platform politics, and the slipperiness of sexual content

Video game live streaming is rapidly growing in popularity. It is a booming business for both the technology industry and content creators and an increasingly robust area of research. Streaming already represents a prominent part of the contemporary cultural landscape around video games, digital entertainment, and online cultures. Loosely defined, video game live streaming is the practice of broadcasting video game play in real time, often to audiences of viewers who can interact with streamers and each other. In addition to streaming their video game play, streamers often stream video footage and audio of themselves playing. Streaming of this sort takes place on a variety of platforms. At the end of 2017, Twitch, currently the most popular platform for video game live streaming, reported that it had more than 2 million streamers actively broadcasting each month and more than 15 million viewers visiting the platform each day (Perez, 2018). Live streaming also represents a growing area of interest in academic research. This is because of live streaming’s increasing popularity but also because it offers a valuable window onto larger issues related to digital media and digital cultures. As Taylor (2018) writes in Watch Me Play, “Game live streaming intersects many contemporary issues not only around media transformations but also larger considerations of cultural production and everyday users” (p. 11). Existing research on video game live streaming has addressed a number of topics, such as how streaming fosters participatory communities (Hamilton et al., 2014), how experiences with disability and mental health impact streamers (Johnson, 2018), and how streaming necessitates forms of affective labor (Woodcock and Johnson, 2019).

Increasingly, scholars are turning attention toward issues of identity, diversity, and marginalization in live streaming. Kishonna Gray (2017) has demonstrated how whiteness and masculinity are imagined as defaults in the cultural discourse that surrounds Twitch. On Twitch, women streamers are frequently subjected to sexual harassment and other expressions of gender-based biases. Women are often the targets of challenges to their credibility (Romine, 2019), tokenism based on gender stereotypes (Witkowski, 2012), and comments about their physical appearances (Nakandala et al., 2016; Ruberg et al., 2019). The predominant demographic of those who participate on Twitch, either as streamers or viewers, is still young white men from North America (Consalvo, 2018: 84). For this reason as well as others, video game live streaming is closely tied to what Mia Consalvo (2012) has termed “toxic gamer culture,” which promotes discriminatory attitudes around video games, especially toward women, LGBTQ people, and people of color. Video game live streaming can also be considered in the context of research about other types of online streaming, such as Sarah Banet-Weiser’s (2012) work on post-feminism and “lifecasting” and Theresa M. Senft’s (2008) study of “camgirls.” This related work serves as a valuable reminder that, while video game live streaming may be on the rise, many of the issues surrounding live streaming are tied to Internet technologies and their cultural impacts more broadly. It also suggests connections, which often go unaddressed, between video game live streaming and other forms of gendered and/or sexual digital streaming performances.

Yet, it is not only the participants in video game live streaming who determine its characteristics. Digital platforms themselves have a strong hand in shaping the activities and cultures of their users. In addition to participating in conversations about live streaming and video game culture, this work also builds from and contributes to another growing area of scholarship: research on the structures, policies, and politics of digital platforms. Many researchers in this area have focused on how platforms themselves, as corporate and legal entities, regulate content. In Custodians of the Internet (2018), Tarleton Gillespie describes how online platforms seem to offer users opportunities for “free” expression, but in fact enact top-down control. The work of content moderation itself can have devastating effects for those employed by social media sites (Roberts, 2019). This existing work demonstrates the importance of attending not only to users and content as they behave or appear on digital media platforms, but also to the policies and practices that operate behind the scenes of those platforms.

One of the key ways in which platforms regulate their content and shape their cultures is through policy documents, such as community guidelines and terms of service. Almost all digital platforms publish documents of this sort, though the details of these documents differ from platform to platform. Documents such as community guidelines, even when they are presented as tools for improving a platform’s culture, can have a disproportionate or even discriminatory effect on women, people of color, and others who are already marginalized in digital spaces. This is especially true for the elements of these guidelines that relate to sex, sexuality, sexualized bodies, and by extension gender. In the context of video game live streaming, researchers have shown how community guidelines that regulate streamer attire can reinforce biases against women streamers and privilege the perspectives of straight, male users (Ruberg and Cullen, 2019; Taylor, 2018; Zhang and Hjorth, 2019). The guidelines of other social media platforms have produced similar effects. Influenced by the passing of the Stop Enabling Sex Traffickers Act, or “FOSTA-SESTA” (Romano, 2018), sites like Tumblr, Instagram, YouTube, and the video game distribution service Steam have all pushed to remove or limit sexual content shared by users. Many commentators have critiqued these sites’ definitions of sexual content as ill-conceived and biased against women, LGBTQ people, and sex workers (Grayson, 2019; Hale-Stern, 2018; Hunt, 2017). Thus, the issue of how digital platforms define sexual content has implications that echo across the Internet.

Whether in video game live streaming or other media forms, the definition of sexual content is itself far from self-evident. Across the history of popular media, this definition has been the subject of numerous cultural and legal debates. In its terms of service document, as described at more length below, Twitch characterizes sexual content as “pornographic, obscene . . . or otherwise questionable.” The meanings of these terms are subjective and constantly shifting. Indeed, both “pornographic” and “obscene” are words whose meanings have been demonstrated by historians and legal scholars to be highly context-dependent and reflective of hegemonic and often problematic ideologies. The very concept of pornography is culturally constructed, as Lynn Hunt (1993) states in the Invention of Pornography. Fittingly for a consideration of platform policies, Hunt explains that definitions of pornography are themselves inextricable from how state and institutional forces attempt to regulate content creation. She writes,

Although desire, sensuality, eroticism, and even the explicit depiction of sexual organs can be found in many, if not all, times and places, pornography as legal and artistic category seems to be an especially Western idea with a specific chronology and geography . . . Pornography was not a given; it was defined over time and by conflicts between writers, artists, and engravers on the one side and spies, policemen, clergymen and state officials on the other side. Its political and cultural meanings cannot be separated from its emergence as a category of thinking, representation and regulation. (Hunt, 1993: 10, 11)

Definitions of obscenity are just as slippery as definitions of pornography. Many media objects that may today strike us as uncontroversial were previously put on trial for obscenity: from avant-garde French cinema to 1950s broadcast television to “classic” novels (Morrow et al., 2016: 377–381). Far from establishing concrete norms for what constitutes pornography or obscenity, these cases have demonstrated the impossibility of establishing objective measures for defining sexual content. As Justice Roth famously stated in his ruling for the 1964 supreme court case Jacobellis v. Ohio, what makes a piece of media pornographic is fundamentally a matter of interpretation–or, in Roth’s words, “I know it when I see it” (Morrow et al., 2016: 381).

Digital content production brings these challenges of defining sexual content, and their related ideological questions, onto emerging 21st century media platforms. The case of video game live streaming, in particular, highlights the messy and culturally constructed nature of these definitions, as platforms like Twitch rush to keep up with the regulation of millions of channels of multi-layered, user-generated content. Making the regulation of content on a streaming service like Twitch all the more complex is the instantaneous nature of its production. With an overwhelming number of streams broadcasting live, and millions of viewers participating in real-time chat, Twitch must rely on a network of moderation techniques that extend the boundaries of traditional community guidelines. Many streamers have human or computer moderators who intervene when users behave inappropriately, for example. In addition, due to the live nature of the streaming medium, Twitch cannot retroactively censor content. It is precisely what makes video game live streaming so compelling for many streamers and viewers—the opportunity for self-expression at a rapid pace and on a grand scale—that also prompts platforms like Twitch to focus such considerable effort on regulating its content. Regulating all Twitch content is indeed an impossible task; the result is that these regulations are unequally applied, leaving already marginalized streamers most vulnerable to censure. Meanwhile, streamers and commenters from dominant groups remain comparatively free to broadcast problematic content. When such streamers do experience consequences, such as bans for using homophobic language, they are often given public platforms for their indignation (Alexander, 2018), sparking absurd, entitled “debates” about whether hate speech should be considered offensive within the context of online culture in the first place.

Analyzing the policy documents of digital platforms

Understanding and unearthing how Twitch defines sexual content requires an analysis of a network of policy documents. The particular documents I discuss here have been selected either because they directly address the issue of sexually explicit or sexually suggestive content, both of which are prohibited on Twitch, or because they contain information about procedures and consequences for user violations related to sexual content. In working with these documents, I have used the humanistic method of close reading as an entry point into a critical cultural analysis that reveals both the patterns in how these documents define sexual content and the broader implications of these definitions. Rather than studying live streaming or gameplay practices themselves, my object of study is a subset of the “paratextual” artifacts (Consalvo, 2007) that surround live streaming. In this way, my methods are similar to those found in work by T. L. Taylor (2018), Tarleton Gillespie (2018), and Ge Zhang and Larissa Hjorth (2019), all of whom have conducted research that draws from analyses of digital platforms’ community guidelines as a way to interrogate the cultural politics of these platforms. As part of the practice of intersectional feminist critique, I also recognize the particular, situated perspectives and privileges that I bring to this research as a white, gender non-binary, queer tenure-track faculty member with a long history of experiencing identity-based harassment in online spaces.

These documents represent an interconnected network of textual data points: webpages that frequently link to and reference one another. Specifically, my analysis encompasses the following six documents:

Together, these webpages constitute Twitch’s current public-facing documentation regarding restrictions on sexual content.

Analyzing materials of this sort, which are dynamic and can be updated by the platforms that publish them at any time, poses unique challenges. Community guidelines and related policy documents are subject to regular changes. Twitch’s community guidelines document itself opens with the following disclaimer: “This is considered a living document that we regularly update based on the evolution of the Twitch community and service.” Therefore, by nature, any discussion of a platform’s community guidelines represents a snapshot: a consideration of how these guidelines appeared at a specific moment in the history of the platform, with implications that speak out across the platforms ongoing history. In analyzing Twitch’s community guidelines, I am referring to what is, as of this writing in July 2019, the latest iteration of this document, which was published on 14 November 2018. Twitch has also archived older versions of their community guidelines from 2017 and 2018. Though the other documents listed earlier (such as Twitch’s terms of service and its statement on “Nudity, Pornography, and Other Sexual Content”) have surely also been updated over the lifetime of the platform, Twitch does not provide public-facing documentation of these updates.

For the purposes of this article, I have limited my analysis to policy documents from Twitch. Twitch is not the only online platform that supports video game live streaming. Other game-focused platforms include Caffeine and Mixer; Mixer in particular is currently growing in popularity and public attention. In addition, there are a number of platforms that support live streaming as part of a suite of other features, such as YouTube and Facebook. However, Twitch is still currently the largest platform for video game live streaming. Because of its size and the reach of its influence, both cultural and commercial, Twitch is a valuable focus for analysis (Taylor, 2018: 10). At the same time, my decision to limit my initial research to Twitch’s documentation is pragmatic. Narrowing in on my object of study has allowed me to dig more deeply into a qualitative analysis of these documents and to deconstruct the cultural implications of their contents. Future work might expand on this research by comparing Twitch’s policies to those from other platforms, tracing the evolution of Twitch’s policies over time, or interviewing streamers about how they have been affected by Twitch’s nebulous definition of sexual content.

Each of the documents analyzed here contributes to the overall functional definition of sexual content as articulated by Twitch. In the sections that follow, I group my analysis according to important elements that emerge across these documents. I begin by identifying where and how Twitch’s policy documents engage with the question of sexual content, with a focus on how that this content is categorized and assessed. I, then, map out key concepts that recur in these documents, including obscenity, pornography, and nudity, and demonstrate the slipperiness of their definitions. In critiquing these documents, I highlight the ways that they are structured around contradictions, double standards, and notable absences, such as the absence of engagement with issues of sexual harassment. Finally, I draw out the underlying cultural logics that shape these documents, which are framed through problematic rhetorics of community protection and common sense. In conclusion, I discuss how the definitions of sexual content found within these documents reflect and perpetuate bias.

Categorizing sexual content: explicit, suggestive, violent

A prominent feature in Twitch’s policy documents is the categorization of different types of sexual content. This can be seen, for example, in the “Community Guidelines” document (Twitch, 2018). Twitch’s community guidelines cover a range of topics, most of which relate to the types of content or behaviors that are not allowed on the platform. Within these community guidelines, the issue of defining and regulating sexual content is addressed in a subsection labeled “Nudity, Pornography, and Other Sexual Content.” This subsection lists three categories of sexual content: content that is sexually explicit, content that is sexually violent, and content that is sexually suggestive. “Nudity and sexually explicit content or activities, such as pornography, sexual acts or intercourse, and sexual services, are prohibited,” the document states. In this sentence, sexually explicit content is defined as pornography and/or the representation of sex acts or sex work (as indicated by the phrase “sexual services”). Next, this subsection addresses sexual violence and child exploitation. Here, the guidelines state, “Content or activities that threaten or promote sexual violence or exploitation are strictly prohibited and may be reported to law enforcement.” By inserting this statement between descriptions of sexually explicit and sexually suggestive content, the community guidelines document makes the not-so-subtle suggestion that sex and sexuality are closely related to violence and abuse. Finally, this subsection addresses sexually suggestive content, stating simply that “sexually suggestive content or activities are also prohibited, although they may be allowed in educational contexts or for pre-approved licensed content.” Thus, the definition of sexually suggestive content remains vague. Unlike for sexually explicit and sexually violent content, the community guidelines provide no examples to illustrate what Twitch does or does not consider sexually suggestive content.

Linked to from the community guidelines is a supplemental document titled “Nudity, Pornography, and Other Sexual Content,” which addresses Twitch’s policies on sexual content in more depth (Twitch, n.d. d). This document contains information on a number of topics, with subsections dedicated to topics like nudity and attire, as well as expanded writing on the categories of content described earlier. The document opens with the statement: “We restrict content that involves nudity or is sexual in nature, and are committed to ensuring that Twitch is not used for sexual exploitation or violence.” Though the “Nudity, Pornography, and Other Sexual Content” document represents the most substantial text within the network of Twitch’s policy documents for addressing sexual content, its description of what constitutes explicit sexual content remains vague and broad. Further down in the document, sexually explicit content is described merely as entailing “pornography, sexual acts, and sexual services, including solicitation and offers for such content.” According to this document, background imagery can also count as explicit sexual content; streamers are prohibited from broadcasting in areas where “nudity or sexual activities may be taking place, even if such conduct or activity is not at the direction of the broadcaster.” However, no further explanation is provided for what constitutes “nudity or sexual activities.”

By contrast, the subsection in the “Nudity, Pornography, and Other Sexual Content” document on sexually suggestive content acknowledges that more information may be needed for Twitch’s moderators to determine what counts as sexual content. Here, the document states that, when reviewing potentially sexual content, moderators will consider the following factors:

This extensive list of the factors speaks to the complexity, challenge, and active work entailed in defining such sexual content—and therefore the extent to which, far from being obvious in nature, it remains open to varied or subjective interpretation. Strikingly, this list does not tell readers what about these factors moderators look for (e.g. what does or does not represent an appropriate camera angle?). In addition, this list reveals that definitions of sexually suggestive content are inextricable from the features of streaming itself. How a streamer frames a shot, what kinds of metadata they use, and even what emotes they make available to their viewers can all influence whether their content is deemed sexually suggestive. In this sense, definitions of sexual content are inextricably tied to the technical infrastructures and affordances of the streaming platform, which are interpreted as carrying their own cultural meanings.

In addition, these guidelines for assessing sexually suggestive content reveal the (by nature hidden) labor that is demanded of streamers who must work to distance themselves and their streams from associations with sexual content. Regardless of her performance or appearance, a woman streamer is likely to receive sexual comments from viewers. These comments will publicly appear on screen if they are not caught quickly enough by the streamer or her moderators. If she does not have the resources to quickly remove these comments—for example, if she does not have effectively tailored moderator “bots” or a group of her own trained, active chat moderators—she herself could be reprimanded by Twitch’s higher level moderators, since the platform’s guidelines dictate that “what chat messages [streamers] permit in their community” can be used as a factor in removing streamed content or banning a streamer. Under these guidelines, the burden falls to streamers themselves (and disproportionately to streamers who are the objects of sexual commentary) to fend off sexually suggestive remarks or face the consequences. In this way, Twitch’s guidelines present streamers as having complete control over their streams, while overlooking the fact that moderating user behavior effectively is much harder for some streamers than others. These guidelines in fact remove a layer of agency from streamers, placing it instead in the hands of sexist or otherwise biased viewers, who have the power to make suggestive remarks for which streamers are then held accountable.

Obscenity, pornography, and nudity

The language of Twitch’s policy documents relies on a constellation of terms that are deployed to differentiate acceptable and unacceptable content. Prominent among these terms are obscenity, pornography, and nudity. The use of these terms can be seen, for instance, in Twitch’s “Terms of Service.” Like the community guidelines, the “Terms of Service” document articulates the rules and regulations of the platform (Twitch, n.d. e). By comparison, it is long and legal in tone, whereas the community guidelines are written for a more general readership. Many of Twitch’s other policy documents refer and link back to the terms of service, making this a core text for establishing policies and definitions on the platform. Despite this, the topic of sexual content is addressed only briefly in this document, often with strikingly little explanation of the language it uses—particularly worrisome, given the power of the terms of service document itself, which describes itself as “a binding contract to which all users implicitly agree.”

The issue of sexual content appears in the terms of service document in a section labeled “Prohibited Conduct.” Here is found a brief yet wide-reaching clause that prohibits users from creating, uploading, or distributing content that is “inaccurate, unlawful, infringing, defamatory, obscene, pornographic, invasive of privacy or publicity rights, harassing, threatening, abusive, inflammatory, or otherwise objectionable” (emphasis mine). Whereas other documents expand on different types of sexual content, in this arguably most official of policy documents, sexual content is represented only through the words “obscene,” “pornographic,” and “otherwise objectionable.” No examples are provided to illustrate what Twitch considers obscene, pornographic, or objectionable. Elsewhere in the document, a reference to obscenity and pornography appears again in a clause that states,

[Twitch] assumes no liability for any User Content or for any loss or damage resulting therefrom, nor is Twitch liable for any . . . obscenity, pornography or profanity you may encounter when using the Twitch Services. Your use of the Twitch Services is at your own risk.

Once again, the document declines to define obscenity and pornography. It also couches such content within the notion of “risk,” both legal and emotional, suggesting that sexual content represents a potential danger to users and the platform alike.

The subject of nudity also appears frequently in these documents. Nudity itself is far from a self-evident category. In recent years, there have been many debates across social media platforms about what constitutes appropriate or inappropriate attire—and which parts of a person’s body can or cannot be shown before an image is flagged and deleted by moderators. Overwhelmingly, these regulations and their uneven enforcement fall along gender lines. Infamously, as part of its own ban of sexual content in 2018, Tumblr released policies that prohibit “female-presenting nipples” (Walker, 2018), a turn of phrase that is at once absurd and reflective of a gendered double standard in definitions of nudity in online spaces. These broader issues form a backdrop to Twitch’s discussion of nudity in their policy documents.

Twitch’s “Nudity, Pornography, and Other Sexual Content” document itself contains two subsections that address nudity. The first, “Nudity and Attire,” focuses on the clothing and bodies of streamers. It states that “attire intended to be sexually suggestive and nudity [are] prohibited.” This subsection foregrounds the idea of “appropriate public attire,” reminding readers that streaming is a “public activity” and stating that streamers should wear “attire appropriate for public settings, such as what you would wear on a public street.” In a rare moment of specificity, this document lists individual articles of clothing and body parts that, when displayed on screen, are considered sexually suggestive, including, “undergarments, intimate apparel, or . . . on male or female genitals, buttocks, or nipples.”

Yet, an FAQ at the bottom of the document reneges on this specificity. In response to the question “Why aren’t there specific rules about what clothing is and isn’t allowed on Twitch?” (an odd question, given that there are such rules, quoted above), the document states, “As Twitch continues to expand the variety of content we feature, so comes the need for updating the range of attire that is acceptable.” In particular, this FAQ response stresses the importance of context for determining what constitutes acceptable clothing, such as whether a streamer’s attire would be appropriate for specific “real-world” activities. The document explains that Twitch’s moderators will consider “not just the attire itself, but also the contextual setting in which it is worn and the intent of the person wearing it.” This raises an important question: Who determines what a streamer’s “intentions” are in wearing a certain article of clothing? How does this open streamers up to mis-readings, biased assumptions, and potentially discrimination?

The second subsection on nudity in the “Nudity, Pornography, and Other Sexual Content” document concerns sexual content as it appears in video games themselves. It reads,

Games featuring nudity, pornography, sex, or sexual violence as a core focus or feature, and gameplay modified to feature these elements are entirely prohibited. Occurrences of in-game nudity are permitted, so long as you do not make them a primary focus of your content and only spend as much time as needed in the area to make progress. Games rated Adults Only (AO) by the ESRB are not permitted in gameplay-oriented broadcasts or complete or unedited format on Twitch.

It is worth observing that this discussion of sexual content in games differs from other comments on sexual content found elsewhere in these documents. Though Twitch’s community guidelines supposedly prohibit all sexually explicit or sexually suggestive content, this document states that “nudity, pornography, sex, [and] sexual violence” in video games are only considered unacceptable if they represent a “core focus or feature” of a game. In contrast to the subsection on streamer attire, this subsection also allows in-game nudity, provided that streamers do not direct “too much” attention to this imagery. This leaves considerable leeway for interpretation in assessing whether a game is appropriate for the platform, creating possibilities for bias in determining what constitutes a “normal” amount of sex in a video game versus excessive or otherwise remarkable (and therefore unacceptable) sexual content. Examples that illustrate how these guidelines are implemented appear in an additional document, Twitch’s “List of Prohibited Games,” which describes and enumerates specific video games that streamers may not play (Twitch, n.d. c), either because of sexual content, violent content, or both—44 titles and counting, as of this writing.

Double standards and cultural logics

Comparing these documents reveals a number of discrepancies and contradictions in how Twitch defines sexual content, which have problematic implications. One such discrepancy is between the definition of sexual content in games versus the definition of sexual content in the performances of streamers. Twitch’s community guidelines state that both sexually explicit and sexually suggestive content are prohibited. Accordingly, the “Nudity, Pornography, and Other Sexual Content” document confirms that streamers are not allowed to wear “sexually suggestive attire” on camera. Yet, the “List of Prohibited Games” states that “occurrences of in-game nudity are permitted, so long as you do not make them a primary focus of your content” (emphasis mine). This indicates a double standard in how Twitch defines sexual content—or, perhaps more accurately, in the types of sexual content it deems acceptable for viewing. Ways of presenting the body that are considered unacceptable for streamers, such as a woman streamer wearing underwear, are, at same the time, considered acceptable in games. Consider Grand Theft Auto V (Rockstar Games, 2013), the third most popular game being streamed on Twitch at the time of this writing, with 145,000 active viewers. Grand Theft Auto V is only one of innumerable examples of video games that Twitch allows players to stream in which women characters wear attire or display levels of nudity for which women streamers themselves would be banned. In one scene in Grand Theft Auto V, for example, the player-character visits a strip club and receives an up-and-personal lab dance from a performer wearing a g-string. Her breasts overflow the top of her corset; they are featured so prominently in the sequence that, at moments, they fill up the entire screen. Women streamers who dress or dance in this way would be subject to ban. Yet, when it comes to in-game content in an established, mainstream, and widely popular video games series, such content is permitted. This illustrates how Twitch’s definition of sexual content shifts depending on who creates the content in question and whether it is considered “normal” within the confines of contemporary gaming. According to Twitch’s guidelines, it is more acceptable for women to be represented in a sexual manner (in games) than for them to present themselves (as streamers) in this way.

A second contradiction that emerges across these documents is a marked contrast in the vague language they use to define sexual content and the concrete language they use to state procedures for reporting and punishing users who violate those policies. Whereas documents like the community guidelines and terms of service reference sexual content using broad terms, other documents, like “How to File a User Report” (Twitch, n.d. b) and “About Account Suspensions and Chat Bans” (Twitch, n.d. a) are clear and direct. When it comes to what counts as sexual content on Twitch, the answer is open to interpretation. However, when it comes to reporting content or what will happen to those who post it, the answer is comparatively specific. This demonstrates how the language of policy documents like Twitch’s can vary depending on the platform’s goals. Leaving the definition of sexual content vague gives Twitch considerable leeway to determine, on a case by case basis, what content (and, by extension, which content creators) they do or do not want to represent their community. Describing reporting mechanics and consequences for violating policies in concrete terms also allows Twitch to enforce its guidelines with little room for pushback from users. This serves as a reminder that, as Hunt (1993) states regarding the history of pornography, no definition of sexual content exists separately from the conditions of its regulation.

Upon close consideration, it becomes apparent that there are also notable absences in Twitch’s policy documents and their discussions of sexual content: topics that, while highly relevant to the platform and the experiences of its users, these documents decline to cover. One example can be found in the “How to File a User Report” document, which provides instructions for filing a complaint about a broadcaster, another user on the platform, or the content of a video game being streamed. To a large extent, this document is technical; it walks readers through the steps of accessing the reporting tool from different devices. Nonetheless, this document is influential in structuring the ways that users are directed to conceptualize what constitutes a violation on Twitch. Notable is the absence of references to sexual content in one particular part of the document: the instructions for how to report user-to-user harassment. The document states that reporting harassment is “most suited to when a user is [engaging with] you in a manner that breaks the Terms of Service, such as being abusive or threatening.” No reference appears here to sexual harassment, a common form of harassment experienced by women (among others) on Twitch—which, while it may not always be characterized as “abusive or threatening,” is nonetheless harmful. This document suggests that only certain types of sexual content fit within the definitions and the concerns of the platform. It also suggests that, when it comes to sexual content, Twitch is primarily focused on regulating streamers rather than viewers, even in the face of sexual harassment directed at streamers.

An analysis of these documents reveals not only how they define sexual content but also the cultural logics that shape how these definitions are articulated and justified. Often, Twitch frames these documents to emphasize community, safety, and positivity. The community guidelines document, for example, begins with a statement that the guidelines have been designed to “protect the integrity of our community” against “inappropriate or harmful” conduct or content. The document continues,

At Twitch, our mission is to provide the best shared social video experience . . . where creators and communities can interact safely. To achieve this goal, we ask that all users participate in such a way that promotes a friendly, positive experience for our global community.

Through statements such as this, these documents present sexual content as a matter of danger or even violence, and position sexual content in opposition to values like “community” and “positivity.” Echoes of this ethos appear throughout these texts, such as when the “Nudity, Pornography, and Other Sexual Content” document groups sexual content with sexual violence and the exploitation of children or when the terms of service present “obscenity and pornography” as types of content that may cause users “harm.” By contrast, the absence of sexual content is presented as a characteristic of a safe and positive community. This framing suggests a limited and biased view of who counts as a legitimate member of the Twitch “community.” Streamers who broadcast sexual content (or, as explained earlier, whose content is perceived as sexual) are implicitly positioned in contrast to “safety” and “positivity.” Such language goes beyond stating the platform’s policies for the moderation of content. It ties the issue of inappropriate content to morals and values, suggesting that those whose content gets caught in the net of vague and biased policies (often women) represent threats to the unity of the (male-dominated) Twitch community.

In addition to framing Twitch’s regulations around notions of community and positivity, these documents also present the task of defining sexual content as a matter of common sense. The community guidelines document, for example, states matter-of-factly: “These guidelines fall under a common sense philosophy.” By referencing common sense, Twitch implies that it does not need to provide more concrete definitions of sexual content because its readers share a reasonable, universal, stable understanding of what constitutes such content. However, it is important to ask: Whose common sense is this? As these documents demonstrate, it is the hegemonic, heteronormative common sense of the population that makes up Twitch’s core demographic: white, straight, cisgender men. Twitch is not alone in appealing to common sense logics to regulate acceptable forms of sexual expressions on digital platforms; notions of common sense are often part of technological moral panics, especially those related to women’s use of digital media to engage in sexual expression (Hasinoff, 2015). Twitch’s appeal to common sense also enlists readers in the work of defining sexual content. After all, community plays an active role in identifying and regulating sexual content on Twitch, given that Twitch’s moderators review content based on user reports. As described in these documents, to be a community member on Twitch is also to participate in the work of community policing and peer-to-peer surveillance. Indeed, it takes the labor of an entire community to define sexual content on Twitch and to put this definition into action. Far from being obvious and self-evident, though it is often presented as such, the definition of sexual content must be constantly reestablished and renegotiated, requiring considerable, ongoing work to uphold a subjective and often sexist status quo that passes as common sense.

Conclusion: biases in definitions of sexual content

Ultimately, what emerges across these documents is a vision of Twitch’s definition of sexual content as vague, subjective, shifting, scattered, and often contradictory. One of the most striking observations that can be made from this body of material is that, though Twitch addresses sexual content and related issues in six different public-facing policy documents, no clear, concrete definition of that sexual content can be found across this network of texts. Practically speaking, from the perspective of a digital platform like Twitch, this vagueness may be considered helpful for giving moderators leeway to determine which pieces of content (and, by extension, which users) to remove from the platform. However, for precisely these reasons, such blurry definitions are problematic because they leave considerable room for bias and discrimination–for example, in how Twitch’s guidelines are enforced. Without a clear definition of sexual content, the matter of what counts as “sexual” is in many respects left open to interpretation. As the work of interpretation plays out across user reports and platform moderators, the perceptions of straight male viewers become the presumed default, leaving marginalized streamers open to disproportionate regulation and subjective perspectives, shaped by explicit or implicit bias.

Though these guidelines affect marginalized streamers of many identities and backgrounds, the case of women streamers is particularly helpful for illustrating how these documents place responsibility on streamers’ shoulders, while also ignoring these streamers’ agency to determine the nature of their content and their own self-presentations. Women streamers still regularly experience harassment on Twitch, including challenges to their legitimacy as gamers and sexually charged comments about their bodies. In addition, women streamers’ self-presentation is more likely to be met with negative responses and/or deemed sexual by straight men viewers (Ruberg et al., 2019) or the platform’s own moderators (D’Anastasio, 2019). Depending on whose perspective Twitch adopts, different forms of content may be considered sexual and therefore prohibited from the platform. To the extent that the vagueness of these policies provides a kind of freedom, it is the freedom of men to regulate the bodies of women on their own shifting terms.

In addition to the ways that Twitch’s policy documents expose women streamers (as well as others) to potential discrimination, these documents are themselves constructed around gendered cultural biases and anxieties. This can be seen, for example, in how these documents address the topic of sex work–the professional exchange of sex or related body work for compensation. At times, sex work is only referenced in these documents euphemistically, such as in the “Nudity, Pornography, and Other Sexual Content” document, which lists representations of “sexual services” as one type of content that Twitch considers sexually explicit. At other times, sex work is referenced explicitly and incorrectly grouped with violent or exploitative activities. Most egregiously, the subsection on “Sexual Violence and Exploitation” in the “Nudity, Pornography, and Other Sexual Content” places “sexual services, including prostitution, escort services, [and] sexual massages” in the same list as “sexual assault and molestation” and child pornography. These passages demonstrate how Twitch’s policies reflect and replicate larger cultural biases, like those against sex workers and especially women sex workers (Grant, 2014). In such moments, these documents reveal their sexist, ill-informed nature—equating sex labor done by women with violence and the abuse of children.

Sex work is a particularly charged topic for a video game live streaming platform like Twitch. The issue of defining sexual content in live streaming inevitably also raises the question of the similarities between streamers and webcam models: online sex workers who perform through webcam. As Senft (2008) describes it, the work of “camming” in many ways echoes that of live streamers today. Camming itself has changed and evolved in the decade since Senft’s study; webcam models and streamers now use many of the same technologies and techniques to attract and retain viewers and to earn money—for example, tipping systems built into their respective streaming platforms. They also face many of the same challenges around emotional labor (Lord, 2018) and performances of authenticity (Jones, 2016). Indeed, a number of professional streamers are also webcam models. Much of Twitch’s attitude toward sexual content on their site can be understood as an anxiety about this blurry divide. Implicit in Twitch’s policy documents is the need to distance live streaming from camming. This also helps explain why Twitch focuses their discussion of sexual content on the performances of streamers but remains comparatively lax about in-game content and almost entirely silent on viewers’ sexual harassing comments. Even when these documents present themselves as addressing a range of issues, one of their main purposes is to regulate the ways that streamers perform (with) their bodies on Twitch, lest these performances look too much like camming. This too reflects a sexist attitude that rejects—while also leaving unspeakable—the suggestion that the hot new phenomenon in the supposedly straight, male-dominated realm of “real” gaming (Consalvo and Paul, 2019) could in fact have much in common with women’s online sex work.

Clearly, more concrete and thoughtfully constructed community guidelines are needed for a platform like Twitch to become truly inclusive or “diverse.” As they currently stand, these guidelines exemplify how discrimination can operate on an online platforms at a systemic level. Documents like community guidelines and terms of service may seem tucked away in the background of digital platforms, but these documents in fact have an active hand in shaping the content and cultures found on these platforms. The case of video game live streaming offers a clear example of how regulations, and especially those related to sexuality and gender, have the power to enact platform politics. On Twitch, these regulatory documents set the terms for which streamers and which content will be seen as acceptable, valuable, and (in the parlance of these documents themselves) “positive.” My purpose here has not been to argue that any particular piece of content, whether sexual or otherwise, should or should not be allowed on Twitch—though I do believe that these regulations should be altered and made more specific, ideally in collaboration with those most affected by these regulations: marginalized streamers. I also recognize that it is immensely challenging to establish one concrete, comprehensive, unchanging, and socially just definition of sexual content, whether on Twitch or any media platform. Nonetheless, it is crucial to address the ways that sexual content is defined in these policy documents and others like them so that we can move away from the concept of “common sense.” Doing this allows us to reveal how definitions of sexual content, whether on Twitch or other digital platforms, are fundamentally constructed, subjective, and reflective of broader cultural ideologies, values, and biases. Promoting diversity, inclusion, and social justice in video game live streaming, as well as on contemporary digital platforms across the Internet, requires confronting the biases that underlie their regulatory policies, which often have real and immediate effects on those who participate on these platforms.