Community-based policing and emerging informal systems of justice: Battling against the dark side of Twitch

Introduction ‘We are reviewing each case that has come to light as quickly as possible, while ensuring appropriate due diligence as we assess these serious allegations. We’ve prioritised the most severe cases and will begin issuing permanent suspensions in line with our findings immediately. In many of the cases, the alleged incident took place off Twitch, and we need more information to make a determination’ (Twitch statement, 2020) The above statement was made by the streaming platform Twitch in June 2020, after a significant number of streamers, moderators, users, and employees took to social media (primarily Twitter and Tweets linking to either Twitlonger or Medium) to share stories of abuse, harassment, bullying, and, in some extreme instances, sexual assault and rape. This study set out to explore these allegations in detail, ultimately seeking to understand how individual motivations and concepts of justice may have aligned with utilisation of informal, community-based systems of justice rather than traditional, formal, offline systems. This study uses concepts of community-based policing (also referred to as peer-to-peer surveillance, calling out, and the related concept of cancel culture whereby content creators are ‘cancelled’ or forced to leave the industry based on public opinion), the role of the gaming industry in challenging harmful stereotypical norms, and the exposition of power dynamics between streamers, moderators, viewers, and employees in these new ecosystems to frame our discussion. These social media power dynamics combined with an unstable regulatory environment can both entrench and challenge patriarchal heteronormativity - entrenching through user-based regulation but challenging through movable power dynamics. It was therefore unsurprising when, in June 2020, this flood of allegations of a wide variety of abuse linked to the streaming site Twitch began to emerge on Twitter. While difficult to pinpoint an exact start, media outlets reported that it began with a Tweet on June 19, 2020, calling out an unnamed Destiny streamer as being a ‘scum lord’ (Lorenz, 2020). It started a tidal wave, culminating in the #TwitchBlackout hashtag which saw users boycotting Twitch for 24 hours (Bullard, 2020; Hall, 2020; Hernandez 2020). Individual allegations of this nature are not unusual, but it was the tidal effect of so many stories being shared at the same time that makes this time period ripe for study. The media attention was enough that, six months later, Twitch updated their policies in response (Perez, 2020; Rayome, 2020). These allegations manifested the extent of toxic behaviours in the Twitch world where power, popularity, and money are openly up for grabs, the lines between online and offline relationships are blurred, and the current approach of community-based policing seemingly fails to properly regulate the environment (Chess et al., 2021; Chess & Shaw, 2015; Gillespie, 2010; Marinett, 2021; Massanari, 2017; Powell et al., 2018; Ruberg, 2021; Salter, 2018; Shaw, 2015; Thomas et al., 2021; Zolides, 2021). While Twitch can bring huge benefits, including a whole new labour market, communities of support, diverse representation, and, of course, fun, there is clearly a ‘dark side’ which is yet to be addressed. We use these June 2020 Twitter allegations to explore if and how the nature of the allegations relates to individual notions of justice, attempting to understand why informal, social media based, loose systems of justice were used rather than traditional, structured, formal systems (See, e.g. Crawley & Simic, 2019; Powell, 2015; Powell et al., 2015). We explored 150 allegations, first with descriptive statistics and then with inductive qualitative coding. We found that, perhaps contrary to initial assumptions, social media, community-based informal systems of justice were quite similar to traditional formal systems of justice. Individuals sought similar outcomes centred around consequentialism, deterrence, and protection of society, while organic elements of due process emerged. Literature and theoretical framing We present three different scholarly areas of study to help situate and theorise our study. First, the reliance on community-based policing to regulate acceptable and unacceptable behaviours for users, both on and offline. Our study explores how this regulatory environment may affect justice-seeking behaviours. Second, the movement to challenge stereotypes in the industry, as our study also set out to explore how the allegations may or may not have continued the work to challenge gender stereotypessitting alongside other studies in this area. Finally, we look at the nature of relationships and power dynamics in the streaming industry to situate the allegations within wider applicable theories and concepts relating to relationships. Community-based policing Twitch, like most major social media platforms, privately governs online conduct: it sets community guidelines, terms of service, and harassment policies that define both acceptable and unacceptable behaviours. Yet in practice, enforcement of these rules is outsourced to users themselves. Twitch makes the rules, then distances themselves, leaving the rule implementation and enforcement to the community (Zolides, 2021, p. 3010). This delegation positions community-based policing as a natural response, posing the conundrum of whether such is a supplement to platform moderation, or whether it represents a failure of Twitch to adequately enforce its own standards of safety and accountability. Following Ruberg’s (2021, p. 1694) description of ‘peer-to-peer surveillance,’ community-based policing on Twitch often takes the form of chat moderators—volunteers who ‘work’ for streamers, typically without pay, to enforce channel rules and cultivate a welcoming environment while retaining viewers (Wolff & Shen, 2022). Moderators draw on Twitch’s official guidelines and reporting systems, but creators and their communities bear the burden of implementation (Obreja, 2021). As Zolides (2021, p. 3010) argues, Twitch maintains distance by making the rules while shifting enforcement responsibilities downward. This governance structure embeds asymmetrical power relations: moderators and creators do the labour, while Twitch retains control over platform-wide rules and monetisation. However, restricting community policing to moderators alone understates the breadth of practices through which users govern each other. Viewers report misconduct, communities ‘call out’ harmful behaviour, and creators sometimes organise coordinated responses to harassment, such as the collective campaigns against ‘hate raids’ documented by Meisner (2023). Scholars have shown how callouts and accusations can function as a form of creator-driven governance, where communities seek to expose, discipline, or delegitimise harmful actors (Hallinan, Reynolds, & Rothenstein, 2024; Hallinan et al, 2025). These practices link Twitch to broader online cultures of accountability, including ‘cancel culture’ and activist movements such as #MeToo, which rely on networked visibility and solidarity to confront abuse. The terminology used to describe these practices requires careful reflection. The phrase ‘community policing’ is borrowed from physical-world law enforcement and risks importing its punitive and carceral associations. Alternative framings—such as a norms-based regulation (Radu et al, 2021), ‘peer-to-peer surveillance’ (Ruberg, 2021), or ‘calling out’ (Meisner, 2023) may capture the dynamics of Twitch more precisely, where governance often revolves around maintaining reputations, enforcing community norms, and organising collective responses rather than formal legal authority. Our approach uses ‘community-based policing’ as a provisional term, while emphasising its overlaps with online activism, callout practices, and participatory forms of governance. We also feel this term more accurately reflects both the positive and negative aspects of the system. On the one hand, recognition of the role of the community in regulation can be positive; those at the frontline are more likely to understand context and, arguably, care about outcomes. On the other hand, the use of the term ‘policing’ is intended to highlight the negative aspects, whereby anyone can abuse the privilege of regulating from within. As an example of this negative aspect, while women gamers once celebrated the introduction of harassment policies as a step forward from the platform’s laissez-faire early years (Cote, 2017), subsequent scholarship shows that these rules are themselves gendered and exclusionary. Guidelines and reporting mechanisms privilege some forms of action while marginalising others (McLean & Griffiths, 2019), embedding discriminatory assumptions in ways that are often invisible (Zolides, 2021). As Ruberg (2021, p. 1682) argues, such policies replicate broader biases against women, LGBTQ people, and people of colour, often granting male creators’ greater freedom and less scrutiny (Zolides, 2021, p. 3002). These exclusions are not only cultural but economic: when marginalised creators are constrained in their self-expression, they are effectively denied equal access to the monetised opportunities of the platform (Zolides, 2021, p. 313). Our research contributes to this body of work by exploring how community-based policing is entangled with informal systems of justice (Crawley and Simic, 2019). By this, we mean practices of accountability that operate outside of formal legal or platform-sanctioned frameworks—ranging from moderation and reporting to public callouts, coordinated campaigns, and collective refusals to support abusive creators (Hallinan, Reynolds, & Rothenstein, 2024; Hallinan et al, 2025). Such practices are often symbolic or reputational rather than judicial, yet they function as ways in which communities seek redress, protection, or fairness in the absence of effective platform enforcement. In the context of gender-based violence, global occurrences of ‘hashtag activism’ serve the purpose of survivors feeling heard and seen, constructing narratives of strength which occur outside of formal justice systems (Powell, 2015). In situating Twitch within broader debates on cancel culture, platform governance, and networked activism, we highlight both the possibilities and the limits of informal justice as a response to abuse in digital spaces. Challenging norms Gaming spaces were historically a contentious space for anyone deviating from the white heterosexual male status (Chess et al., 2021; Chess & Shaw, 2015; Cross et al., 2022; Gray et al., 2017, 2018a, 2018b; Salter, 2018; Shaw, 2012, 2014, 2015; Taylor, 2018a, 2018b). This space has been hostile and toxic, and, coupled with the anonymity of social media, incidents of harassment and abuse remain rife (Fox & Tang, 2014; Paaßen et al., 2017). Ruberg notes that the harassment ‘manifests and functions in ways that are specific to live streaming platforms and the player cultures that surround them’ (Ruberg et al., 2019, p. 467). According to Yoganatham, ‘though some aspects of individual identity are shielded by the anonymity afforded by the virtual world, sexism and racism are a manifest, and ultimately normalized, feature of power and dominance in online gaming marketplaces’ (Yoganathan et al., 2021, p. 1033). Yet recent research is starting to show that sexism in the industry is not always a given, and (while they should not have to) those other than white heterosexual males are successfully negotiating, co-opting, and subverting these spaces, confronting patriarchal heteronormativity through empowerment and resilience (Cote, 2017; Gray et al., 2018a; McCullough et al., 2020; McLean and Griffiths, 2019; Richard and Gray, 2018; Vilasís-Pamos and Pires, 2021). For example, a study conducted by Ferguson and Glasgow explored the demographic profile of participants in #Gamergate, aiming to empirically understand if the movement really was driven by alt-right, white, heterosexual men (the generally accepted popular opinion). Their findings very much challenged this stereotype, where white males did not make up the majority of GamerGate supporters (Ferguson and Glasgow, 2021: 246; See also Salter, 2018). A systematic review undertaken by Thomas et al. continued to challenge these stereotypical identities and assumptions. They found that men experience a significant amount of online abuse as well, including physical threats and being called offensive names. According to the authors ‘these results highlight that it is critical to avoid potential stereotypes of who faces harassment online, and that experiences differ across genders’ (Thomas et al., 2021: 255). Our study aimed to explore these stereotypes specifically in the streaming industry, potentially challenging the mainstream media reports which continued to perpetrate a harmful narrative of woman/victim, man/perpetrator (Lorenz, 2020). Power dynamics in on- and off-line streaming relationships The monetisation of sites such as Twitch has created a hotbed of moveable power (Ask et al., 2019). Live streaming is unstable but highly profitable work, where the stakes are high and hierarchies of power are tall (Cote, 2020). It is not a traditional one-way system of media distribution and income generation, rather there are unique mechanisms for audience participation and ‘influence’ over the channel (Jackson, 2021; Obreja, 2021; Sjöblom et al., 2017; Wolff & Shen, 2022; Zhao et al., 2018). Viewers who pay more can exert more influence over the content creation and the channel itself. Donations are a typical example of this influence, as high valued donations present the expectation that the streamer should celebrate in a particularly spectacular fashion. Failure to celebrate adequately presents the risk of looking ungrateful, which creates an environment of constant behaviour modulation in response to their audience’s demands (Partin, 2019). Other methods of monetisation include gifting Twitch subscriptions, where viewers can pay money to gift one of the various ‘Sub Tiers’ to individual viewers or several all at once. These subscriptions become visible in the livestream’s chat log, and tend to be displayed in the livestream itself. Particularly unique creators such as CodeMiko (Twitch, n.d. -b), integrate various animations in response to these gifts, including animations that manipulate her virtual avatar. Bits are another method of supporting creators, where viewers can pay for the Bits virtual currency in exchange for ‘cheering’ which is a method of engagement through the likes of voting, sending Cheermotes and celebrating. Content creators receive revenue every time a Bit is used by a viewer to ‘Cheer’ on their channel. Creators may also opt to integrate Bits extensions onto their channel, which provide additional means for viewers to engage with them. Content creators can also opt to run advertisements on their channel, in addition to providing affiliate links for Amazon products (Twitch, n.d. -a). Revenue generating, or professional streamers on the Twitch platform, are also bound to the Twitch Monetized Streamer Agreement, which outlines the contractual terms and conditions that must be adhered toin order to remain applicable for monetisation. This agreement has been scrutinised as a power imbalance between the platform and the content creator (Aade, 2025), as it grants Twitch the authority, at its own discretion, to suspend created accounts and remove uploaded content for any reason. This level of discretionary control over content creators gives Twitch the flexibility to enforce such rules, regardless of whether there’s sufficient justification for that enforcement. A notable example of Twitch exercising this power was with the banning of the streamer Dr Disrespect in June 2020, which was done so without explicit reason at the time, the abruptness of which created shock amongst fans and other content creators. In broadcast categories such as “Pools, Hot Tubs, and Beaches” popular streamers have been banned from Twitch, with Twitch itself being reported as being unresponsive to the harassment and abuse viewers inflict onto streamers (Diaz, 2021). Attempts to measure viewer-lead toxicity have been attempted for other platforms, with the results indicating implications surrounding monetisation despite the increased engagement rate toxicity potentially brings (Bertaglia et al., 2024). Content removal has also been observed as being disproportionately high for marginalised users (Thach et al., 2024), and this is especially so when there is a lack of transparency surrounding content moderation (Haimson et al., 2021). Consequences of this form of content moderation were investigated further through interviews with banned-from-Twitch streamers, who all expressed the anger, stress and sadness felt by account suspension. It was highlighted in those interviews that being banned, particularly when the streamer relied on the platform for their full-time profession, was psychologically harmful, as streamers had to deal with both the suspension and the judgement of others in relation to the allegations made. Twitch structures also blur the lines between public and private. Streamers open the doors to their private lives in the most public way, and private identities become public property (Woodcock & Johnson, 2019). Often, streamers broadcast from very personal spaces, with women more often broadcasting from their bedrooms (Ruberg & Lark, 2021, pp. 680, 692). This constructed sense of intimacy tends to then translate into increased social participation in the channel, which then eventually also increases revenue. In inviting viewers into their bedroom (literally and figuratively), the relationship between creator and audience is already obfuscated – while the streamer is ostensibly in charge, there is also a dependency on the actions of the viewers (Dargonaki, 2018; Zolides, 2021, p. 3002). The platform thus conflates economics and personal relationships, with tensions ensuing (Bingham, 2020). Equally, there is a false sense of emotional connection between viewers and streamers built into the functionality of the platform itself (Wolff & Shen, 2022; Woodcock & Johnson, 2019). This can then become immensely complex when viewers and streamers engage in an offline, in-person, possibly romantic relationship (Trepte et al., 2012; Wulf et al., 2020; Yoganathan et al., 2021, p. 1007). Our findings very much align with this literature exploring the multiple layers of destructive power dynamics in the streaming industry. Study design In the immediate aftermath of the allegations in June 2020, gamer Jessy Quil began cataloguing the allegations in a publicly available Google spreadsheet titled ‘Survivor Stories of Harassment/Abuse/Assault within the gaming live-streaming industry’. She wrote in the document: People came forward with their stories. I am not casting judgement, nor condone a witch hunt of those accused of alleged abusive behavior. This is to give survivors a voice so they don't feel alone or gaslit based on their experiences in this industry. Responses to allegations have been added to give impartiality to these events (Quil, 2020). While we note that this was just one collection of allegations, it was comprehensive, contemporary, and both sensitive and respectful to those involved. In our initial scoping of this study, her spreadsheet appeared to be the most referenced and shared. Her catalogue, as compared to two others we found, was the most neutral, including the most number of stories, and was the only one to include responses. However, there are limitations to using a catalogue compiled by someone external to the study, in that there may have been elements of her data collection that introduced potential bias outside of our control. By the time Quil was done cataloguing the allegations, there were ~420. According to her catalogue, most allegations were made on Twitter, many with links to Twitlonger, Medium, or YouTube, though this was not exclusively the case. We limited our initial scope to the first 48 hours, where the bulk of allegations (250) were made. The allegations then tailed off significantly and we were interested in the collective aspect of those that all shared together en masse to capitalise on the calling out and group mentality aspect of the allegations. This also helped to create a workable sample size. With a 95% confidence interval and a 5% margin of error of the 250 allegations in the first 48 hours, we used a z-score for descriptive statistics which indicated an ideal sample size of 150 (Chaudhuri, 2022). This was also feasible given the size and scope of the project. 150 allegations were purposively selected from Quil’s catalogue by the principal investigator (PI) based on Creswell’s approaches to ensure representative coverage, reviewing all 250 from the timeline of interest and selecting those that included a wide range of types of allegations, individuals involved (aiming to include a range of gender identities, ages, and roles in the streaming industry), depth of information available (some of the stories shared on Quil’s spreadsheet did not contain any usable information), and location (so as not to be solely a US study) (Creswell, 2024). The PI then completed the full enquiries, using Quil’s archive/catalogue to find the initial allegation and any possible responses. It is hoped that by using a manual, purposive selection process, bias was minimised as the goal was to ensure a representative and wide ranging dataset. Descriptive statistics involved deductive coding of data, determining who was making the allegations, who the allegations were made against, the content of the allegations, and the legal status of the allegations. The PI also located demographic information publicly available on Twitter, information publicly available on Twitch, and in-depth internet and media searches to collect as much additional evidence for each story as possible. Once descriptive statistics were collected, the PI then accessed all of the original allegations from the 150 selected stories. This study was initially undertaken purely as a descriptive statistical study, but was redesigned later on to more fully engage with the rich qualitative data, which also involved a further ethics discussion. This unfortunately meant that the full breadth of qualitative stories were not all still publicly available. Additionally, some of the stories were in video format only or did not include enough text for qualitative analysis. Therefore of the initial 150, 123 were qualitatively analysed. Qualitative analysis took an inductive approach, coding the data iteratively with thematic analysis in Atlas.ti (Braun & Clarke, 2006). This analysis was conducted to learn about why individuals choose to engage with informal systems of justice, what their notion of justice was (if indicated), and how an informal system might align more closely with their conceptualisation of justice than a formal system. This was informed by our theoretical work on community-based policing and the power structures built into platform architectures (specifically Twitch), addressing some of the critiques of thematic analysis (Bazeley & Bazeley, 2020). Inter-rater reliability was used for consistency and validity (Gwet, 2012; O’Connor & Joffe, 2020). The PI engaged in initial free coding to develop a coding frame which specified the high-order categories, definitions, component sub-themes, and examples from the data. The coding frame was first tested by the PI and then a selection of data (20%) was coded by two independent researchers to determine IRR using pre-defined quotations (O’Connor & Joffe, 2020). This was done to ensure reflexivity and dialogue more than as a rigid measure of objectivity, using a simple percentage agreement. Given the sensitive nature of the subject matter and the use of survivor stories (though publicly available), the study was given ethical approval by the University of Suffolk Research Ethics Committee in November 2020. To protect the individuals involved, the Ethics Committee in further discussions suggested that we do not use Twitter handles when using quotes in this paper. Quotes are directly from the publicly shared stories linked in Quil’s catalogue and are indicated with indented italics. Also note that original text has not been altered; there may be spelling and grammatical errors but these are preserved to reflect the user voice. Findings Descriptive statistics The 150 allegations studied were made by 136 unique individuals, against 83 alleged perpetrators. This means that 35 individuals (42%) had more than one allegation made against them, with three individuals having more than five allegations against them. On two occasions, alleged victims were also named as alleged perpetrators in separate allegations. Twelve specific characterisations of abuse were found, as seen in Table 1 . Table 1: Twelve characterisations of abuse Number Percent Toxic relationship 37 25% Bullying and harassment (non-sexual) 21 14% Inappropriate advances 20 13% Sent or requested lewd texts or photos 17 11% Inappropriate touching and/or kissing 14 9% Rape 12 8% Felt pressured to engage in sexual activity 11 7% Sexual harassment or assault (other than rape) 10 7% Blacklisted from industry 3 2% Poor/unethical business behaviour 2 1% Inappropriate comments 2 1% All of the stories took place in the UK (England) or the US. As one of the investigators is a US-trained lawyer teaching law in the UK, they were able to determine based on knowledge of criminal law and necessary legal requirements and elements to determine that approximately 15% of the allegations rose to the legal level of a potentially prosecutable crime under sexual harassment, sexual assault, or rape. These were coded based on jurisdiction (England or US) and, to the extent possible, used the facts given to map against the relevant criminal codes. Bullying and harassment allegations amounted to 15%, corroborating the employment lawsuits (described more fully later in this paper) and could, perhaps, have legal implications. This means that a total of 30% of the allegations may have violated either criminal or employment law, with the potential for more under coercive control doctrines. Five allegations studied involved the police with one perpetrator eventually found guilty of a sexual crime. This leaves 70% of allegations without a clear formal justice route, relying only on community-based policing and naming and shaming. One of the aims of this research was to, as much as possible, explore the outcomes of the allegations for those involved. In the short term, 33 accused individuals publicly responded to allegations, again most using Twitter or links to blogs or videos. Of those who responded, 12 (36%) apologised, 10 (30%) refuted or denied the allegations, seven (21%) countered the allegations with evidence, and four (12%) responded but in another way (apologised but did not take responsibility for serious accusations, angry rant at cancel culture, etc.). Note that the use of the word ‘accuse’ to describe the individuals involved reflects the original language in Quil’s spreadsheet. In the long term (two years), we used Twitch statistics and industry prominence to approximate whether individuals involved in accusations remained in the industry, and, if so, were in a comparably stronger position within two years of the accusation (i.e. more subs on Twitch or YouTube, higher viewership on any streaming platform, partner status with Twitch, or more senior employment position in industry). 1 Table 2: Long term impact Accuser less active Accuser more active Accused less active 11% 33% Accused more active 30% 26% These will be explored further in the discussion section, but it is important to note here that there were a wide range of outcomes. Thematic analysis Almost all the stories shared were framed by some sort of explanation as to why the individual was choosing to use social media, and many went so far as to clearly indicate what outcome they desired. Most stories had multiple and overlapping stated motivations, though not all. Some gave no reason whatsoever and just wrote out what happened, while others shared very lengthy accusations with up to 11 different explicitly stated motivations. Most ranged between three and six, with an average of three and a half. In total, there were 434 stated motivations across all 123 stories analysed. Table 3: Stated motivations for sharing on social media Motivation % (N = 123) Join 54% Silent 37% Warn 30% Support 28% Heal 27% Afraid 21% Believe 21% Apology 20% Industry 17% Confront 15% Risk 15% Deplatform 13% Report 12% Define 11% No Negative Outcomes 11% Get Better 9% Improve 7% None 3% False Claims 2% Community-based policing, calling out, and cancel culture Over half of all stories (54%) specifically stated that part of their motivation was to join the others – either in general or noting that they know there are others who have experienced similar forms of abuse or mistreatment from the same perpetrator. This reflects the community aspect of social media, both in the positive aspect of shared identity and shared experience, and in the aspect of turning to community-based policing and calling out. This was the most common motivation evidenced in our data set. First of I want to with adding my experiences with some of the people that have already been spoken about. While some of my experiences are not a awful as other people’s, I still think it would be worth sharing. Reading your accounts has been devastating, but at the same time, it has made me feel so much less alone. I’m more grateful than words can ever say for the bravery of the women who came out first. Because of you, my chains are about to break. 37% noted the importance of speaking out or being heard. There was something inherently important about telling the story out loud for others to hear (read). It was about being heard and making people aware that this happens I couldn't take it any longer - I had to speak up, for them & for myself. Speaking out is the best way to take the power away 30% said that they wanted to warn others about their specific alleged perpetrator. This is a clear example of using community-based policing or calling out as a deterrent or to attempt to curb the abuse. I am writing this because women, especially vulnerable women, deserve the right to avoid dating and associating with someone who has a very recent past of being abusive. If he does decide to come back at all to the gaming space, women need to know everything before jumping into anything with him- so here's my story. I'm not going fully into it but if you're talking to them, be warned. Closely related, 28% offered support to others while 27% said that they needed to share their story in order to heal, find closure, or stop blaming themselves. The notion of catharsis is very prevalent in the qualitative findings, with explicit references to healing (27%), supporting others (28%), and apologising (20%). Challenging norms 113 of those making the allegations publicly identified as female (83%), 16 male (12%), and seven did not provide enough public information to determine their gender identity. 72 alleged perpetrators were male (87%). Six of those accused of abusive behaviour publicly identified as female. The assumption that heterosexual men are always the perpetrators and women are victims was challenged in our study, reflecting emerging scholarship (Ferguson & Glasgow, 2021; Thomas et al., 2021). Mainstream media and earlier scholarship tends to frame these issues as stereotypically misogynist and heteronormative - while there are certainly elements of misogyny and patriarchal heteronormativity in power dynamics, limiting the lens to this dichotomy risks silencing minority stories. As seen in the quote below, intersectionality of gender, race, and sexual identities complexify the previously dualistic lens while also underscoring the blurring of public and private, and offline and online in the streaming industry: A reminder: assaulters are women too. At [an] after-party last year a woman forcefully shoved her tongue down my throat in front of coworkers and people I was trying to network with. It was utterly humiliating and embarrassing. I know this pales in comparison to many other stories, and as a white man saying this to a black woman I can't help but feel like absolute garbage for saying all this. I openly and honestly answer that I was a bi man…and I was happily married… She pushed me against the wall, attempted to kiss me, I pulled away. At that moment I was trapped. I felt truly helpless… I remember franticly looking past her hoping that I would see someone coming or that someone would see me. Power dynamics The other prevalent concepts tend to underscore the problems in the power dynamics in the industry as outlined in our literature review –afraid (21%), believe (21%), risk (15%). The level of harassment with this person was so intense I often felt she wanted to do everything in her power to destroy my career. The fact that you find up-and-comers who are new to the industry and leverage your power over them, knowing that they’ll all ever be too afraid to say anything, needs to STOP. It started out with a bit of flirting but later I felt very pressured in to sexting and exchanging photos and looking back there was a clear exploitation of the offset viewer — streamer power dynamic, which is unacceptable We also looked at code co-occurrence at narrative level; in other words, are there patterns in the stated motivations as each story often had more than one? Wanting to join others in sharing and a need to break silence were often seen together, emphasising again the ‘social’ aspect of social media (26% of all accounts). This was the most common co-occurrence. 11% of all stories had both silent and heal as a motivation – this links the idea of telling the story to being able to heal and move on, the catharsis effect. Perhaps unsurprisingly, 59% of those who discussed the risk they were taking also expressed fear, and most of these were from females. Half of those who wanted to improve the industry also encouraged the community to believe others telling their stories. We then explored any trends in the relationship between outcomes and motivations. Although there were not many notable trends, one particularly intriguing relationship showed that of those who expressed a desire for the alleged perpetrator to get better or improve themselves, 91% remained in the industry in the long term. We also saw that those who used social media to confront the accused also tended to stay in the industry, perhaps reflecting the monetised power dynamics explored in our literature review. In this way, those who used social media to have some sort of direct communication with the accused were more likely to stay in the industry long term. Finally, we put the pieces together to see if there was a relationship between the nature of the allegation and the motivation for publicly sharing. Those who alleged sexual assault or rape tended to express a desire to break the silence and be heard and to want to actively support others. Those whose allegations went as far as rape also tended to want to heal and felt that publicly sharing their story would help in that healing process. Individuals who had experienced inappropriate advances notably felt it important that those reading believed them and others. This may be because inappropriate advances may be subjective or less ‘visible’ than something going as far as sexual assault or harassment, but it is still problematic. This may explain why these individuals needed to ask readers in the community to believe the allegations. Perhaps unsurprisingly, those whose stories were about feeling pressured into engaging in sexual activity also tended to express fear, again underscoring what we have already seen in the literature and earlier findings about problematic power dynamics in the industry. Those who had experienced bullying and harassment that was not of a sexual nature seemed to want to warn others of the specific behaviours of the accused. However, overall, the data indicates that there were no strong associations between the nature of the allegation and the motivation for sharing or making an allegation on social media. Discussion Community-based policing If individuals are entrenched in an ecosystem that relies on community members for front-line policing and the interpretation and implementation of policies and regulations, then will they invariably turn to community-based, informal systems of justice? Rather than engaging with criminal law or employment law to right the wrongs of abuse and harassment, individuals in our study instead used the ‘Court of Twitter’ and public opinion to police, call out, or cancel. Without formal systems of justice to turn to (only 30% met possible legal thresholds for sexual assault or employment law) and with immersion in community-based policing, it is no wonder that victims turned to another online community to seek informal justice. This calls into question traditional notions of ‘justice’ which tend to focus on formal systems which may not be accessible, particularly to minority or already silenced voices. If we use individually defined notions of justice, there may be very compelling reasons for turning to accessible, informal systems. The concern is, of course, if these systems are too informal in that critical aspects of due process and balancing of rights are lost. However, perhaps the most interesting outcome of this study was the prevalence of traditional notions of justice in this community-based informal sphere. Most legal frameworks look to justice as a means to right a wrong. The concepts of compensation and incarceration are linked to not only having the wrong-doer 'pay' back society for the harm caused in a reparative (Zedner, 1994) sense, but also to work towards rehabilitation, deterrence, and the protection of the 'majority' of society (Bentham, 1780). Additionally, a cornerstone of many Western formal systems of justice is the ability to confront the wrongdoer (whether this actually occurs is not without controversy, but it remains a constitutional right in many jurisdictions). This is a useful finding, which challenges an understanding of informal systems of justice as deviating from traditional notions. Perhaps these social media, community-based informal avenues to seek justice are not viewed differently from formal legal systems. This challenges notions that public spheres fragment public discourse (Habermas, 2006), rather Salter’s (2013) sentiments surrounding ‘counter-public’ discourse are useful here. It is argued that the boundaries between the two are permeable, and often alternative ‘counter-publics’ can create genuine and meaningful discussions around sexual violence (Fraser, 1990, Powell, 2015; Powell et al., 2015; Salter, 2013). Challenging norms Our findings also align with the work of Ferguson & Glasgow and Thomas, empirically demonstrating that, while the abuse of power most definitely has gendered dimensions, the assumption that the abuse is purely misogynist and heteronormative silences many voices with multiple identities, and ultimately is not supported by our data (Ferguson & Glasgow, 2021; Thomas et al., 2021). At an even more macro-level, many media headlines used inflammatory language such as ‘abuse’ and ‘sexual’. Our findings do not support this representation. Many of the stories were about bullying, harassment, toxic relationships (romantic, economic, and employment), and inappropriate behaviours. As seen in the data, only 15% could legally be categorised as sexual abuse or rape. It is vital that the real nature of the problems are accurately labelled and understood, so as not to overlook less salacious challenges in the industry and to consider appropriate responses. Power dynamics Although perhaps a small piece of our work, the initial finding that 42% had more than one allegation made against them and that some alleged victims were also alleged perpetrators in other allegations paints a complicated picture with overlapping roles, identities, and motivations. We must question whether, for 35 individuals who had more than one allegation, they really were abusing their power, or whether the calling out culture led to mob-mentality false allegations. Quil herself used the phrase ‘witch hunt’, bringing up important connotations about this informal community based approach. To then see that two individuals who said that they were victims were then accused by others as being perpetrators presents an immensely complex hierarchy. We loosely hypothesised that many of those making accusations would leave the industry (this was indirectly based on the initial reading of many allegations which alluded to a desire to leave the industry), so the prevalence of individuals who shared stories of abuse who remained in the industry after two years was a surprise. There are a few possible explanations for this potentially surprising result. One is that there is a selection bias in those that feel willing and able to share their stories of abuse. These are individuals who have the social and emotional capital required to speak out; perhaps this also means that they have the capital and the support to thrive in the streaming industry (Gray et al., 2017). Another possible explanation is that in sharing their stories, they feel that they have achieved their desired notion of justice and that experience has allowed them to continue in the industry in a more positive light. This was underscored in the qualitative analysis where those who used social media to ‘contact’ (positively or negatively) their alleged abuser were more likely to remain in the industry, again relating to formal systems of justice and the right to confront your accused. Here it is possible that using an informal community-based system achieved a sense of justice, righting a wrong. A third and somewhat dark explanation is that they have no other options for meaningful employment or income-generation. On the opposite end of the spectrum, there was also a loose relationship between a stated desire to deplatform and the alleged perpetrator leaving the industry, where only a small number were officially deplatformed by Twitch but a larger number left the industry anyway. This raises serious concerns about due process and validity of claims (relating to cancel culture discussed earlier), where it becomes too opaque to determine if this was an achievement of justice for the victim or an injustice for a wrongly accused innocent. This captures the risk of rapid interactivity within social media, whereby prejudices remain reinforced due to the multitude of interpretations regarding the ethics and morality of an individual’s allegation (Lloyd, 1993). At platform level, Twitch did respond with changes to their policies in January 2021. According to Twitch, ‘we developed the new policy to take a clearer and more consistent stance against hate and harassment, and to give you greater insight on what is and isn’t acceptable on Twitch’ (Twitch Blog, 2020). Prior to 2021, Twitch policy stated that, in some serious cases, the platform could take action against users for conduct occurring offline. From January 2021, they partnered with an unnamed law firm (a ‘highly regarded third party investigative partner’) to investigate allegations of ‘serious offenses that pose a substantial safety risk to the Twitch community, even if these actions occur entirely off Twitch’ (Twitch Blog, 2020). Here we can say that the June 2020 allegations contributed to the change in policies. As Marinett posits, ‘policies that undertake to investigate and sanction off-platform abuse…raise unique problems in comparison to on-platform content moderation that increase the difficulty of balancing the positive aims of such policies with maintaining accountability and fairness’ (Marinett, 2021, p. 3). There is still an overreliance on community-based policing, and concerns have been raised over the lack of transparency when it comes to investigating offline abuse that affects members of the Twitch community. In this vein, the allegations did have an impact, we just do not know yet whether that impact is positive or not. Finally, there may be a loose correlation between the June 2020 allegations and the macro-level legal efforts to improve these spaces. Bringing so many of these stories to light via Twitter may have added to the overall industry-wide efforts to address discrimination, bullying, harassment, and abuse from the top-down. 2 There is also continued widespread media coverage of abuse in the industry, including live streaming and Twitch. Sustained and persistent attention to the problems may, eventually, add to solutions. Conclusion Social media, community-based informal systems of justice may in fact be closely aligned to traditional formal systems of justice. Individuals sought similar outcomes centred around consequentialism, deterrence, and protection of society, while organic elements of due process emerged. Our data also, however, indicate deeply rooted problems with community-based policing as many accused individuals lost their livelihoods and left the industry without a proper investigation or balancing of rights. Our study ultimately contributes to our understanding of the conflicting power dynamics in the streaming industry, challenging some of the gender stereotypes attached to this industry, and adding a new dimension to the critiques of the use of community-based policing and informal systems of justice, where we see that justice, in some instances, can indeed be achieved, but potentially at a cost. Further work is needed to understand whether the resulting changes to Twitch’s policies has had any effect to address the problems in these informal systems of regulation, whether allegations have continued and what they look like now, and to further probe the experiences of those involved through more in-depth qualitative work with participants. References Aade, L. (2025). To ban or not to ban?: Understanding the impact of platform discretion in content moderation on professional Twitch streamers. In The Hashtag Hustle (pp. 88–106). Edward Elgar Publishing. Ahmed, S. (1995). Deconstruction and law’s other: Towards a feminist theory of embodied legal rights. Social & Legal Studies , 4 , 55–74. Ask, K., Spilker, H. S., & Hansen, M. (2019). The politics of user-platform relationships: Co-scripting live-streaming on Twitch.tv. 14 . Bazeley, P. (2021). Qualitative data analysis . SAGE Publications Ltd. Bentham, J. (1907). An introduction to the principles of morals and legislation . Oxford: Clarenden Press. Bertaglia, T., Goanta, C., & Iamnitchi, A. (2024). The monetisation of toxicity: Analysing YouTube content creators and controversy-driven engagement (p. 9). Bingham, C. M. (2020). Talking about Twitch: Dropped Frames and a normative theory of new media production. Convergence: The International Journal of Research into New Media Technologies , 26 (2), 269–286. boyd, danah. (2008). Can social network sites enable political action? International Journal of Media and Cultural Politics , (2), 241. Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology , 3 (2), 77–101. Bullard, B. (2020, June 24). ‘Twitch blackout’ spotlights claims of harassment, discrimination among streamers . SYFY. Chaudhuri, A., & Pal, S. (2022). A comprehensive textbook on sample surveys . Springer Nature. Chess, S., Massanari, A., Kruse, H., Consalvo, M., & Boudreau, K. (2021). Pleasure + gender + play. AoIR Selected Papers of Internet Research . Chess, S., & Shaw, A. (2015). A conspiracy of fishes, or, how we learned to stop worrying about #GamerGate and embrace hegemonic masculinity. Journal of Broadcasting & Electronic Media , 59 (1), 208–220. Clarke, V., & Braun, V. (2013). Successful qualitative research: A practical guide for beginners . Cote, A. (2017). ‘I can defend myself’: Women’s strategies for coping with harassment while gaming online. Games and Culture , 12 (2), 136–155. Scopus. Crawley, K., & Simic, O. (2019). Telling stories of rape, revenge and redemption in the age of the TED talk. Crime, Media, Culture , 15 (2), 259–278. Creswell, J. (2024). Qualitative inquiry and research design . SAGE. Cross, L., Kaye, L. K., Savostijanovs, J., McLatchie, N., Johnston, M., Whiteman, L., Mooney, R., & Atherton, G. (2022). Gendered violence and sexualized representations in video games: (Lack of) effect on gender-related attitudes. New Media & Society , 14614448221075736. Dargonaki, S. (2018). Performing gender on Twitch.tv: Gendered playbour through Butlerian theory. International Journal of Media & Cultural Politics , 14 (1), 103–110. Diaz, A. (2021, June 21). Twitch bans popular streamers Amouranth and Indiefoxx after yoga ASMR streams . Polygon.Com. Ferguson, C. J., & Glasgow, B. (2021). Who are GamerGate? A descriptive study of individuals involved in the GamerGate controversy. Psychology of Popular Media , 10 (2), 243–247. Fox, J., & Tang, W. Y. (2014). Sexism in online video games: The role of conformity to masculine norms and social dominance orientation. Computers in Human Behavior , 33 , 314–320. Scopus. Fraser, N. (1989). What’s critical about critical theory?: The case of Habermas and gender. In Unruly Practices (NED-New edition, pp. 113–143). University of Minnesota Press. JSTOR. Fraser, N. (1990). Rethinking the public sphere: A contribution to the critique of actually existing democracy. Social Text , (25/26), 56–80. JSTOR. Gillespie, T. (2010). The politics of ‘platforms’. New Media & Society , 12 (3), 347–364. Gray, K. L., Buyukozturk, B., & Hill, Z. G. (2017). Blurring the boundaries: Using Gamergate to examine “real” and symbolic violence against women in contemporary gaming culture. Sociology Compass , 11 (3), e12458. Gray, K. L., Voorhees, G., & Vossen, E. (2018a). Feminism in play . Springer International Publishing AG. Gray, K. L., Voorhees, G., & Vossen, E. (2018b). Introduction: Reframing hegemonic conceptions of women and feminism in gaming culture. In K. L. Gray, G. Voorhees, & E. Vossen (Eds), Feminism in Play (pp. 1–17). Springer International Publishing. Gwet, K. (2012). Handbook of inter-rater reliability: The definitive guide to measuring the extent of agreement among raters . Habermas, J. (2006). Political communication in media dociety: Does democracy still enjoy an epistemic dimension? The impact of normative theory on empirical research. Communication Theory (1050-3293) , 16 (4), 411–426. Haimson, O. L., Delmonaco, D., Nie, P., & Wegner, A. (2021). Disproportionate removals and differing content moderation experiences for conservative, transgender, and black social media users: Marginalization and moderation gray areas. Proceedings of the ACM on Human-Computer Interaction , 5 (CSCW2), 466. Hall, C., & Polo, S. (2020, June 25). The game and comics industries are grappling with widespread allegations of harassment and abuse . Polygon. Hallinan (Blake), Reynolds (CJ), Kuperberg (Yehonatan), & Rothenstein (Omer). (2025, March 31). Aspirational platform governance: How creators legitimise content moderation through accusations of bias [Info:eu-repo/semantics/article]. Alexander von Humboldt Institute for Internet and Society gGmbH. Hallinan (Blake), Reynolds (CJ), & Rothenstein (Omer). (2024, June 26). Copyright callouts and the promise of creator-driven platform governance [Info:eu-repo/semantics/article]. Alexander von Humboldt Institute for Internet and Society gGmbH. Hernandez, P. (2020, June 25). Twitch starts banning streamers over sexual abuse allegations . Polygon. Jackson, N. (2021). Understanding memetic media and collective identity through streamer persona on Twitch.tv . Lorenz, T., & Browning, K. (2020, June 23). Dozens of women in gaming speak out about sexism and harassment. The New York Times . Marinett, M. (2021). The new frontier of platform policy. Internet Policy Review , 10 (3). Massanari, A. (2017). #Gamergate and the Fappening: How Reddit’s algorithm, governance, and culture support toxic technocultures. New Media & Society , 19 (3), 329–346. McCullough, K. M., Wong, Y. J., & Stevenson, N. J. (2020). Female video game players and the protective effect of feminist identity against internalized misogyny. Sex Roles , 82 (5–6), 266–276. McDonald, N., Schoenebeck, S., & Forte, A. (2019). Reliability and inter-rater reliability in qualitative research: Norms and guidelines for CSCW and HCI practice. Proceedings of the ACM on Human-Computer Interaction , 3 , 1–23. McLean, L., & Griffiths, M. D. (2019). Female gamers’ experience of online harassment and social support in online gaming: A qualitative study. International Journal of Mental Health and Addiction , 17 (4), 970–994. Meisner, C. (2023). Networked responses to networked harassment? Creators’ coordinated management of “hate raids” on Twitch. Social Media + Society , 9 (2), 20563051231179696. Obreja, D. M. (2021). Toward a multidimensional streaming: A thematic case study of two Twitch channels. New Media & Society , 146144482110206. O’Connor, C., & Joffe, H. (2020). Intercoder reliability in qualitative research: Debates and practical guidelines. International Journal of Qualitative Methods , 19 , 1609406919899220. Ott, K. (2018). Social media and feminist values: Aligned or maligned? Frontiers: A Journal of Women Studies , 39 (1), 93–111. Paaßen, B., Morgenroth, T., & Stratemeyer, M. (2017). What is a true gamer? The male gamer stereotype and the marginalization of women in video game culture. Sex Roles , 76 (7–8), 421–435. Partin, W. (2019). Watch me pay: Twitch and the cultural economy of surveillance. Surveillance & Society , 17 , 153–160. Perez, S. (2020, December 9). Twitch updates its hateful content and harassment policy after company called out for its own abuses. TechCrunch . Powell, A. (2015). Seeking rape justice: Formal and informal responses to sexual violence through technosocial counter-publics. Theoretical Criminology , 19 (4), 571–588. Powell, A., Henry, N., & Flynn, A. (Eds). (2015). Rape justice: Beyond the criminal law . Palgrave Macmillan. Powell, A., Stratton, G., & Cameron, R. (2018). Digital criminology: Crime and justice in digital society . Routledge. Quil, J. (2020). Survivor stories of harassment/abuse/assault within the gaming live-streaming industry . Google Docs. Radu, R., Kettemann, M. C., Meyer, T., & Shahin, J. (2021). Normfare: Norm entrepreneurship in internet governance. Telecommunications Policy, Norm Entrepreneurship in Internet Governance , 45 (6), 102148. Rayome, A. (2026, March 5). Twitch updates harassment policy to curb online abuse . CNET. Richard, G. T., & Gray, K. L. (2018). Gendered play, racialized reality: Black cyberfeminism, inclusive communities of practice, and the intersections of learning, socialization, and resilience in online gaming. Frontiers: A Journal of Women Studies , 39 (1), 112–148. Ruberg, B. (2021). “Obscene, pornographic, or otherwise objectionable”: Biased definitions of sexual content in video game live streaming. New Media & Society , 23 (6), 1681–1699. Ruberg, B., Cullen, A. L. L., & Brewster, K. (2019). Nothing but a “titty streamer”: Legitimacy, labor, and the debate over women’s breasts in video game live streaming. Critical Studies in Media Communication , 36 (5), 466–481. Ruberg, B., & Lark, D. (2021). Livestreaming from the bedroom: Performing intimacy through domestic space on Twitch. Convergence: The International Journal of Research into New Media Technologies , 27 (3), 679–695. Salter, M. (2013). Justice and revenge in online counter-publics: Emerging responses to sexual violence in the age of social media. Crime, Media, Culture: An International Journal , 9 (3), 225–242. Salter, M. (2018). From geek masculinity to Gamergate: The technological rationality of online abuse. Crime, Media, Culture: An International Journal , 14 (2), 247–264. Shaw, A. (2012). Do you identify as a gamer? Gender, race, sexuality, and gamer identity. New Media & Society , 14 (1), 28–44. Shaw, A. (2014). The internet is full of jerks, because the world is full of jerks: What feminist theory teaches us about the internet. Communication and Critical/Cultural Studies , 11 (3), 273–277. Shaw, A. (2015). Gaming at the edge: Sexuality and gender at the margins of gamer culture . University of Minnesota Press. Sjöblom, M., Törhönen, M., Hamari, J., & Macey, J. (2017). Content structure is king: An empirical study on gratifications, game genres and content type on Twitch. Computers in Human Behavior , 73 , 161–171. Taylor, T. L. (2018a). Twitch and the work of play. American Journal of Play , 11 (1), 65–84. Taylor, T. L. (with Princeton University Press). (2018b). Watch me play: Twitch and the rise of game live streaming . Princeton University Press. Thach, H., Mayworm, S., Delmonaco, D., & Haimson, O. (2024). (In)visible moderation: A digital ethnography of marginalized users and content moderation on Twitch and Reddit. New Media & Society , 26 , 4034–4055. Thomas, K., Akhawe, D., Bailey, M., Boneh, D., Bursztein, E., Consolvo, S., Dell, N., Durumeric, Z., Kelley, P. G., Kumar, D., McCoy, D., Meiklejohn, S., Ristenpart, T., & Stringhini, G. (2021). SoK: Hate, harassment, and the changing landscape of online abuse. 2021 IEEE Symposium on Security and Privacy (SP) , 247–267. Trepte, S., Reinecke, L., & Juechems, K. (2012). The social side of gaming: How playing online computer games creates online and offline social support. Computers in Human Behavior , 28 (3), 832–839. Twitch Blog. (2022, July 5). Introducing our new hateful conduct & harassment policy . Twitch Blog. Vilasís-Pamos, J., & Pires, F. (2021). How do teens define what it means to be a gamer? Mapping teens’ video game practices and cultural imaginaries from a gender and sociocultural perspective. Information, Communication & Society , 1–17. Wolff, G. H., & Shen, C. (2022). Audience size, moderator activity, gender, and content diversity: Exploring user participation and financial commitment on Twitch.tv. New Media & Society . Woodcock, J., & Johnson, M. R. (2019). The affective labor and performance of live streaming on Twitch.tv. Television & New Media , 20 (8), 813–823. Wulf, T., Schneider, F. M., & Beckert, S. (2020). Watching players: An exploration of media enjoyment on Twitch . Games and Culture , 15 (3), 328–346. Yoganathan, V., Osburg, V., & Stevens, C. J. (2021). Freedom and giving in game streams: A Foucauldian exploration of tips and donations on Twitch. Psychology & Marketing , 38 (6), 1001–1013. Zedner, L. (1994). Reparation and retribution: Are they reconcilable? The Modern Law Review , 57 (2), 228–250. Zhao, Q., Chen, C.-D., Cheng, H.-W., & Wang, J.-L. (2018). Determinants of live streamers’ continuance broadcasting intentions on Twitch: A self-determination theory perspective. Telematics and Informatics , 35 (2), 406–420. Zolides, A. (2021). Gender moderation and moderating gender: Sexual content policies in Twitch’s community guidelines. New Media & Society , 23 (10), 2999–3015. Footnotes 1. Examples of descriptors used to evaluate: ‘active streamer on Twitch, 1.4K followers’, ‘eventually banned from the community, based on other allegations and complaints’, ‘eventually banned from the community, based on other allegations and complaints - hasn't streamed since Oct 2020 and no social media presence detectable’, ‘artist in the industry, seems to do work for User X’, ‘hasn't streamed in two years, but works for Activision (in marketing)’, ‘very active Twitch streamer with 20.9K followers - Mixer partner with 49.7K followers - Partnered to Team X’, or ‘was banned from Twitch but unbanned after 8 days, no longer active’. These descriptors were used as a broad brush stroke code of industry status. 2. Two major lawsuits were brought against Activision Blizzard in the US, where most of the AAA companies are HQed. Legal actions in the EU at the moment are focussing on consumer protection, relating to microtransactions, child protection, and addictiveness.IGN provides a good overview of the Activision lawsuits: .