Why Can't Black Women Just Be?: Black Femme Content Creators Navigating Algorithmic Monoliths

In tandem with the rise of social media, new pathways have emerged for the creation of video content for monetary gain [87]. In this section, we explore the history of content creation and what the creator economy affords to content creators. Vlogging (video blogging), fashion, books, gaming and other interests have allowed content creators to explore their personalities and interests. Content creation typically falls into two categories: (1) those who pursue it for full-time financial support and (2) those who use it casually for their own enjoyment [9]. Both options grant financial compensation, however, those who take content creation seriously often create schedules to prioritize their content, investing their own time and money to maintain engagement. During the early 2010s, social media platforms such as YouTube and Instagram created an expansive market for influencers to advertise products, fashion, and ideas to adolescent teens [45]. This content tended to be popular with Gen-Z and millennial viewers. In the early 2010s, YouTube became a popular platform for content due to its ability to reach a wide audience across various social media channels and its support for long-form videos. Crystal Abidin defines influencers as "everyday, ordinary internet users who accumulate a relatively large following on blogs and social media through the textual and visual narration of their personal lives and lifestyles," and monetizing their activity through the promotion of commercial brands in their content [1]. Content creators differ from mainstream celebrities, who are established in other popular entertainment domains (such as modeling, music or acting). The term ‘microcelebrity’ indicates the broadening of content creators’ audiences through social networking sites. A micro-celebrity is defined as a social media user with a relatively large following and a niche audience.[58, 59]. Although they may not have the status of mainstream media celebrities, their fashion, beauty and lifestyle content on social media serves a similar purpose. On TikTok, the creator fund is allocated to content creators who meet the platform's eligibility criteria, allowing everyday people to get paid for sharing their daily lives [62] .

While being able to monetize different aspects of one's life can be rewarding, many researchers have reported negative effects associated with being a content creator. The stigma surrounding influencers often paints content creators as either self-absorbed or contributing to issues of over-consumption. [84]. Additionally, influencers are vulnerable to harassment, unfair content moderation, and doxing. Brooke Duffy explores the psychological impacts that affect content creators that work within the creator economy [27]. More specifically, the stress and boundary conflicts that are afforded on social media. She argues that these issues arise when creators try to differentiate their personal and professional lives online. Duffy indicates that the demands of increased social media engagement can lead to significant emotional labor, which in turn contributes to feelings of exhaustion and burnout among those who contributes to the creator economy. Researchers have specifically explored how LGBTQ+ content creator communities have been affected by these practices. For example, many marginalized content creators feel that they need to prove that their content is being moderated to justify their low engagement [23]. In our research, we examine the experiences of Black femme content creators to shed light on the growing issue of unfair content moderationwhile exploring their unique experiences as content creators. In addition, we explore TikTok through Black online communities to better understand the inequities affecting content creation in marginalized communities.

There exists a substantial body of digital humanities literature investigating the experiences of Black online communities on social media. Here we review Black experiences in online spaces as well as briefly introduce Black Feminist Theory and Digital Black Feminism as key frameworks for interpreting these experiences.To start, many digital humanities and race scholars have argued that race and ethnicity are inextricable from the "culture” of the Internet. For example, André Brock's work utilizes Critical Technocultural Discourse Analysis (CTDA) to explore the relationships of Black users on Black Twitter [11]. In his work he argues that Black Twitter is a collective of Black users who come together in discourse bounded by shared cultural experiences. Brock and Sarah Florini explain that Black Twitter's "live tweeting” works as storytelling [11, 32] where hashtags like #ThanksgivingforBlackFamilies and #BlackBoyJoy create a digital common space of shared stories that are archival [13] due to the finiteness of the internet. This emergence of Black technoculture relates to the Black Feminist tradition of storytelling and signifying [43] through preservation. Black online communities have always played a significant role in digital humanities and the culture and even preservation of the internet.

Online spaces also give way to new negotiations of Black identity - both providing space for identity formation as well as presenting new challenges in racial bias and oppression. For example, Maryann Erigha and Ashley Crooks-Allen researched Black girls’ use of online platforms to engage with identity formation, empowerment, and community through three digital campaigns : Black Girls Rock !, Well-Read Black Girl, and SayHerName [29]. In these campaigns, they see how online spaces enable Black girls to create care networks, assert their voices, and challenge dominant narratives [43] of race, gender, and media representation.

However, at the same time, the racial oppression of our offline world is often remade and reppetuated through digital systems [67]. In the most extreme yet often still "everyday” cases, Black online communities face persistent problems including racist harassment, infiltration of Black spaces, theft of Black cultural production, the digital commodification of Blackness, and excessive forms of content moderation. For example, the online recommendation systems that proliferate social media reinforce racialized biases against Black users [38]. Camille Harris et al. used Twitter datasets for African American English (AAE) dialect and hate speech classifiers to investigate the fine-grained relationship between the specific language of AAE, such as word choice and grammatical features, and hate speech predictions. Harris’ work pushes for more inclusive training data to reduce biases in automated content moderation tools, which disproportionately affect Black social media users [40]. Similarly, Mutale Nkonde argues that racial bias is embedded in facial recognition technologies, impacting Black communities. She explores how these systems are deployed in heavily surveilled urban areas such as Brooklyn, disproportionately misidentifying Black people due to the surveillance technologies’ reliance on biased datasets. Nkonde goes on to argue that this exacerbates racial profiling, over-policing in Black neighborhoods and wrongful arrests [65]. Allison Koenecke's et al. studied how racial disparities in mainstream and broadly used automated speech recognition technologies have significant performance gaps particularly for AAE speakers [54]. Along with this, Tyler Musgrave's et al. demonstrates that online harassment and unwarranted behavior toward Black users led to distrust in social media platforms, resulting in negative relationships with self and others [64]. Gabriela Richard and Kishonna Gray explore how video games repurpose the social structures of race and gender, thus perpetuating exclusionary practices, particularly toward marginalized groups [72]. Hibby Thach et al. explores the often unseen processes of content moderation on Reddit with where moderation practices are disproportionately experienced by marginalized communities [79] . We seek to add to this extensive scholarship by exploring Black femme content creators’ experiences with content moderation on TikTok and the perceptions TikTok imposes on them.

We primarily utilize Black Feminist Theory (BFT) [18, 21, 43] and digital Black feminism [76] to position our work and understand its context. BFT engages the complexity of identity as well as incorporates and understanding of systems of power and pathways for resistance that are common threads throughout the above reviewed Black experiences online. Specifically, we endorse and utilize tenets of BFT such as community building, resisting controlling narratives, and centering Black women's experiences [43] as well as complex and intersectional understandings of identity (i.e., how race, sexuality and class oppression are intersected for specifically marginalized peoples identities [18, 21]) - where our focus particularly attends to race and gender. More specifically, Patricia Hill Collins’ four domains of the matrix of domination illustrate how oppression operates through interconnected systems: structural domains, which organize and institutionalize oppression; disciplinary domains, which administer and enforce it; hegemonic domains, which sustain it through cultural and media narratives; and interpersonal domains, which manifest in individual experiences. Together, these domains reinforce the systemic marginalization of oppressed groups [43].

Catherine Knight Steele argues that the digital intersects within Black feminist tradition and practice through digital Black feminism [76]. Digital Black feminism prioritizes Black women's experiences, voices, and perspectives in digital spaces. It seeks to address and amplify the erasure and marginalization they often face in online discourse (e.g., by recognition of trendsetting by Black creators on the internet [76]). Digital Black feminism seeks to confront and challenge online spaces where misogynoir (i.e., the confluence of racism and sexism specifically targeting Black women [3]) is perpetuated, advocating for platforms to address harassment, discrimination, and hate speech targeting Black women [76]. Digital Black feminism promotes the creation of supportive online communities where Black women can connect, share experiences, and build solidarity. These spaces are essential for fostering empowerment and mutual support [76]. Encouraging and promoting the creation of diverse and accurate representations of Black women in digital content is crucial. This involves challenging stereotypes, showcasing a variety of experiences, and fostering positive narratives that disrupt the idea that Blackness is monolithic [76]. Digital Black feminism utilizes online platforms for activism and advocacy, leveraging social media and digital tools to raise awareness about issues impacting Black women and to mobilize for social change [76].

There remains a lack of exploration into the experiences of Black femme content creators specifically on Tiktok, despite Black Women content creators driving many of the trends on TikTok [7, 37]. We see our work as a contribution toward challenging the online perceptions that affect Black femmes on social media, while also addressing the need for diverse representation of Black women online. Next, we investigate the relationship between critical algorithmic studies and marginalized communities.

Many scholars in critical algorithmic studies have introduced folk theorization as a way of understanding everyday users’ expectations and mental models of how digital platforms work [25, 30, 47, 60, 88]. More specifically, algorithmic folk theories help us interpret how users navigate and document perceived algorithmic unfairness and bias [88]. Brita Ytre-Arne and Hallvard Moe describe five main folk theories the average user believes about algorithms [88]:

• Algorithms are confining - This belief holds that algorithms narrow the user's worldview by feeding them more and more of what the user is interested in [88].
  
• Algorithms are practical - This belief puts algorithms in a positive light where user's rely on them to make technology more efficient and productive. [88].
  
• Algorithms are reductive - This belief holds that algorithms optimize towards stereotypes and make mistakes based on their lack of human experience and identity [88].
  
• Algorithms are intangible - This belief is based on the notion that algorithms’ power is opaque and difficult to understand [88].
  
• Algorithms are exploitative - This belief holds that algorithms use personal data that enables companies to sell users more products [88].
  

In the context of social media, algorithmic folk theories are used and created both by general users and content creators. TikTok, for example, uses recommender systems such as content-based filtering to generate their For You Page [8]. Investigations into the algorithm show that TikTok uses users’ metadata such as likes, comments, followers, and following counts to curate a user's For You Page [8]. Moreover, content creators are aware of this metadata collection and use and have strong folk theories or "assumptions” about how posting time, hashtags, comments, likes, and shares impact the likelihood their videos will be "pushed” to other users’ For You Page [53].

Folk theories are especially critical for marginalized users. We know that our offline world's systemic oppression is very often reproduced and remade through digital systems [67, 75]. For example, Michael Ann DeVito explains in her work that the unfair treatment of transfeminine hypervisibility online leads to adverse reactions on the TikTok platform, such as hate speech, harassment, and aggressive content moderation [24]. Motahhare Eslami's et al. "eye of providence" theory argues that social media platforms, such as Facebook, act like panopticons² that determine what content users see in their feeds [30]. Users believe that the platform watches everything, controls what is shown, and filters out or highlights content based on its own criteria. Essentially, people see the platform as having total control over what appears in their social feeds. Nadia Karizat and colleagues developed "identity strainer theory," which argues that algorithms suppress certain identities while privileging others, allowing some identities to become prioritized [47]. Using this folk theory, Karizat et al. explores how social identities are suppressed.

Throughout the growth of the internet, many instances of physical, emotional, and mental abuse have been perpetuated on social media [51, 61]. As a result, social media platforms have implemented content moderation to mitigate harm by removing content that is considered inappropriate [35]. Traditionally, companies have used human decision-making for content moderation, but this exposes moderators to excessive amounts of harmful content[55]. To reduce human contact with this content, social media platforms utilize machine learning models to classify harm [2]. However, in our research we see that HCI researchers have noticed marginalized content creators reporting these ML models moderate their content unfairly [35, 40].

Samuel Mayworm et al. work found that LGBTQIA users are regularly abused online, leading to negative perceptions of and lack of trust in the social media platforms they use[60]. Many content removals directly contradicted the platforms’ stated goals of inclusion, safety, and freedom of self-expression [22]. Adding to this, Reina Gossett, Eric Stanley, and Johanna Burton delve into this issue in their book Trap Door, by examining the visibility politics surrounding the transgender community [36]. The "Trap of the Visible" in transgender representation influences various aspects of transgender lives when such representation deviates from societal gender norms. Consequently, the representation of marginalized communities is often weaponized or their identities reduced to simplistic narratives. This has prompted discussions around the use of algorithmic folk theorization as a framework for evaluating the design of these platforms, as research shows that many marginalized communities do not trust social media platforms to treat them equitably, keep them safe from abuse, or fairly enforce content moderation policies.

Shadowbanning is another form of content moderation under scrutiny. Researchers in critical algorithmic studies have been working to better define shadowbanning. Content creators have defined it as a decrease in engagement on their profiles or the exclusion of their content from the recommendation algorithm, with the platform often using community guidelines as a justification for the suppression of content [20]. Throughout critical algorithmic research, content creators and everyday users have created theories that justify that definition. Our research contributes to the conversation of shadowbanning by exploring how content moderation affects Black femme content creators relationship with social media.

This project was approved by the authors’ Institutional Review Board. Once approved, we began recruiting participants from March 2024 to July 2024. We created a recruitment screening survey to better filter the study criteria. We prioritized Black non-binary people and Black women who reported content moderation issues or potential shadowbanning. We used the term ‘femme’ to represent gender identities that include queer femininity, [57] rather than adopting a cis-normative perspective.

We promoted the study by sending the survey to content creators from the authors’ personal TikTok following and using the TikTok API to filter through #BlackWomenTikTok and #BlackTikTok. We also promoted the study by searching Black content creators in the authors’ city using the TikTok search bar. Our last form of promotion was through social media platforms such as Instagram and TikTok. The majority of our success came from emailing content creators who listed their emails in their bios and incorporating snowball sampling techniques by asking participants to send our survey to others who would be interested in the study.

After participants filled out our recruitment survey, we adhered to our study criteria for selection. Out of the 30 applicants, 10 were selected for interviews. The remaining 20 creators were not selected because they either had not experienced shadowbanning or had only started creating content within the last six months of the study. We wanted participants with more than six months of experience and a considerable following. We divided participants into two groups: those with 5k to 10k followers and those with more than 10k followers. We did this to ensure an even distribution of newer content creators who started within the past year and those with more established content creation history.

After completing the survey screening, participants scheduled a one-hour interview. Upon completion, they received a $20 gift card.

We conducted semi-structured interviews with 10 participants from various locations across the United States. On the next page we have Table 1 to represent the demographics of our participants.

All interviews were conducted via Zoom video conferencing with only audio recordings retained. All recordings were stored in a password-protected folder and only accessible to the research team. Once interviewees consented to participate in the study, they were asked questions in five high-level categories: Content creator's account and audience dynamics, growth and performance, content moderation, identity and TikTok content, and future on the platform. Below, we define these five categories in more detail.

We used reflexive thematic analysis to analyze our interview data [10, 14] with a mixed inductive and deductive coding approach. Our deductive analysis focused on content moderation experiences, resistance, and folk theories among Black femme content creators on TikTok, while inductive coding revealed broader experiences.

During the interview stage of the study, the first author shared summaries at weekly meetings for the team to review data, refine the approach, and discuss relevant themes. Once data collection was complete, we used Otter.ai to transcribe, copyedit, and clean our interviews. To address the potential biases of transcription AI services [69], the first author carefully reviewed the transcriptions generated against the original recordings. This additional step allowed us to identify and correct any misrepresented words or phrases, ensuring accuracy for our findings. Then the first author used the Atlas.ti software to manually open code all of the interviews. Next, the second author was invited in to further open code three of the interviews to lend a new perspective on the data and further develop themes. The last two authors reviewed some data informally to provide additional perspectives, especially where they had particular topic expertise or experience related to the creators’ content. Throughout the process, we discussed codes and themes in weekly research meetings. We iteratively developed the themes alongside data extracts until our findings were coherent, consistent, and distinctive.

This section addresses the theories Black femme content creators had while on TikTok. We present how participants perceived themselves online as well as how they felt algorithms perceived them online. We then present the emotional and invisible labor that participants felt was associated with being a Black femme Content Creator.

Overmoderation of participants’ content was observed consistently throughout our study. In this section, we report how participants felt about their content being moderated. We split this section into 2 parts: ambiguity surrounding why their content was moderated, and the idea that their looks and ideas affect the heightened moderation they experience compared to non-Black femme content creators.

The importance of making a collaborative community online was frequently mentioned by Black femme content creators as a driving force for creating content. This theme emerged in response to questions about why these creators stayed on the platform and what brings them joy as creators. When asked how she measures success on the platform, one participant said “I would say that kind of how I measure my success is through like the interpersonal relationships that I'm able to cultivate” (P05). She went on to explain that much of her success in growing a community online came from her followers personally reaching out to her for advice on applying for graduate school applications. This relationship between the content creator and her followers was interesting because it extended beyond online interaction into real life. Another participant noted that she never really thought about leaving TikTok, going on to say “even if I take a break. I always go back just because, I kind of feel like it's my little family. You know? I actually care for my followers” (P05). This participant, like many others, also noted that she tended to use the comment section to keep track of her online community.

Participants that had been on the platform for over 2 years expressed how hard it is to build community online.One said that “The hardest part when you start is building a community. First of all is figuring out exactly like what they want to see from you” (P05). Here, we see how the content creator’s understanding of her followers helped form a strong community. This was common for many of the content creators due to their under-representation in the content they create. A fitness content creator that had been making content since 2019 said “I really believe in the building the community part, that's so important to me, because at the end of the day, maybe my content will switch up in a few years, and I want to explore something else, you know, but if I have people that follow me, because they just like who I am as a person, they'll ride with me for anything, you know, instead of, somebody that just followed me because they saw my posts on my 10 pound weight loss” (P04). The theme of authenticity and community building is apparent in this content creator’s observation that in her time building a following i.e. her followers’ tendency to see who she was as a person beyond her fitness content correlated with a community that appreciated her as her.

We also see aspects of community in the ways participants navigated shadowbanning or overmoderation. When defining shadowbanning participants believed that shadowbanning was a passive aggressive way of suppressing ones content. To counteract this tactic, BFCCs relied on their followers and friends in real life to check if their content was being banned. One participant describes this strategy: "So I've asked my friends, like my other content creator friends to go on my page, and they don't see it [my content]. So that, to me, is shadowbanning, like some of my, users can see it, some people can't, even if they look me up" (P10). In this example, the content creator uses her network of other content creators to determine whether her content had been moderated or shadowbanned. P10 also noted that TikTok designed a campaign to give Black content creators a platform to report instances of overmoderation; however, the campaign's effectiveness was limited because it primarily benefited creators with larger followings, leaving those with smaller audiences with less influence and visibility.

“Everyone coming of age in the digital era has practiced this online performance of self. But Black women considered deviant and “other” in American society had extra practice in navigating their sense of self in stark contrast to societal expectations.” [76]

Our participants want freedom from the constraints imposed by the algorithm. They describe TikTok as a result of our larger societal inability to view Blackness as complex and diverse, preferring instead to hold onto stereotypes of race, gender, and class and viewing anyone outside of those stereotypes as anomalous. We connect these sentiments to Andre Brock's work in Critical Technoculture discourse analysis (CTDA) [12, 13] and more specifically the notion of the “libidinal economy” [12, 13, 63, 86]. CTDA investigates the internet and online behaviors with a critical perceptive on culture and has been increasingly used in HCI research (e.g., [16, 70, 81]).  This framework draws focus away from the Western deficit perspective on minority technology use but instead emphasizing the knowledge and experiences of underrepresented groups of technology users. In reference to Brock, a libidinal economy emphasizes “the role of emotional and psychological intensities in driving anti-Blackness, rather than the more rationalist models of human behavior derived from political-economic approaches” [78]. Brock argues that the constant portrayal of Black suffering that is pervasive on the internet and in other forms of media reinforces the anti-Blackness we see today. Social media algorithms amplify images of Black suffering, thus perpetuating a narrow, one-dimensional, monolithic view of Blackness. Our participants clearly echo this sentiment, as they face challenges due to some of their content not fulfilling monolithic representations. The datafication of injustices described by Ruha Benjamin makes the negative viewing of Blackness a normality on the internet[4].  In this way, it is not just the users, but the algorithm on TikTok optimizing Black femme content creators towards these confining margins. The TikTok algorithm prioritizes digital narratives [76] of harmful stereotypes of Blackness rather than optimizing towards joy and participants’ own nicheness.

Monolithic views of Blackness are prevalent on TikTok and most social media platforms [31]. Those with more algorithmic privilege [47] can be multidimensional on the internet. For example, TikTok facilitates microtrends [6] ( e.g., Fleabag Aesthetic, Clean Girl, Aesthetic, BarbieCore, and Coquette ). Formal content creators and general users can display different sides of their personality by engaging with micro trends through different posts, including different outfits, styles, sounds, and other creative choices. However, these micro-trends, at best, only materially benefit cis white women and, at worst, explicitly exclude Black femmes. While on the surface, microtrends are just a new form of social engagement online, trends such as these have always seriously impacted the public's identity exploration, formation, and feelings of representation and belonging [6, 26].

Reina Gossett, Eric Stanley, and Johanna Burton explore identity formation in their book Trap Door , where they discuss visibility politics for the transgender community [36]. The ’Trap of the Visible’ for transgender representation affects aspects of transgender lives when the representation does not fit into societal norms of gender. As a result of not fitting into these norms, marginalized communities’ representation is weaponized and/or identity flattened. Identity flattening, coined by DeVito, is the minimizing of someone's identity, not allowing them to be multifaceted [24]. In the context of social media, the identity of Black femme content creators is confined to monoliths, not allowing their content to be represented in other micro-trends afforded to those more algorithmically privileged. Black femme content creators’ visibility is scrutinized if they do not fall into the monoliths the algorithms designed for them. Our participants’ "trendy" content was actively suppressed, and when it was not, they faced repeated rebuffs (e.g., surprise at Black femmes’ interest in the trend, feedback that they don't actually fit the trend, etc.). For social media algorithms to advertise the freedom to be oneself but quietly select a few to promote speaks to Ruha Benjamin's argument that the datafication of Black pain shapes how Blackness is seen through the internet [4]. This cycle can become dangerous due to social media's influence on identity formation, which affects Black online communities’ perception of self and can affect young Black users who are highly influenced by the internet.

In addition to CTDA, we observed participants’ experiences through their resistance to monoliths. Black feminist scholars have discussed the importance of pluralism when explaining Black people's experiences within the everyday world. In our study, participants expressed their intersectional identities through the nicheness of their content. They described leaning into their unique interests and experiences through their content and audience curation. They embody Patricia Hill Collins's emphasis on the multiplicity of Black women's experiences [43]. Collins argues against the notion of a single, universal experience of Black women, instead advocating for the recognition of the diverse ways in which Black women live and resist oppression.

bell hooks describes self-actualization as “aware[ness] of oneself and the influence one has in the transformation of the lives of others through the construction and dissemination of knowledge” [44]. We observe Black femme content creators self-actualizing via their online platforms by embracing their talents, interests, and lifestyles free of the assumptions that are perpetuated by racism. However, BFCCs’ identities tends to be reduced by social media through the stereotypes of the jezebel ⁴, mammy ⁵ or sapphire ⁶, making it hard to reach self actualization. We argue that the algorithm encourages this rhetoric. [85]

Computer scientists and HCI researchers alike have already found that algorithms are biased. They are not purely logical and rational, and instead they perpetuate racial, gender, and other identity-based discrimination (e.g., [40, 66, 67]). Black femme content creators specifically navigate an algorithm that prioritizes the “Black Barbie Aesthetic” (P05) or a certain type of Black woman that is exceptional in what they do (P02, P03). In Collins's matrix of domination, we see how social media algorithms hold these domains of algorithmic power by organizing, sustaining, enforcing, and manifesting aesthetics, thoughts and trends on the internet and deciding who can be the face of them.

This networked circulation of what is a Black femme content creator is developed in Brock's argument of “racism without racists,” in which imagery of anti-Blackness is distributed via social media algorithms rather than through a person perpetuating the racism. Through this monolithic view of Black femme content creators online, pluralism and diversity are hindered. However, despite the algorithm prioritizing pain over joy, we see content creators resisting through building tight-knit communities of care among other content creators and their followers.

Identity was formed both around participants’ day to day lifestyles and around their role as Black content creators. However, our participants speculated that the algorithm made their physical appearance and Blackness hypervisible to everyday users. This contextualizes their over-moderation and shadowbanning, sparking conversation around what gets views and what does not. For example, P05 felt frustration with her predominantly male audience and felt very sexualized on the platform. She believed that, as a Black woman, her body was more sexualized than most due to the stereotypes associated with Black women's bodies. While this participant aimed to share content that would resonate with those interested in her journey as a student athlete and contribute to the representation of Black girls in higher education, she felt that much of her content unintentionally catered to the male gaze. At times, she even felt pressured to lean into the male gaze to gain more views.

Other participants noted how their appearance was made incredibly visible by the algorithm in respect to either their hair or their clothing. A non-binary content creator expressed that they tended to “play up femme” (P08) because they noticed that their content would be promoted more on the platform if they adhered to traditional forms of femininity. They later went on to say that offline they present more androgynously. Another participant, who created content about graduate school, resisted the “Black  excellence”  narrative she saw being pushed on TikTok for Black content creators pursuing post-graduate education. Building on this, we believe the Black exceptionalism narrative further complicates the experience for Black creators, as it imposes expectations of what it means to be a "good" Black person online. This leads us to ask: What is a "good" Black content creator in the context of social media feeds? The Black exceptionalism narrative is imposed on Black creators online as a way of of defining what it means to be a good Black person. Upward Black mobility, both online and offline, is indicative of Black people attaining positions of high power that disrupt negative perceptions of Blackness. However, these "negative" perceptions are applied to those who do not follow social norms brought on by white hegemonic capitalist power structures of success, which results in the othering of Black people who are perceived as "different" or "ghetto". Scholars have explored the damages these narratives can create as they push the idea that to be Black and successful, you must be  extraordinary [19, 71]. The participants in our study experienced frustration regarding how they think the algorithm perceives them. The central theme that emerged from our study was: Why can't Black women just be? (P02, P03, P04, P05) We observed that the offline pressures and perceptions of what it means to be a successful Black woman have now extended to the online space, causing Black femme content creators to be acutely conscious of how they present themselves, what they say, andwhat they do before posting.

Our results regarding algorithmic folk theories for Black femme content creators reflect past research on folk theories of social media algorithms, such as Karizat's strainer theory [47] as well as Brook Duffy's work on the labor of social media consumption [27]. Karizat argues that social media algorithms recognize, classify, sort, and suppress social identities due to the social construction of these identities. In this folk theory, the algorithm is represented as a strainer impacting what types of identities are prioritized on social feeds. Building upon strainer theory, Karizat recognizes algorithm privilege, which refers to the advantages that arise from algorithms prioritizing certain identities over others. We argue that stereotypical views associated with Black women are more optimized on these feeds. The unwarranted hypervisibility driven by the algorithm forced the content creators in our study to become highly aware of their Blackness. This hypervigilance was manifested in their clothes, hair, gender expression, sexuality, or career. In these instances, we see how content creators who lack algorithmic privilege must perform additional labor compared to other content creators. Building on Brooke Duffy's work [27], we see how Black femme content creators navigate the dual pressures of offline societal norms imposed on Black femmes and the online stereotypes they must either resist or embrace within social media spaces.

As our study revealed, the mental and emotional hurdles that marginalized individuals must consider before posting online are significant. We build on the work of HCI scholars like Brock, DeVito, and Klassen in trauma-informed computing. At the same time, other scholars have described how TikTok's interface empowers users to engage in a Black feminist praxis (e.g., through the green screen feature that enables users to teach and critique each others’ content [68]). We also advocate for a shift in how Black communities are viewed online, moving away from a deficit mindset towards a more joyful perspective [81]. In the following sections, we outline the benefits of building social feeds rooted in joy rather than struggle.

In our research, we observed that Black femme content creators navigate the platform's over-moderation by consciously adjusting their appearance or lifestyle, often aware that they are being monitored and moderated. This hypervigilance can be exhausting, but these content creators are inspired by the relationships they have formed online. For example, some participants described being more comfortable with their followers knowing that they were interested in the creator's authenticity, while others created their own measures of success (e.g., the number of followers who reach out for advice). Those who felt like their content was being shadowbanned or overly moderated would rely on their followers for feedback, posting videos to ask if their content was being shown on their feeds or, in extreme cases, asking followers to follow them on other platforms. These are care networks the creators cultivated through their followers online and offline. Despite the negative experiences BFCCs had while navigating TikTok, we saw BFCCs relying on their followers (e.g. having meaningful conversation through comments, giving or taking advice as well as sharing inside jokes relevant to the Black experience) and showcasing their diverse cultures online to cultivate joy (e.g., dancing, cooking, makeup and haircare). We propose that these meaningful connections could help with exploring the implications of social feeds that are rooted in Black joy:

‘Black joy allows us space to stretch our imaginations beyond what we previously thought possible and allows us to theorize a world in which white supremacy does not dictate our everyday lives.’ [46]

Prioritizing Black joy as the ultimate goal in the design of future social technologies, we look to the many interrelated approaches proposed by HCI and technology researchers and designers for centering BIPOC and specifically Black experiences. First, trauma-informed design encourages us to confront head-on that many users have histories of trauma that are often re-perpetuated or aggravated by social technologies [70]. To that end, Randazzo and colleagues discussed tangible ways for researchers and designers to build for marginalized communities in ways informed by their traumas, such as adopting principles of trustworthiness, transparency, and direct engagement with social and cultural differences [70]. In the case of Black femme content creators, the hypersexualization and other stereotyping that is imposed on Black women offline translates to algorithmic constraints and overmoderation online. Here, transparency in the user interface for creators—specifically, sharing more about how the algorithm reads BFCCs’ content—could mitigate the physiological and psychological harm associated with uncertainty regarding digital racial microaggressions [82].

Second, To, Smith, and colleagues suggest moving beyond a deficits framing to design that focuses on the flourishing and joy of people of color [81]. They integrate value-based design with transformative justice and other design protocols, allowing care networks for marginalized users to be centered around joy rather than struggle. We encourage designers to focus on building tools for the care communities already curated by content creators. For example, allowing followers to opt-in to creator notifications for new content (as YouTube does with the notification bell) and emphasizing the visibility of the creators that users already follow could support creator-community relationships.

Finally, focusing on Black experiences, researchers have proposed Afro-futurism as a framework for exploring ways to help Black women and non-binary users have joyful experiences with technologies [52, 75]. In our research we saw that Black joy was represented through dancing, shared experiences and care networks between followers and creators. We believe that focusing on these pluralistic Black experiences on the internet will create a positive user experience online for BIPOC users. Many researchers have been laying the foundation for social feeds that are informed by Black joy rather than constant forms of stereotypes, struggle, and trauma. Lisa Egede and colleagues have explored how to design technology for Black experiences that are "For Us By Us" by studying Black technologists’ work on joyful technologies [28]. In their research, they studied Black technologists who curated resources to support lived Black experience rooted in joy. Through their study, they recognized that Black culture is shared through social media. They also observed that the monetization of Blackness on social media is often paired with a failure to center Blackness in the design of these applications. With our participants, we observed that BFCCs’ Blackness was hypervisible, often resulting in consequences like overmoderation or shadowbanning. This prompts a discussion about designing for Black users on social media platforms like TikTok, where their physical appearance is prominently displayed, as opposed to platforms like Twitter, which has been more extensively explored in Black Techno-studies. This sentiment was evident when one participant (P10) reflected on how her content might be perceived differently if her Blackness were not so visibly expressed. The dimming of one's Blackness should not be the solution to addressing injustices on the internet, as it undermines the push in HCI research towards building for joy rather than struggle for marginalized communities.

Egede and colleagues also argue for the need to be intentional about the recruitment of minoritized groups. We learned from one participant that TikTok created a resource for Black content creators in 2020 to help with overmoderation and shadowbanning. However, the resource has not been maintained, making Black content creators rely on their care networks. Egede et al. observes that this phenomenon of a lack of full support for Blackness in design is likely due to non-Black participants in their study lacking an understanding of the nuanced needs and aspects of Black communities. We also believe that the push for diversity and inclusion tactics within the United States is greatly used as a one size-fits all approach to combating racial contentions. However, without naming and understanding the intersectional identities and histories rooted on why these technologies are racialized a cyclical pattern will occur. Making more space for unsustainable design recommendations for BIPOC users. We advocate for increased efforts to develop sustainable support systems for content creators, with a particular emphasis on including Black designers and engineers in building these social feeds.