2 Theoretical framework: Baradian relational ontology, intra-action, diffraction and the constitutive power of law

One of the least visible and most neglected spaces or subjects in current scholarship on content moderation is, of course, the constitutive potential of the juridical system of the Global South – that is, its capacity not only to reflect and mirror existing subjects, objects and relationships, but to actively constitute them. As Judith Butler, in her book Gender Trouble, explains (drawing on Foucault’s relational conceptualisation of power):

Thus, Butler, drawing on post-structuralist theory, argues against the notion of pre-existing, stable legal subjects and objects. Instead, she holds that identities and social categories (especially those related to gender) are not inherent but are performatively constituted through discursive, juridical practices. This means that the subjects, objects and contexts of law are not pre-given but are produced and reproduced through the very legal discourses that they claim to only describe, interpret, regulate or represent. This process of subject formation is what makes juridical discourse not just a form of negative power – where power is seen purely as restrictive or oppressive – but also a productive power because it generates and constitutes the subjects, identities and relationships within its purview and, therefore, actively participates in the creation of social realities rather than merely reflecting or controlling them. In other words, Butler is suggesting that it is within the relational dynamics between law and the subjects of its power – that is, the processes of naming, categorising, adjudicating and regulating – that ontology (what things are, or the very nature and existence of entities) emerges (Bakker Reference Bakker2007; Braedley and Luxton Reference Braedley and Luxton2021; Buchanan Reference Buchanan2008; Dinner Reference Dinner2023; Silverstein Reference Silverstein2003; Zhurzhenko Reference Zhurzhenko2001). This is, in many ways, consistent with Karen Barad’s relational ontology and concept of ‘intra-action’ in the field of new feminist materialism (Barad Reference Barad2007).

Much like Barad’s concept of intra-action, Butler’s idea of the constitutive power of law, and of co-constitution, blurs the boundaries between the ‘inside’ and ‘outside’ of law. According to Barad, just as particles in quantum physics do not exist as independent entities but are co-constituted through their intra-actions, relational entanglements and unfolding, so too are the subjects and objects of law co-constituted through their entanglement with and within juridical discourse.

By way of aquatic analogy, which is quite common in Baradian discourse and analysis (Arlander Reference Arlander2020; Barad Reference Barad2007; Murris and Bozalek Reference Murris and Bozalek2022; Tisha Dejmanee and Dejmanee Reference Tisha and Dejmanee2016), imagine a wave approaching a shoreline. In traditional liberal ontologies, we might think of the wave as an independent entity and the shore as a static boundary. However, Barad’s concept of intra-action shifts this understanding: the wave’s characteristics – its height, speed and form – are shaped by its dynamic intra-action with the shoreline. The shore’s slope, texture and composition influence how the wave behaves, while the wave, in turn, re-shapes the shore with each collision in ways that alter its contours and generates new landforms over time. This ongoing process of co-constitution means that the wave and the shore continuously shape and re-shape each other and blur the boundaries between them. They are both cause and effect: each influences and is influenced by the other in a continuous feedback loop, where the actions of one become the conditions for the other. They operate in a dynamic system where their identities, forms and behaviours are in constant flux in ways that make it difficult to distinguish where one ends and the other begins.

In this context, the notion of distinct, self-contained entities gives way to a more fluid understanding of relationality, where what each ‘is’ cannot be separated from what it ‘does’ in relation to the other (Mauthner Reference Mauthner2021; Murris and Bozalek Reference Murris and Bozalek2019). The wave and the shore are no longer seen as independent actors but as co-constitutive elements of a single, evolving process. Each encounter leaves an imprint that transforms both, creating new possibilities for future intra-action and making it nearly impossible to define one without reference to the other.

As such, what Karen Barad is suggesting (much like with Butler and juridical discourse) is that constitutive power – meaning the capacity to shape, form and bring into existence – lies in the process of entanglement itself: the intra-actions, relationships and connections that give rise to the entity or phenomenon in question (Murris and Bozalek Reference Murris and Bozalek2022). It’s within these processes of entanglement that the identities, properties and boundaries of things are formed and continually (re)constituted. So, what something ‘is’ can be understood as a relational outcome of the constitutive power of its intra-actions with other entities and forces. As Barad put it: ‘We are being churned by the soil, the wind, the foggy mist. A multiplicity, an infinity in its specificity, condensed into here-now. Each grain of sand, each bit of soil is diffracted/entangled across spacetime’ (Barad Reference Barad2014: 184).

In other words, relationality is the constitutive power of ontology, meaning that it is the force or agency that actively creates and defines the very nature of being and existence. This is, of course, what feminist new-materialism and post-humanism, especially cyborg studies have suggested in collapsing rigid distinctions between body and object, human and machine and nature and culture. The boundaries we perceive between these categories are not fixed but are constantly being renegotiated through intra-actions or socio-technical assemblages (Amrute Reference Amrute2019; Arndís Bergsdóttir and Bergsdóttir Reference Arndís and Bergsdóttir2017; Åsberg and Lykke Reference Åsberg and Lykke2010; Barad Reference Barad1998; Hamraie Reference Hamraie2015; Haraway Reference Haraway1985; McNeil Reference McNeil1992).

And, if everything is constituted through relational processes and entanglements (and as such is best understood as what I term a ‘relatant’), then it becomes nearly impossible to position anything – including social identities and power and the very idea of autonomy – as existing independently or outside of these relationships. As Black feminist scholarship in the field of intersectionality studies has made clear (Crenshaw Reference Crenshaw1991; Cho et al. Reference Cho, Crenshaw and McCall2013) positioning identities and power as existing outside of these relationships would mean trying to consider them in isolation while ignoring the very constitutive processes that define and shape them (Barclay Reference Barclay, Mackenzie and Stoljar2000; Fletcher Reference Fletcher1998; Mauthner Reference Mauthner2021).

What Barad suggests is – if we read this into Butler’s constitutive understanding of law – that it is at these points of relational encounter, where law intra-acts with its subjects and objects, that ‘diffraction’ emerges (Barad Reference Barad2014). That is, imagine light passing through a small aperture or encountering an obstacle. In classical physics, you might expect the light to simply pass through or reflect off the obstacle, maintaining a clear boundary between the light and the obstacle. However, what actually happens is more complex: the light bends, spreads out and creates an interference pattern of alternating light and dark bands, where the waves of light overlap and are entangled. In some areas, the waves reinforce each other, producing brighter, more intense bands of light, while in other areas, they cancel each other out, forming darker, shadowy regions, creating a shimmering, ripple-like effect that radiates outwards. This pattern is known as diffraction (Barad Reference Barad2007, Reference Barad2014; Bozalek and Zembylas Reference Bozalek and Zembylas2017; Merten Reference Merten2021). In physics, diffraction patterns provide insights into the wave-like nature of light and how it behaves under different conditions. These patterns are a visual representation of how even small changes in the environment or conditions – such as the size of an aperture or the presence of an obstacle – can significantly alter outcomes and lead to new, emergent properties (Bozalek and Zembylas Reference Bozalek and Zembylas2017).

According to Karen Barad – drawing from physics as well as the post-humanist and feminist materialist work of Donna Haraway (Åsberg and Braidotti Reference Åsberg and Braidotti2018; Haraway Reference Haraway1985; Sehgal Reference Sehgal2014) – diffraction, when applied to social life, allows us to see differences (or ‘interference patterns’) that matter – that is, to study social patterns in a way that captures or acknowledges differences that matter. Barad uses the concept not just to describe a physical phenomenon, but as a methodological tool for analysing how various forces, identities and power structures interact and produce outcomes. In this context, diffraction is about more than just observing how things interact; it’s about paying attention to the specific ways in which these intra-actions generate new forms of meaning, identity and power (Barad Reference Barad, Giffney and Hird2016). By examining these patterns, we can identify the differences that matter – that have significant social, political or ethical implications, for as Haraway puts it, ‘diffraction is a mapping of interference’ (Haraway Reference Haraway, Grossberg, Nelson and Treichler1992). Its specific characteristics and patterns, where there is constructive or destructive interference, tell us something about how these elements intra-acted, or, as Barad explained it, are ‘materialised differently through different practices of contingent ontology’ (Barad Reference Barad2007: 89).

In this sense, the diffractive process is more than a simple reflection or simple extension of the pre-existing entities involved in the process. Instead, it represents a transformative moment or space of potentiality where the intra-action between these entities produces something new – something that contains elements of the original components, but also differs in significant ways. Just as in physical diffraction, where waves of light overlap, intertwine and either reinforce or cancel each other out, this social diffraction can challenge existing structures and create new pathways that are brighter and more intense in certain areas, or forming darker, shadowy regions where certain elements are diminished or negated. As Barad explains, it is like examining the ripples created by a stone thrown into a pond (Barad Reference Barad2007: 28; Murris and Bozalek Reference Murris and Bozalek2022).

In the context of legal processes, Barad’s concept of diffraction can, as Butler argues (Butler Reference Butler1990: 2), be understood similarly. When law encounters its subjects and objects, it does not only encounter pre-existing structures or simply reflect what is already there. Instead, these encounters can reinforce existing social and legal patterns (like established norms and practices, inequalities and systemic and structural, socio-economic and cultural formations), generate entirely new pathways and forms (such as new interpretations of the law or novel legal precedents or socio-cultural or economic formations) or produce outcomes that both extend and transform what existed before. That is, the encounter doesn’t just lead to a complete overhaul or replacement of what was there before (transformation). It also implies that there’s a continuation or stretching out of certain elements from the original context (extension). So, the result is a blend of the old and the new, where some aspects are carried forwards and modified, while others might be entirely novel, which is why Butler cautions that ‘an uncritical appeal to such a system [i.e. the constitutive and diffractive power of law] for the emancipation … will be clearly self-defeating’ (Butler Reference Butler1990: 2), for it holds both the capacity to reproduce and to constitute new legal subjects, objects and social relations.

If we apply this to the context of the Global South – and take a decolonial orientation by centring the juridical systems and formations of the Global South (Shahid and Vashistha Reference Shahid and Vashistha2023; Zokaityte and Mbioh Reference Zokaityte and Mbioh2023), especially in the context of their encounters with global capitalist entities and forces (Crawford, Reference Crawford2018; Robin Bellingham and Bellingham Reference Bellingham and Bellingham2022; Weir Reference Weir2017) –diffractive readings allow us to go beyond subjection-centred framings by revealing the potential for agency, resistance and social reproduction, which is often overlooked in much of current scholarship on content moderation and the Global South (Crawford, Reference Crawford2018). By focusing on the differences that matter (Barad Reference Barad2007: 89), a diffractive reading can uncover the subtle but significant ways in which legal encounters in the Global South produce new forms of relational dynamics and social configurations.

This of course folds in important ways into long-standing post-colonial engagements with resistance and subaltern agency. Scholars such as Gayatri Spivak (Reference Spivak, Cain and Harrison2023), Balakrishnan Rajagopal (Reference Rajagopal2003), Boaventura de Sousa Santos (Reference de Sousa Santos2015) and James Scott (Reference Scott1985) have foregrounded the epistemologies and political practices of the Global South – whether in the form of subaltern silence, legal pluralism, insurgent jurisprudence or everyday forms of resistance. My approach builds upon this literature but departs from it methodologically (de Sousa Santos Reference de Sousa Santos2015; Rajagopal Reference Rajagopal2003; Scott Reference Scott1985; Spivak Reference Spivak, Cain and Harrison2023). Where these traditions often begin with the presumption of situated subaltern agency acting within or against imposed juridical orders, the Baradian notion of intra-action refuses any pre-existing ontology of actors or systems. Legal power, on this view, does not simply act upon marginal subjects or become the terrain for their resistance; rather, subject and system are co-constituted through their iterative entanglement. In this sense, diffraction does not only attend to resistance within structure, but to how both ‘resistance’ and ‘structure’ emerge through dynamic relationality. This ontological commitment distinguishes the Baradian/diffractive reading from the otherwise allied critiques in post-colonial and Third World legal thought. That is, while Scott, de Sousa Santos, Rajagopal and Spivak show how subaltern or peripheral actors resist, reinterpret or endure within asymmetrical global legalities, a Baradian/diffractive approach invites us to examine how the very categories of ‘subaltern’, ‘resistance’ and ‘legal order’ are themselves materially enacted through intra-active encounters. It allows us to trace how juridical subjects and normative structures do not pre-exist their engagements, but are constituted through them – often in ways that both reproduce and exceed the logics of domination they appear to inhabit. Diffraction, in this sense, foregrounds the constitutive power of entangled difference: not as simple deviation or opposition, but as a generative patterning through which multiple, often conflicting, relational possibilities are enacted simultaneously – producing overlapping, interfering and co-existing configurations of subjectivity, legality and agency that do not resolve into singular or stable forms, but allow for what Barad calls the ‘cutting together-apart’ (Barad Reference Barad2014).

What this brings – specifically, and in distinction from otherwise allied critiques in post-colonial and subaltern legal thought – is a shift in how we frame the analytic terrain itself. Where post-colonial scholarship has convincingly shown how Global South actors resist within and push back against dominant legal orders, a Baradian/diffractive approach asks how those very legal orders – and the figures who appear within them as ‘subaltern’, ‘agent’ or ‘victim’ – are themselves materially enacted through their entanglement. Rather than beginning with presumed asymmetries and then asking how those might be resisted or endured, a diffractive reading begins with the intra-active process through which such asymmetries – and the legal forms that sustain or contest them – come to matter at all. In doing so, it offers not simply a different stance but a different question: not whether law resisted or reproduced domination, but how law – as apparatus – cut differently, enacted difference and generated relational patterns with uneven and often messy, contradictory effects.

This does not, accordingly, negate (or even attempt to question) the insights of post-colonial legal critique – it depends upon and extends them – but it reframes the inquiry: from the politics of resistance within structure, to the ontological conditions through which resistance, structure and subjectivity come into being as diffractive effects. It invites us to stay with the instability of juridical forms, not to resolve them, but to read their patterned interferences – to trace how what appears as closure might also mark the trembling aperture of a yet-unfolding cut. A good illustrative example of this is the Arendse case in Kenya, to which I shall now turn.

3 Arendse case: subjection-centred accounts, slave-like conditions and disposable, affective labour

There are multiple ways to interpret or tell the story of the Arendse case. Depending on the aspects of the situation that one chooses to focus on, different narratives or accounts can emerge. If we take a subjection-oriented reading (based on the testimonies of the workers in the case and the findings of fact of the Kenyan court), this story begins with a powerful multinational corporation, Meta, which faced the considerable challenge of managing an immense volume of disturbing and graphic content on its platforms – Facebook, Instagram and WhatsApp. To keep these platforms running smoothly twenty-four-seven, Meta needed to ensure that this content was swiftly and effectively moderated – a task that was both highly labour-intensive and expensive. To meet these demands, Meta turned to a Kenyan company, Samasource Limited (Sama), among others in the Global South.

The content moderators who took on this harrowing work were recruited by Sama from some of the most ‘war-torn countries’ (Arendse, para. 3(k)) of East Africa – Somalia, Sudan, Eritrea and Ethiopia. Many of them were refugees, having fled their homes in search of safety and a better life (Arendse, para. 6). They were lured into this job under the false pretence that they would be working as call centre agents. Upon arrival in Kenya, however, they found themselves thrust into the role of content moderators (Arendse, paras 30–31). For example, one moderator, Mahlet Yilma Lemma, recounted that in April 2019, while residing in Ethiopia, she learned that Sama was recruiting for the job of an Amharic-speaking call centre agent in Nairobi. The job description was vague, but the organisation promised to process work permits, provide accommodation for one month and cover the airfare. Successful applicants were expected to be ready to move within ten days. Desperate for employment, she applied, only to find during the training in Nairobi that the job was not what she had been led to believe. It became clear that she would be working as a content moderator for Facebook – a job that would require her to confront some of the most horrific content online.

The work was gruelling. Every day, these moderators were exposed to the most graphic and violent content imaginable. One content moderator recounted: ‘Since the day I was assigned to review contents for the Tigrinya/Amharic market, my daily routine at Sama Source was to watch graphic videos back-to-back from the Ethio-Tigray war … the majority of the content I reviewed daily included mutilated or dismembered bodies, sadistic videos depicting manslaughter and burning of persons alive among others’ (Arendse, para. 4) – a task made even more harrowing by the fact that the moderator came from this region, making the atrocities not just distant horrors but deeply personal, as they could relate to the cultural, social and emotional contexts of the violence, and may even have recognised places, people or situations depicted in the footage.

The psychological toll was immense, as the moderator not only had to process the horror of what they were seeing but also grapple with the emotional resonance of knowing that these events were occurring in their homeland. This experience triggered feelings of survivor’s guilt, anger and an overwhelming sense of powerlessness, as they were tasked with the impossible task of moderating content that depicted the suffering of their own people (Arendse, paras 3–4). The job felt like a relentless assault on their mental and emotional well-being, with the blurring of the line between professional work and personal trauma leading to a deep internal conflict and lasting impact on their mental health. Moderators described how they came to the job ‘whole’ but were leaving as ‘broken shells’ of themselves (Arendse, para. 5). One moderator called the job ‘a dreadful experience draining them physically, mentally, and emotionally’, leaving them with ‘personality and psyche completely changed’ – from ambitious and sociable individuals to ones who were ‘bitter, avoidant, cynical, paranoid’ with anti-social tendencies (Arendse, para. 4). Likewise, another moderator vividly recalled their first experience witnessing a ‘live manslaughter’ on video (a real-time killing broadcast over the internet) related to their homeland: ‘I unconsciously stood up and screamed … I almost forgot where I was and who I was’ (Arendse, para. 4). Yet, the ‘wellness counsellors’ provided were not qualified to help, with one moderator lamenting that:

The relentless pressure imposed by Meta’s enterprise resource planning (ERP) system only deepened their despair. This system tracked their every move – including when they took breaks – which, by implication, meant even monitoring their toilet breaks (Arendse, para. 20(I)) – and tied their job security to their ability to keep up with an unrelenting pace. Content moderators were expected to ‘action between 500 and 1000 pieces of content per day’, with the understanding that ‘the average handling time (AHT) for a ticket was 60 seconds and below per ticket [a term used to describe individual content items flagged for review, such as posts, images or videos]’ (Arendse, para. 4). This meant that moderators were expected to process one piece of content every minute, amounting to sixty pieces of content per hour, and up to 480 to 720 pieces over an eight-hour work shift. Any deviations from this benchmark were considered poor performance (Arendse, para. 4). The system was so unforgiving that moderators testified to being put under ‘Performance Improvement Plans (PIP)’ – a process designed to correct perceived underperformance by setting new targets and deadlines, which pushed them to meet these unrealistic targets, with the constant threat of termination hanging over their heads if they failed to comply (Arendse, para. 15).

The ‘extremely toxic and dangerous’ (Arendse, para. 3) work environment only heightened their sense of being expendable tools in Meta’s global operations. As Mahlet Yilma Lemma, one of the moderators, explained: ‘The workplace was incredibly toxic … I felt voiceless … treated like garbage … subjected to a system where the content sent to me was extremely random, vulgar, graphic, and truly not content meant for human consumption’ (Arendse, para. 7). She further described how this work had left her suffering from insomnia, forcing her to rely on sleeping pills, with her mental health irreversibly damaged (Arendse, para. 6). That is, as another moderator put it: ‘What Meta and Sama are doing to African youth is so heartless … we are being collected from all over Africa to be tortured in slave-like conditions … We are building the world’s largest social media platform, Facebook, and in turn, Meta and Sama make sure we will never live normal lives again’ (Arendse, para. 4(81)).

To work in such an environment must have felt like being trapped in an unending nightmare – a place where every day brought new horrors, and there was no escape, even in sleep. The atmosphere would have been one of constant dread and anxiety, where the moderators were continuously bombarded with the worst of humanity, forced to witness acts of unimaginable cruelty and violence, knowing that each piece of content they reviewed was just one of hundreds they would have to endure that day. The relentless pace, combined with the random and graphic nature of the content, would have created a sense of overwhelming helplessness, as if they were mere cogs in a vast, merciless machine that cared nothing for their well-being.

The sense of being voiceless and ‘treated like garbage’ (Arendse, para. 7) speaks to the dehumanising experience of being reduced to a function, stripped of individuality and forced into a role where their suffering was invisible and their pain disregarded. The comparison to ‘slave-like conditions’ conveys not just the physical and mental toll of the work, but also the deep betrayal they felt – contributing to the success of a global giant like Facebook while being discarded as if their lives held no value. This atmosphere of exploitation and despair, where their every action was monitored and their every mistake could mean the loss of their job, would have suffused their daily existence with a pervasive, crushing sense of fear and futility.

Yet, the situation took a darker turn when one of the content moderators, Daniel Motaung, brought legal claims against Sama and Meta, alleging that Sama had breached his constitutional rights to unionise, and had raised serious concerns about the harsh working conditions faced by the moderators (see LRC Constitutional Petition No. E071 of 2022, Daniel Motaung v Samasource EPZ Ltd t/a Sama & Others). This legal action shed light on the exploitative environment where any attempts to organise or speak out were met with resistance. This resistance manifested as increased surveillance and subtle threats of job loss, and the establishment of a culture of fear, where workers were made to feel that any form of dissent or collective action would lead to immediate and severe repercussions, such as being singled out for harsher treatment, losing their jobs or even facing deportation due to their precarious immigration status (Arendse, paras 3(o), 9–10, 13, 15). When Sama and Meta became aware of Motaung’s lawsuit, they apparently responded by terminating the contracts of the employees en masse – an action that moderators perceived as a retaliatory move designed to stifle dissent and prevent further unionisation efforts (Arendse, paras 8–9, 17(e), 20).

This decision placed the moderators – many of whom were already vulnerable because they were refugees and/or dependent on their employment for legal residency in Kenya – in an extremely precarious position. Losing their jobs did not just mean the loss of income; it meant the loss of their legal right to remain in the country. Without employment, they faced the terrifying prospect of deportation – potentially back to the very regions they had fled due to conflict and instability. As the Court noted:

Thus, what this testimony suggests is that – much like a violent wave (drawing from Baradian aquatic analogies) crashing relentlessly upon a shore – the experiences of the content moderators align closely with the subjection-oriented narrative in capturing the overwhelming sense of disempowerment, trauma and helplessness that defined their work environment. If we liken the experience of the Kenyan content moderators to a wave crashing against a passive shore, the wave – representing the relentless pressures and traumas of their work – surged forwards with unrelenting power, repeatedly striking the shore with devastating impact.

Each crash of the wave chips away at the shore’s integrity, eroding its structure bit by bit. The sand and rocks – once stable and cohesive – begin to fragment and dissolve under the continuous onslaught. The wave does not simply interact with the shore; it overwhelms and consumes it, gradually breaking down its defences until the shoreline is left battered, fragmented and irrevocably altered. The force of the wave leaves no room for the shore to recover or rebuild. Instead, the shore passively absorbs the impact, its form and substance slowly disintegrating into the relentless tide.

Over time, the repeated pounding of the waves wore away the shore’s very essence, stripping it of its original shape and coherence. The shore – much like the moderators – was left with deep scars: its once-solid structure now fractured, its surface marred by the constant erosion. The once-familiar contours of the shoreline were now unrecognisable, replaced by jagged edges and gaping voids where the waves had washed away the land. The shore’s transformation was not one of growth or renewal, but of gradual, inevitable decay – a process driven by the relentless force of the waves that continued to pound away without mercy, leaving behind ‘broken shells’ of what once was whole, ‘slave-like conditions’ and deep psychological trauma. Flashbacks of mutilated bodies, sadistic videos depicting manslaughter and the ‘burning of persons alive’ and images of unimaginable cruelty lingered like debris. The shore – much like the moderators – felt like fragments, cast away by the wave: ‘treated like garbage’, drained physically, mentally and emotionally, their lives irreversibly damaged, left as mere remnants of their former selves.

4 Arendse case: jurisdictional contestations, separate legal personality and the question of employer status

While such a subjection-based account provides one lens to understand the overwhelming systemic power imbalances at play, it is not the only way to interpret and it does not capture the full diffractive possibilities and readings of the Arendse case. Another account – one that accepts but moves beyond this subjection view – would see the relationships between Meta, the content moderators and Kenya not as just interactions (without co-constitution) between distinct, self-contained entities, but as intra-actions where these entities are tied to the relational processes of their becoming and diffractive patterns of interference or difference in their emergent effects. In this lens, the relationship between Meta and the content moderators, as well as Meta’s engagement with Kenya as a whole, is not a simple case of a powerful multinational corporation – like a relentless, forceful wave crashing with immense power, eroding everything in its path – imposing its will on a passive recipient/shore: a shore that absorbs all types of violence and trauma from the wave, gradually wearing away under the unyielding impact, its defences crumbling and its form disintegrating into fragments. Instead, this account acknowledges that these entities are co-constituted through their encounters with each other, with Kenyan law playing a significant, ontological role in this process. Through the constitutive power of Kenyan law – what Butler would call ‘discursive juridical formations’ – the intra-actions between Meta and the content moderators are diffracted: refracted in ways that open up new possibilities for agency, resistance and, simultaneously, social reproduction.

In that, against the structural power and violence they suffered, the Kenyan content moderators filed a constitutional petition against Meta in 2023. They argued that their termination due to redundancy was unlawful; that they faced discrimination and unfair treatment in breach of their constitutional rights – specifically, their rights to fair labour practices (Article 41), dignity (Article 28) and freedom from discrimination (Article 27) under the Constitution of Kenya 2010; and that their work had caused severe harm to their mental health, and that both Meta and Sama were, as joint employers, responsible for the unsafe working conditions, the psychological trauma inflicted upon them and the violation of their rights to fair treatment and mental well-being – specifically, their constitutional right to dignity.

It is important to clarify that the Arendse case actually constituted three interrelated cases. However, the Daniel Motaung case provided the backdrop to the Arendse cases. Daniel Motaung, a former content moderator, had earlier filed a lawsuit against Sama and Meta, alleging unsafe working conditions, violation of labour rights and mental health harm. This case – which gained significant attention – was believed by content moderators to have precipitated the mass redundancies among content moderators, as they suspected these layoffs were retaliatory in nature. In response to the perceived retaliation, the content moderators initiated the first of the Arendse cases, seeking legal recourse.

The first case, which I will refer to as Arendse 1 (Arendse & 182 Others v Meta Platforms, Inc & 3 Others; Kenya Human Rights Commission & 8 Others (Interested Parties) (Constitutional Petition E052 of 2023) [2023] KEELRC 921 (KLR) (20 April 2023) (Ruling)), primarily focused on an application for an interim injunctive relief sought by the content moderators. The moderators aimed to obtain a court order to temporarily halt Meta and Sama from proceeding with the redundancy process until the substantive issues could be resolved. This case was particularly concerned with the legal question of separate legal personality, as the court had to determine whether the injunctive order should be directed at Meta, Sama or both, given the complex corporate structure and the roles of the various entities involved. The court’s decision in this instance did not go deeply into the factual matters but rather concentrated on the immediate procedural question of who bore legal responsibility for the redundancy and thus who should be subject to the interim order.

The second case, which I will refer to as Arendse 2 (Arendse & 42 Others v Meta Platforms, Inc & 3 Others; Kenya Human Rights Commission & 8 Others (Interested Parties) (Constitutional Petition E052 of 2023) [2023] KEELRC 1398 (KLR) (2 June 2023) (Ruling)), served as the primary forum where the substantive legal and factual issues were contested and decided. Unlike the preliminary procedural focus of Arendse 1, this case looked in detail at the core matters at hand, including the legality of the redundancy process, the conditions under which the content moderators worked and the roles and responsibilities of the involved parties, including Meta and Sama. The court examined extensive testimony and affidavits submitted by the content moderators, which provided detailed accounts of their experiences, the psychological impact of their work and the alleged violations of their constitutional and employment rights. These testimonies, which were crucial to understanding the lived realities of the moderators, are the same ones that I referenced in the previous section.

Finally, there was the appeal, which I will refer to as Arendse 3 (Meta Platforms, Inc & another v Samasource Kenya EPZ Limited t/a Sama & 185 Others; Central Organization of Trade Unions Kenya & 8 Others (Interested Parties) (Civil Application E178 of 2023) [2023] KECA 999 (KLR) (28 July 2023)), where Meta appealed the decisions made in Arendse 1 and Arendse 2. This appeal was heard by the Kenyan Court of Appeal, and it primarily focused on jurisdictional questions that had originally been raised in relation to the interim orders issued in Arendse 1. Meta sought to challenge the court’s jurisdiction to issue orders against it, given its status as a foreign corporation. Importantly, the appeal did not engage deeply with the factual disputes of the case, but instead centred on whether the Kenyan courts had the authority to adjudicate and enforce decisions against Meta. However, the appeal was unsuccessful on all accounts, with the Court of Appeal affirming all the findings of law and fact from Arendse 1 and Arendse 2, and thus upholding the jurisdiction and decisions of the lower courts.

With that background set, the Kenyan courts in the Arendse cases had three key legal questions to address. First, Meta challenged the court’s jurisdiction over it as a foreign corporation in Arendse 1 and 3, arguing that, because they did not have a physical office or directly conducted business operations in Kenya, the Kenyan court should not have the authority to hear the case against them. Meta’s argument relied heavily on the doctrine of separate legal personality – especially in Arendse 1, para. 85 – a principle enshrined in the landmark UK case of Salomon v Salomon & Co Ltd [1897] AC 22, which has been incorporated, as a legacy of colonialism, into Kenyan common law (Arendse 2).

The doctrine of separate legal personality holds that a corporation is an independent and autonomous entity, distinct from its shareholders, directors, parent companies, creditors and entities with whom it enters into commercial agreements, particularly in terms of legal liability (Magaisa Reference Magaisa2002; Mohanty and Bhandari Reference Mohanty and Bhandari2011; Watson Reference Watson2019). However, this principle may be set aside, or the ‘corporate veil’ lifted, in certain exceptional circumstances, such as when the corporation is used for fraudulent purposes, to evade legal obligations or when there is clear evidence of the parent company exercising direct control over the subsidiary to the extent that the subsidiary is effectively acting as an agent of the parent company (Ahmad Reference Ahmad2021; Banoo Reference Banoo2018; Magaisa Reference Magaisa2002; Mohanty and Bhandari Reference Mohanty and Bhandari2011). Outside of these exceptions, the corporation is considered to have its own legal obligations and responsibilities, separate from those who own it (shareholders) or run it (directors), as well as from those with whom it conducts business – such as suppliers, customers, contractors and other business partners.

This means that when a multinational corporation like Meta enters into commercial agreements with subsidiaries or third-party vendors, these entities are, under the doctrine, treated as separate and autonomous, each with its own liabilities and legal obligations. Under this doctrine, the parent company is not automatically held responsible for the actions or liabilities of these subsidiaries, third-party vendors or other business partners, as they are all regarded as stand-alone entities in their own right (Benjamin Reference Benjamin2021; Magaisa Reference Magaisa2002).

In practice, multinational corporations headquartered in the Global North have frequently relied on the Salomon doctrine to shield themselves from legal liability for activities carried out in the Global South (Magaisa Reference Magaisa2002). For instance, in industries such as oil and mining, parent companies often establish subsidiaries in the Global South to manage local operations. When serious and often long-term environmental and health issues arise – such as contamination of soil and water sources with toxic substances like heavy metals (including lead, mercury and arsenic) or chemicals used in the extraction process, which can lead to severe health problems such as cancers, neurological disorders, respiratory diseases and reproductive issues – the parent company can argue that the subsidiary, as a separate legal entity, is solely responsible for the damages (Baars Reference Baars2016; Černič Reference Černič2018; Magaisa Reference Magaisa2002). This effectively insulates the parent company from direct liability, meaning the parent company avoids being held legally accountable or financially responsible for the harmful consequences caused by the subsidiary’s operations.

That is because these corporations often structure their financial accounts so that the assets of the subsidiary – such as its local bank accounts, physical equipment, property and any limited cash reserves – are the only ones subject to legal claims (Magaisa Reference Magaisa2002). The parent company might ensure that the subsidiary operates with limited financial resources or assets, meaning that in the event of a lawsuit, only these restricted assets – such as locally held funds, operational machinery or on-site facilities – are at risk, not the broader wealth or resources of the parent company. Moreover, in anticipation of legal claims, these corporations can sometimes shift resources out of subsidiaries, further limiting the assets available for compensation (Magaisa Reference Magaisa2002). This financial structuring further protects the parent company by making it difficult – if not impossible – for affected parties to obtain meaningful compensation, as any claims are restricted to the subsidiary’s constrained assets, which may be insufficient to cover the damages. Consequently, the parent company remains largely untouched despite its overarching control and significant benefit from the subsidiary’s operations (Magaisa Reference Magaisa2002; Mnyongani Reference Mnyongani2016; Mohanty and Bhandari Reference Mohanty and Bhandari2011).

When dealing with third-party vendors – such as suppliers, contractors and service providers that might handle various aspects of production, logistics or service delivery in supply chains, particularly in the Global South – the application of the Salomon doctrine and the concept of separate legal personality can similarly be used by multinational corporations to shield themselves from liability (Baars Reference Baars2016; Sukdeo Reference Sukdeo2022; Villiers Reference Villiers2023; Zumbansen Reference Zumbansen2019). Just as with subsidiaries, third-party vendors are legally separate entities from the multinational corporation that hires them. This means that each entity has its own legal obligations, liabilities and financial responsibilities. If a multinational corporation engages a third-party vendor to perform certain services – such as manufacturing products, handling logistics or managing content moderation – the vendor operates as an independent business, even if it is under contract to the multinational corporation. If the third-party vendor engages in practices that result in harm – such as labour violations or environmental damage – the multinational corporation can argue that it is not directly responsible for these actions. Since the vendor is a separate legal entity, the legal responsibility for any wrongdoing rests with the vendor, not the corporation that contracted them.

Accordingly, in Arendse 1 and 2, Meta argued that Sama, as a third-party vendor and a legally separate entity, was responsible for the operations in Kenya, including the employment and management of the content moderators (Arendse 1, para. 85; Arendse 2, para. 13). Meta contended that because Sama was a separate legal person, Meta should not be held liable for any legal issues arising from Sama’s operations in Kenya. According to the Salomon doctrine, any liabilities or obligations related to the content moderation work conducted by Sama were solely Sama’s responsibility, not Meta’s.

In addition to invoking the doctrine of separate legal personality, Meta further argued in Arendse 1 and 3 that it did not have a physical presence in Kenya, which it claimed should exempt it from the jurisdiction of the Kenyan courts (Arendse 3, paras 6, 12, 21–32). Meta’s argument was based on the legal principle of territorial jurisdiction – the principle that a court’s legal authority typically extends over entities that have a physical presence or conduct business within its geographic boundaries (Druel et al. Reference Druel, Billé and Treyer2011; Ryngaert Reference Ryngaert2015). Meta contended that since it did not have an office, employees or direct business operations in Kenya, it could not be said to have a legal presence in the country (Arendse 1, paras 6, 63; Arendse 3, paras 31–38).

Meta maintained that its involvement in Kenya was limited to a virtual presence through its enterprise resource planning (ERP) system, which it used to manage and oversee the content moderation work carried out by Sama. It argued that this virtual system did not count as a physical presence – because it was intangible and did not involve physical infrastructure or employees within Kenya – and therefore should not subject it to the jurisdiction of the Kenyan courts, particularly in terms of the interim orders of Arendse 1. Instead, Meta emphasised that its relationship with Sama, a Kenyan company, was simply a commercial agreement with a separate, independent entity that happened to operate in Kenya. This contractual relationship, in Meta’s view, did not establish a sufficient territorial link to be considered a direct presence in the country for jurisdictional purposes (Arendse 3, paras 6, 12, 21–32).

In other words, Meta was arguing that, because it did not have a physical presence in Kenya, the Kenyan courts should not have the authority to hear claims brought against it by the content moderators. This would mean that the content moderators, who were raising serious complaints about their working conditions and the psychological harm they suffered while moderating content for Meta, would be unable to hold Meta accountable in Kenya. Instead, they would likely be forced to pursue their claims against Sama, the Kenyan company that directly employed them. This has significant implications because, unlike Meta, Sama may have far more limited financial resources and assets. Consequently, any claims for damages or compensation would be restricted to Sama’s assets, potentially leaving the moderators with inadequate redress for the harm they suffered. Meta’s position, if accepted, would effectively insulate it from legal responsibility within the Kenyan legal system – and thereby place it outside the reach of the Kenyan Constitution and the fundamental rights it protects – despite the work the moderators were doing being directly tied to Meta’s business activities, and would force the moderators to seek redress from a much smaller, less financially robust entity.

The Kenyan courts – particularly in Arendse, paras 1 and 3 – dismissed Meta’s argument and asserted its territorial jurisdiction on the grounds that Meta’s operations had a significant impact within Kenya, and the work performed by the moderators was directly tied to Meta’s business activities (Arendse 1, paras 52–65); Arendse 3, paras 31–35). The court in Arendse 1 and 3 reasoned that, despite Meta’s arguments relying on the doctrine of separate legal personality, the reality of its control over the content moderation process in Kenya could not be ignored. Meta’s significant involvement in setting the work standards, monitoring performance and providing the digital tools and content for moderation meant, in effect, that it was conducting business in Kenya, and thus fell within the scope of the court’s territorial jurisdiction – regardless of the lack of a physical office – meaning that, even though Meta did not have a physical infrastructure such as buildings or staff directly employed in Kenya, its extensive virtual presence and operational control were sufficient to establish jurisdiction.

This reasoning by the court in establishing territorial jurisdiction also became crucial in addressing the second key legal question Kenyan courts had to answer, which was the focus of Arendse 2: whether Meta was a primary or joint employer of the content moderators. The court’s finding that Meta operated in Kenya in a way that exerted significant control over the content moderation activities implied that Meta was not simply a distant contractor with limited involvement. Instead, it suggested that Meta was deeply integrated into the employment activities of the moderators, controlling key aspects of their work, including setting performance standards, monitoring their output and providing the necessary digital tools.

Even though Sama was the entity that formally recruited the content moderators and handled the direct distribution of salaries, the court held that this did not materially change Meta’s relationship or status as a primary or joint employer (Arendse, paras 31–32). This is because the financial resources for the salaries ultimately came from Meta, and it was Meta, through its ERP system and performance monitoring, that, in effect, determined pay levels, bonuses and rewards based on the moderators’ work performance. Meta’s control over these crucial aspects of employment meant that Sama’s role was limited to administrative functions – essentially acting as an intermediary or facilitator for the employment relationship rather than an independent employer with full control over employment terms (Arendse, para. 31). That is to say, in practice, Sama’s role was narrowly confined to carrying out tasks that Meta dictated, such as processing payroll and handling recruitment logistics. The court’s analysis suggested that Meta’s substantial involvement in day-to-day operational decisions, coupled with its financial control, rendered Sama’s involvement largely procedural (Arendse, para. 31).

This narrow role of Sama indicated that the real control and direction of the content moderation work lay with Meta, which, according to the court, was the primary employer and therefore responsible and liable for ensuring safe and fair working conditions, as well as for providing compensation for any damages resulting from wrongdoing. Specifically, this meant that Meta had a legal obligation to ensure that the content moderators were working in an environment that did not expose them to unnecessary harm or risk. As the primary employer, Meta was required to implement adequate measures to protect the moderators from the severe psychological injury that could result from prolonged exposure to highly disturbing and graphic content (Arendse, paras 36–43).

5 Diffractive aftermaths: between juridical resistance and re-constituted harm

Thus, the Arendse litigation constituted a juridical rupture – an aperture through which trauma, invisibility and global asymmetry became momentarily legible. Against the grain of dominant narratives that position the Global South as a passive terrain upon which corporate power acts, the Kenyan courts affirmed jurisdiction over Meta, lifted the veil of corporate separability and recognised joint employment in a transnational labour configuration. In doing so, the court re-constituted the terms of juridical legibility itself: reclassifying digital platform labour as labour under the law, affirming the dignity of content moderation as a constitutional concern and invoking the authority of the Kenyan judiciary to reach across jurisdictional thresholds shaped by corporate architecture.

This juridical gesture gave rise to a series of intra-active effects. Through the constitutive power of law, harm gained form; responsibility became actionable; testimony entered the official archive. Violence was no longer atmospheric or ambient – it became adjudicable. Moderators’ accounts of trauma, PTSD, displacement and coercion were no longer held at the periphery of recognition. They appeared in court records, rulings and affidavits, shaping the material-discursive apparatus through which legal subjectivity and digital labour coalesced. Through this process, the Kenyan legal system intra-acted with global capital, digital platforms and human vulnerability to generate new forms of recognition, institutional authority and legal precedent. In this sense, Arendse is indicative of the emancipatory dimension of what Butler identifies as juridical discursive formations (Butler Reference Butler1990): the power of law not merely to name or regulate, but to bring new subjects, injuries and relational configurations into being.

This recognition, however, did not resolve the asymmetries it disclosed. The very processes through which harm became legible also initiated new exclusions, silences and vulnerabilities. Moderators who participated in the case were all made, in the end, redundant. For those whose immigration status in Kenya was contingent on employment, the termination precipitated immediate legal and existential risk (Kimeu Reference Kimeu2024). Without alternative employment and with their right to remain in the country under threat, they faced potential deportation. Many suffered the compounding effects of trauma and precarity after their redundancy: denied access to psychiatric care, isolated from their support networks and subject to social abandonment. Some experienced family breakdowns; others reported an inability to maintain intimate relationships due to trauma-induced desensitisation or hypersexuality, including distorted sexual boundaries (Kimeu Reference Kimeu2024). Several were blacklisted from re-employment in the sector after being refused jobs without explanation (Kimeu Reference Kimeu2024). One moderator developed trypophobia (an intense aversion to clustered holes or patterns, often linked to visual triggers like decaying skin or bodily trauma) after repeatedly viewing videos of decomposing bodies and mutilation, and reported hyperventilating and crying uncontrollably since losing their employment (Kimeu Reference Kimeu2024). A mother in her twenties, diagnosed with severe PTSD, recalled biting her own arm and smashing bricks against the wall to self-regulate (Kimeu Reference Kimeu2024). Others reported insomnia, intrusive flashbacks and psychological fragmentation that rendered daily life unmanageable. Religious moderators described deep shame after repeated exposure to sexual violence, while younger workers spoke of being permanently altered by material that constituted their first exposure to sexuality. Moderators assigned to conflict-related content faced threats from armed actors, and in at least one instance, required relocation to a safe house (Kimeu Reference Kimeu2024). Financial redress remained elusive, as settlement talks collapsed – leaving many without compensation or closure. Despite judicial findings affirming responsibility, the material conditions of harm have, thus, persisted, extended and reconfigured.

In response to the public and legal visibility of these events, the Kenyan government introduced a legislative amendment – Section 5A of the Employment Act (2007) – through the Business Laws (Amendment) Act 2024. This section introduced two key provisions. First, it imposed a statutory duty on business process outsourcing (BPO) companies and Information technology enabled services (ITES) providers to ensure that employees, whether working onsite or remotely, are provided with all tools, equipment and support necessary for the performance of their duties – even where the actual beneficiary of the services is a third party. Second, it clarified that the legal responsibility for employment-related claims rests entirely with the outsourcing firm, regardless of whether it is the direct or indirect beneficiary of the labour. In doing so, the law formalised the visibility of digital labour as legitimate employment, requiring minimum standards of provision and care.

This same provision, however, instituted a protective boundary around multinational clients – such as Meta – by insulating them from legal claims that might otherwise have been brought against them by outsourced workers. While the law affirmed the legal status of digital workers as employees and required workplace accountability, it simultaneously redirected liability inwards, towards the outsourcing intermediary, and away from the transnational corporate beneficiary. This legal configuration represents a paradigmatic diffraction: a re-patterning of labour law that includes new recognitions and produces new occlusions.

Through a diffractive reading, the amendment does not register as an unequivocal advance. Rather, it constitutes a site where recognition and disavowal are co-enacted. The law affirms the status of the digital worker while delimiting the reach of responsibility. It reflects the materialisation of trauma as legally cognisable while generating conditions that allow such harm to be redistributed without corporate accountability. Acknowledgement and exclusion travel together, emerging as part of the same juridical process. The diffraction pattern becomes visible in the aftermath: where brighter bands of recognition appear alongside shadows of abandonment, and where Kenya’s juridical system – a situated cut within the entangled material-discursive field of the Global South – is affirmed as capable of resistance while simultaneously reabsorbed into global circuits of platform governance.

This entangled aftermath requires a reading that does not reduce the court’s intervention to resistance or capitulation. It invites analysis that foregrounds emergence, remainder and the co-production of harm and responsibility. Recognition here is not closure – it is a trembling threshold across which new patterns of digital labour, corporate insulation and legislative accommodation are enacted. As Barad reminds us, diffraction is not a reflection of difference; it is a generative apparatus through which difference takes shape. And in this case, that shape includes the simultaneous emergence of legal agency, structural deferral and recursive violence – co-constituted in the same cut.

6 Conclusion: law’s constitutive cuts and the unsettling of emancipatory narratives

What, then, does Arendse suggest? At first instance, the case might appear to crystallise the ambitions of a mode or discourse of transformative constitutionalism – what Butler refers to as the emancipatory potential of juridical formations to, through law’s (re)productive and constitutive power, reconfigure conditions of subjection and materialise new possibilities for agency and recognition (Butler Reference Butler1990). That is, one where a marginalised and invisibilised class of transnational workers turns to the Constitution of Kenya to challenge platform precarity; a court in the Global South asserts jurisdiction over a powerful foreign platform; and new juridical subjectivities emerge grounded in constitutional rights to dignity, non-discrimination and fair labour practices. This reading, prevalent in media discourse, positions Arendse as a redemptive legal moment – a rupture through which law finally speaks to global asymmetry. Or, read through the lens of subaltern resistance, such a moment might be seen as a counter-hegemonic legal intervention – an insurgent jurisprudence (Rajagopal), a subaltern voice momentarily heard (Spivak), a vernacularisation of justice (de Sousa Santos) or a tactical deployment of legality from below (Scott) (de Sousa Santos Reference de Sousa Santos2015; Rajagopal Reference Rajagopal2003; Scott Reference Scott1985; Spivak Reference Spivak, Cain and Harrison2023). On this account, Arendse is rendered a moment of juridical defiance in which the margins speak through law to global power.

Yet, if the analysis offered here holds, that framing becomes difficult to sustain. Transformative constitutionalism – read now in both senses: as a jurisprudential project and as a theory of law’s constitutive force – presumes a temporality of progression. It stages an origin in structural injustice, followed by an intervening act of juridical recognition, and oriented towards a telos of transformation. Constitutionalism, in this double register, is both a legal doctrine and an ontological claim: it does not only respond to injustice; it configures the very subjects through which justice is imagined, claimed and contested. The figure of the precarious worker, the injured plaintiff or the claimant in the Global South does not pre-exist the legal scene; it is constituted through it. What appears as recognition is already patterned by the conditions through which legibility is granted. Within this arc, law becomes the scene of both injury and its proposed redress – a site where subjection is named, only through forms that render it actionable. The legal subject, though wounded or dispossessed, is made visible through rights discourse, and through that visibility, included, acknowledged and disciplined.

We can arrive here because a Butlerian–Baradian reading shifts the analytic from whether law resisted or transformed, to how difference was materially enacted – how law functioned, even momentarily, as a site of ontological reconfiguration. The question is no longer whether the Kenyan legal system achieved a transformation/emancipation, but how it intra-acted with platform capitalism, outsourced trauma and the epistemologies of digital labour to produce patterns of legibility, responsibility and exclusion. When we do that, we see that Arendse generated a diffraction pattern: it reclassified outsourced workers as employees, pierced the veil of separate legal personality and refused the jurisdictional insulation sought by Meta. In doing so, it rendered visible the violence of digital extraction and reframed the constitutional subject beyond the formal categories inherited from colonial private law. Yet, this same juridical cut gave rise to new modes of exclusion. Recognition materialised without redress. Visibility emerged alongside dismissal. Dignity was invoked even as protection was withdrawn. Workers lost jobs, legal residency, access to health care and avenues for future employment. The trauma that had once been made visible was displaced, deferred and redistributed.

If there was a transformation here, it was neither linear nor totalising. It was relational, recursive and ambivalent – produced through intra-actions that both disrupted and re-entrenched asymmetry. To invoke transformative constitutionalism as the frame is to risk reimposing the very closure that the case resists. It is to seek resolution where only relational remainder exists. It is to narrate progress in a space of ontological instability.

The point is not to reject the emancipatory potential of law to intervene and engage (re)world making. Rather, it is to remain with the complexity. Arendse invites the reader to resist the desire for narrative resolution and instead stay with the diffractive aftermath: the shimmering interference of constitutional recognition and material harm, of ontological emergence and juridical deferral. This is not a story of redemption. It is an account of patterned entanglement – where law, labour, trauma and capital collide without finality. If constitutionalism or emancipatory discourses function here, it does so not as a promise fulfilled, but as an apparatus of becoming. It does not transform in the way doctrine imagines. It diffracts. And in that diffraction, it opens – and simultaneously forecloses – possibility.