Processual epistemic burdens and the struggles for recognition: Epistemic injustice in the platform economy

Abstract

This article asks how contemporary labor domination is sustained through the organization of knowledge, evidence, and procedure. Building on work in epistemic injustice, it introduces the concept of “processual epistemic burdens,” meaning the cumulative mental exhaustion and iterative cognitive labor imposed on disadvantaged actors who must repeatedly interpret events, assemble evidence, translate their experience in ways that are legible to relevant institutions and authorities, and defend their claims over time. Much of the epistemic injustice literature is oriented to discrete and episodic harms, yet insufficiently accounts for the protracted process by which the oppressed must struggle for recognition. Centering the platform gig economy as an archetypal case, this article shows how algorithmic opacity, fragmented accountability, and adversarial litigation compel workers to become perpetual investigators, translators, and litigants. Reconceptualizing domination as epistemically mediated, the article shows that reducing epistemic injustice in the digital economy requires a more equitable distribution of epistemic labor.

Introduction

Since Fricker's (2007) seminal articulation of testimonial and hermeneutical injustice, the literature has broadened rapidly, extending the concept's application to law, medicine, digital systems, and other institutional settings, and developing a family of adjacent concepts (Kidd et al., 2017), such as epistemic oppression (Dotson, 2014), testimonial smothering (Dotson, 2011), willful hermeneutical ignorance (Pohlhaus, 2012), epistemic exploitation (Berenstain, 2016), that track how power shapes who is heard, who is understood, and who must do the work of making injustice legible. However, much of this literature focuses on specific and discrete epistemic harms resulting from diverse epistemic practices. This article highlights and articulates an often-neglected dimension of epistemic wrong: the epistemic labor required to overcome epistemic injustice. Many paradigmatic cases of epistemic injustice do not end at the moment of a credibility deficit or interpretive gap. Instead, they initiate a prolonged and repetitive struggle in which the wronged subject must decipher opaque institutions, translate lived experience into particular framings, anticipate skeptical audiences, and repeatedly press for attention. The central question is whether these prolonged epistemic burdens are merely unfortunate effects of an epistemic injustice, or whether, in at least some cases, they constitute an additional form of epistemic injustice. This article argues that they do. When the burdens are structurally imposed in a way that systematically offloads onto marginalized agents the work required to secure basic epistemic standing, including being seen as intelligible, being taken seriously, and being given reasons and avenues of contestation. In such cases, what is wrong is not only the initial exclusion from epistemic participation, but also the institutionalized requirement that the excluded themselves must do the exhausting work of repairing that exclusion. To capture this phenomenon, this article introduces the concept of “processual epistemic burdens,” defined here as the sustained mental exhaustion and cognitive labor involved in articulating, interpreting, and communicating one's lived experience under structurally unjust conditions. The concept captures the cumulative, laborious, and time-intensive process by which marginalized individuals must make themselves understood to powerful audiences or institutions.

To illustrate the substance and utility of the concept, this article draws on platform workers’ struggles as the central case of analysis. In the context of gig work, these processual epistemic burdens include the ongoing efforts of platform workers to make sense of their exploitation and render it legible, to themselves, to the public, and to legal or public authorities, in order to claim their rights. As will be shown, platform workers must engage in precisely this kind of repeated cognitive labor, often against indifferent or obstructive institutions. Importantly, the processual lens stresses that these harms unfold over time and require iterative and Herculean efforts on the part of the marginalized, rather than being confined to isolated moments such as a single act of silencing or a one-time credibility deficit.

Two clarifications concerning the article's contributions. First, the concept of “processual epistemic burdens” is offered as a general contribution to epistemic injustice theory. It identifies a distinctive locus of wrong: the temporally extended distribution of epistemic labor required to be heard, understood, and redressed. The point is to extend epistemic injustice to account for not only who is discredited due to identity prejudices and disadvantaged due to hermeneutical gaps, but also who must then do the ongoing work of repairing those epistemic conditions. This is theoretically significant because it reframes questions of responsibility and domination in terms of how epistemic labor is allocated across time and social position. Second, the discussion of the platform economy in the article is more than a narrow “application”; it is a diagnostic amplifier. Platform firms govern through algorithmic opacity, fragmented accountability, and legal contestation, and in doing so they bring into view a temporal arc that more familiar cases tend to push into the background. This includes the repeated cycle of sense-making, evidencing, re-framing, and re-presenting that marginalized agents must go through simply to be taken seriously. Gig work shows these burdens in a clearer and concentrated way, even though they are hardly unique to it. The article can therefore be read in two ways: (1) as a general theoretical contribution to epistemic injustice literature, illustrated by a vivid case, and (2) as an applied contribution to gig-work scholarship that highlights an epistemic critique valuable in its own right (see, for example: Kwok, 2021, 2025a; Vallas and Schor, 2020), even for readers less convinced by the distinctiveness of the processual dimension in the general epistemic injustice literature. Even on the latter, narrower reading, tracing how misclassification and algorithmic management shift evidential and interpretive labor onto workers helps explain why grievances struggle to become concrete claims, why litigation becomes a war of attrition, and why opacity functions as a mode of labor domination in the digital economy.

The article is structured as follows. It first argues that the epistemic injustice literature largely treats epistemic harms as episodic and thus misses the long-term processual epistemic labor required to make one's experience intelligible and credible. It then develops the concept of processual epistemic burdens and defends it as a distinct locus of epistemic injustice. Finally, it uses dependent pseudo-gig work as a central case to illustrate the concept, tracing how epistemic burdens accumulate through algorithmic opacity, obfuscated accountability, and legal draining. Finally, it reflects on potential questions and challenges to the central arguments offered in the article.

Epistemic injustice and the missing process dimension

Fricker's Epistemic Injustice (2007) provides an important theoretical lens on how power and prejudice can distort everyday epistemic interactions. Fricker identified two primary forms of epistemic injustice: testimonial injustice, in which a speaker is given deflated credibility due to identity prejudice, and hermeneutical injustice, wherein gaps in collective interpretive resources leave someone's experiences misunderstood or inexpressible. Fricker's (2007: 149–151) often-cited example of hermeneutical injustice is the case of Carmita Wood, a woman in the 1970s who struggled to articulate the experience of sexual harassment before the term “sexual harassment” was available in public discourse. Wood felt something was wrong with how she was treated by her boss, but in an era when such mistreatment was dismissed as harmless “flirting,” neither she nor others had the language or understanding to properly identify the injustice (Fricker, 2007: 153). Fricker notes that in such situations, having the feeling that something was wrong does not necessarily imply the ability to make sense of and understand the experience of oppression. Both the victim and the perpetrator can be, in Fricker's words, “cognitively handicapped by the hermeneutical lacuna – neither has a proper understanding” of the wrong (Fricker, 2007: 151).

Fricker's focus on credibility deficits and interpretive lacunae is often read as centering discrete epistemic wrongs occurring in identifiable exchanges. Yet one influential line of development in the literature pushes beyond this “transactional” framing. Elizabeth Anderson (2012) argues that epistemic justice should be theorized not only as an individual virtue but also as “a virtue of social institutions.” An institutional and systemic perspective matters here because it can account for “the cumulative effects of millions of individual transactions” in testimonial and hermeneutical domains, effects that cannot be handled at the individual level alone (Anderson, 2012: 163). In particular, she stresses that the “cumulative effects” of how an epistemic system elicits and evaluates communicative acts “can be unjust, even if no injustice has been committed in any particular epistemic transaction” (Anderson, 2012: 165). This point is directly relevant to the process dimension stressed earlier: what demands explanation is more than isolated credibility failures; it is also the broader institutional arrangements that predictably generate cumulative epistemic disadvantages and require compensatory labor from some participants rather than others. Medina's (2011: 17) contextualist approach further highlights such temporal dimension more explicitly. He contends that epistemic injustice is both “temporally and socially extended,” suggesting that accounts of epistemic injustice should extend beyond individual moments of testimonial exchange between particular subjects. For Medina (2011: 16), epistemic harms “tend to have temporal trajectories and to reverberate across a multiplicity of contexts and social interactions.” A full account of any given testimonial exchange must therefore attend to the conditions that precede it and the consequences that follow from it, and situating it within broader patterns of social relations. This calls for a socio-historical analysis that can trace how contributions to injustice develop in and across concrete contexts, instead of treating each credibility assessment in isolation.

Against such background, while Fricker insightfully illuminates the existence of these epistemic harms, her framework gives less attention to the active and ongoing efforts that individuals like Wood must undertake to overcome such harms. Hermeneutical injustice is described as a structural gap in collective understanding, but implicit in Wood's story is a laborious process of sense-making and meaning-creation that unfolded over time. In Wood's case, it was through consciousness-raising and feminist advocacy that the term “sexual harassment” was eventually coined, filling the interpretive gap (Fricker, 2007: 150). Fricker's story centers on the injustice of the gap itself, instead of the epistemic labor required of the excluded to fill that gap. Similarly, in cases of testimonial injustice, the emphasis is on the wrong of being discredited in the moment, and yet we learn less about what happens after, for instance, the repeated efforts to establish credibility that a discredited speaker might have to engage in, or the exhaustion that accumulates from constantly having one's testimonies questioned. Thus, while Fricker's framework identifies important discrete epistemic injustices, it does not explicitly theorize the processual burdens placed on marginalized knowers to remedy and cope with those injustices over time.

Subsequent scholars have expanded the repertoire of epistemic injustice concepts. However, a similar issue remains. Dotson (2014: 115) theorizes the term “epistemic oppression,” defined as “persistent epistemic exclusion that hinders one's contribution to knowledge production.” Dotson's definition highlights the persistence dimension of epistemic injustice by demonstrating that epistemic exclusion is not a one-off event, but a systematic and enduring pattern. She articulates multiple “orders” of epistemic oppression, including deep structural conditions that block marginalized people from participating in meaning-making on their own terms (Dotson, 2014). This moves us toward a structural and long-duration view of epistemic injustice. However, Dotson's focus is still principally on the exclusionary structures themselves, the ways marginalized knowers are blocked, rather than on the compensatory epistemic labor required to resist these exclusions. Notably, in a later work, Dotson (2018: 130–131) further explores how certain seemingly “difficult-to-defeat” justificatory arguments can amass epistemic power that sustains “resilient oblivion,” which normalizes oppressive conditions by making them seem reasonable. This insight shows that ignorance itself can be an active and cumulative force; yet even here, the emphasis is on the entrenched nature of ignorance than the continual labor borne by marginalized actors to dismantle it.

Dotson's earlier work on testimonial quieting and testimonial smothering (2011) shows how marginalized people often adapt their speech in the face of hostile and apathetic audiences. Testimonial smothering refers to “the truncating of one's own testimony in order to ensure that the testimony contains only content for which one's audience demonstrates testimonial competence” (Dotson, 2011: 244). In other words, marginalized speakers may preemptively withhold nuance about their experience because they anticipate misunderstanding and prejudice. This suggests a defensive tactic employed under oppression, essentially a form of self-silencing to avoid epistemic exploitation or harm. Yet, while Dotson here acknowledges the effort marginalized people utilize in calibrating and censoring their testimony, the emphasis is on the negative outcome (silence) instead of the processual burden on those who do attempt to speak fully. Testimonial smothering is a strategy to conserve one's energy and safety in a hostile epistemic environment, which implicitly points to the high cost of trying to convey one's experience when conditions are unfavorable. Dotson's account, however, stops at describing the strategy and its rationale; it does not delve into the prolonged struggle that might ensue if, for example, the person chose not to smother their testimony and instead attempted again and again to get a fair hearing. Thus, Dotson's contributions highlight structural and anticipatory dimensions of epistemic injustice, but they still leave under-theorized the longitudinal grind of epistemic labor.

Similarly, Gaile Pohlhau's (2012) notion of willful hermeneutical ignorance shows that powerfully situated knowers, typically dominant group members, often actively resist understanding the meanings and perspectives of the marginalized, even when they have access to epistemic resources that could enable understanding. This is a form of willful ignorance in the hermeneutical domain in that it enables dominant agents to continue misinterpreting and ignoring significant portions of the shared social world, even though the relevant interpretive tools are available in principle and could be acquired through genuinely cooperative epistemic relations. Pohlhaus's critique is aimed at the agency of the powerful in maintaining epistemic injustices. She illuminates how ignorance is not always a passive gap and can be an active project of ignoring. Pohlhaus's framework thus implicitly paints a picture of marginalized knowers who could repeatedly offer understanding only to have it refused by the powerful. Willful hermeneutical ignorance is theorized primarily as a vice of the powerful, and the concept's explanatory focus falls on the mechanisms of that refusal and the question of accountability. That said, Pohlhaus does register the epistemic work of marginalized knowers, and yet she does not make the burdens of that work her central concern. Put differently, the concept tells us why the powerful fail to understand, but it does not directly name the toll on those who, despite this resistance, continue trying to communicate truths about their lives. Rebecca Mason (2011) makes a related point by showing that a gap in dominant hermeneutical resources results in two kinds of “unknowing”: the marginalized group's lack of collective understanding, and the dominant group's epistemically blameworthy ignorance of marginalized experiences. This distinction suggests that dominant ignorance could be actively maintained even as marginalized people develop their own understandings, indicating a process whereby ignorance persists even when the marginalized did attempt to make the unknown known. Like Pohlhaus, Mason highlights the role of the powerful in perpetuating epistemic injustice, which again hints at the ongoing nature of the struggle for the marginalized, that is, those who confront not only a gap in understanding but an opponent with little epistemic interest, or even actively refusing, to understand.

Medina's (2012) also develops a complementary perspective, arguing that hermeneutical injustice is not well captured by treating “the” collective interpretive resource as a single coherent whole. A “polyphonic contextualism,” on his account, is needed to track how different groups may possess partially divergent resources and how marginalization often operates through failures of reciprocity across publics, more than just the absence of concepts. Importantly, for my argument, Medina (2012: 216-217) also disputes the idea that hermeneutical injustice is “purely structural” in a way that insulates interlocutors and institutions from responsibility. He argues that obligations concerning hermeneutical justice are interactive and relational: whether agents live up to “shared hermeneutical responsibilities” depends on patterns of responsiveness and mutual positionality. Medina (2012: 218) explicitly notes that hermeneutical gaps require “collective and sustained efforts across temporally and socially extended contexts,” and that participants in impoverished communicative patterns retain “some limited agency to accentuate the gaps or to contribute to their erosion,” making them potentially co-perpetrators. The implication is that hermeneutical marginalization is maintained through repeated and ordinary failures of interpretive responsiveness, which makes it plausible to treat the burdens of trying again and again to be understood as a predictable and socially produced cost of asymmetric epistemic relations.

Finally, some scholars have begun to identify the exploitative dimensions of epistemic labor. Nora Berenstain (2016) coined the term “epistemic exploitation” to describe situations where privileged persons demand that marginalized persons educate them about the nature of the oppression they experience. She defines epistemic exploitation as an injustice that “occurs when privileged persons compel marginalized persons to produce an education or explanation about the nature of the oppression they face” (Berenstain, 2016: 570). In such cases, the oppressed are compelled to carry the burden of “emotional and cognitive labor” to provide “information, resources, and evidence of oppression to privileged persons who demand it” (Berenstain, 2016: 570). This concept is directly concerned with cognitive burdens: it names the emotional and intellectual work that marginalized people are frequently pressed to do, often without recognition, in order to enlighten the privileged. Epistemic exploitation highlights that exploitation can also occur through the appropriation of epistemic resources and labor, which, essentially, amounts to forcing oppressed individuals to be unpaid educators about their own oppression. This idea is highly relevant to my argument because it suggests the labor and exhaustion that can come with trying to convey one's experience to an uninformed, or willfully ignorant, audience.

Beyond these influential frameworks, more recent feminist epistemologists have explicitly attended to the cumulative and processual nature of epistemic harms. For instance, Berenstain's later work critiques what she calls “white feminist gaslighting,” illustrating how dominant methodological approaches in even feminist spaces can repeatedly invalidate and dismiss the epistemic contributions of women of color, which is a pattern of structural epistemic harm that unfolds over time (Berenstain, 2020). Moreover, MacKenzie (2022: 789) argues that there is “little appreciation” of how routine testimonial injustice can profoundly affect speakers’ wellbeing and ramify with other injustices. Her analysis of the Belfast “Rugby Rape Trial” is instructive as a case where epistemic injustice is not exhausted by a single credibility deficit but is sustained through a long adversarial process, as the trial ran nine weeks, and the complainant was examined for eight days by multiple defense barristers (MacKenzie, 2022). The process, she notes, commonly searches for testimonial inconsistencies and uses constrained formats that “disallow nuance,” while drawing on social myths to discredit the complainant (MacKenzie, 2022: 788). This is a clear example of how institutional procedures can be arranged so that marginalized parties must repeatedly defend intelligibility and credibility under adversarial conditions. It thereby supports the claim that a processual lens is theoretically important: it reveals how epistemic injustice can be reproduced by structured sequences of interaction, evidence-production, and credibility testing.

Ahmed's Complaint! (2021) offers another valuable institutional account explaining how “justifying one's case” is itself burdensome epistemic work: complaint is a temporally extended sequence of tasks that institutions make hard to complete. As she further points out: “making a complaint is never completed by a single action: it often requires you do more and more work,” and "[i]t is exhausting, especially given that what you complain about is already exhausting” (Ahmed, 2021: 5). And because complainants anticipate dismissal, they must make complaints “legible” to relevant institutions and “keep making the same complaints in different ways before they will be heard or in order for them to be heard” (Ahmed, 2021: 6). This captures the processual burdens I emphasize. Epistemic labor stretched over time, multiplied by institutional “stoppages and blockages” (Ahmed, 2021: 27), and shifted from remedy to managing the process itself.

Put more succinctly, these perspectives reinforce the central intuition of the article: epistemic injustice is often enacted through temporally extended patterns. Yet even when these accounts move beyond an episodic frame, they rarely name as such the further burden that follows from those mechanisms, that is, the requirement that the oppressed themselves sustain their participation through an iterative struggle for epistemic recognition. Moreover, what remains comparatively under-articulated is a concept that centers the allocation of epistemic repair work as a distinctive site of epistemic injustice: the sustained effort of making one's experience legible and credible, through diagnosis, translation, and evidential work, within institutional settings that systematically impede recognition.

To fill this theoretical gap, this article proposes the concept of processual epistemic burdens. These burdens are “processual” in that they unfold and accumulate through iterative stages: diagnosing one's situation, translating private pain into publicly intelligible terms, repeatedly telling one's story and presenting evidence, and navigating the often lengthy routes toward recognition or remedy. A reasonable question at this point is that these processual burdens might be better understood as instances of structural injustice more generally, rather than as epistemic injustice in Fricker's (2007) sense, since what is harmed may be the subject's standing as a member of a marginalized group rather than her standing specifically as a knower. Byskov's (2021) account of what makes epistemic injustice an injustice helps address this concern. On his view, epistemic injustice emerges when multiple partial conditions are violated in conjunction, including not only disadvantage and prejudice (Fricker's core conditions), but also a stakeholder condition (those excluded are affected by the outcome), an epistemic condition (they possess relevant knowledge), and a social-justice condition (the pattern is embedded in broader structures of marginalization) (Byskov, 2021: 118). Understood this way, processual epistemic burdens constitute a wrong in a distinctively epistemic sense when they function as an institutionalized epistemic toll: marginalized stakeholders who possess situated knowledge are required to perform disproportionate interpretive and evidential labor merely to have that knowledge count as intelligible, credible, and action-guiding within shared epistemic practices. The wrong is therefore not only that agents are harmed as members of a marginalized group, but that their standing as knowers is systematically constrained. They are placed under asymmetric demands of translation, proof, and repeated justification that more advantaged actors are not required to meet, which predictably undermines epistemic agency and participation over time (see also Ahmed, 2021). This also explains why the burdens are often aggravated by power. When dominant actors or institutional designs introduce opacity, procedural hurdles, and fragmented responsibility, they are reshaping the epistemic environment so that recognition is obtainable only through prolonged and unequal cognitive and expressive labor. In a more just epistemic order, institutions would distribute these interpretive and evidential tasks in a fairer way, instead of simply offloading them onto those who are already vulnerable and least able to bear the cumulative costs.

In what follows, I apply the concept of processual epistemic burdens to a concrete empirical context: the struggles of platform gig workers. By examining this case, I demonstrate how such theoretical lens captures dynamics that existing frameworks might miss and under-emphasize. I show that gig workers, especially those who are misclassified as independent contractors, what I term pseudo-gig workers, face layered and compounding epistemic burdens in their efforts to claim labor rights and justice.

Pseudo-gig work and epistemic burdens

By “pseudo-gig work,” I refer to platform-based labor in which workers perform jobs under conditions functionally similar to regular employment but are deliberately misclassified as independent contractors. This practice is widespread in ride-hailing, delivery, and other app-mediated services: firms exert employer-like control but deny workers the accompanying rights and benefits by exploiting legal grey zones (Bieber and Moggia, 2021; Kwok, 2025b). Pseudo-gig arrangements impose extraordinary interpretive and communicative burdens on workers. By design, the platform model blurs lines of responsibility and obscures how decisions are made, which in turn forces workers into a continuous struggle to discern, document, and communicate the injustices they experience. Worse still, platform companies profit from such epistemic complexity. The fact that it is hard for outsiders, and even for workers themselves, to clearly see or understand how exploitation is occurring allows these firms to evade accountability. Therefore, gig workers who sense they are being wronged must undertake extensive epistemic labor to expose and explain that wrong.

In the dominant platform model, opacity and complexity are core design features. Pseudo-gig work individualizes struggles in that each worker is ostensibly an independent business, so when something goes wrong, they stand alone against a distributive algorithmic system. Unlike some other forms of informal or precarious labor, where a worker might at least know who their direct employer is or visibly see the hazards of the job, gig work obscures these basic facts. In most places, there are usually no unions and formal grievance processes internally. The digital and algorithmic character of gig work introduces a distinctly contemporary dimension to epistemic injustice. Unlike factory labor, where exploitation was at least visible, such as long hours and dangerous conditions, much of what gig workers endure is buried in algorithms and layers of contracting. This makes their oppression unusually difficult to name: workers must, in effect, become students of algorithms and corporate law simply to articulate what is being done to them. New technology and business models, in other words, generate new forms of epistemic burden.

The article now turns to three major stages of processual epistemic burden in gig work: (1) the opacity of algorithmic management, (2) the complication of accountability through structural design, and (3) the draining process of legal challenges. Each of these represents a phase of gig workers’ long struggle for recognition, and together they illustrate how platform-specific features amplify the epistemic labor required of workers. In articulating these burdens, I aim to also demonstrate that they are cumulative. A gig worker seeking justice may start with confusion about why they were treated in a certain way (algorithmic opacity), grow frustrated trying to get answers or reach a responsible person (accountability obfuscation), and end up in a convoluted legal battle (legal and epistemic draining). At each stage, new epistemic tasks are added without the prior ones being fully resolved. By the time they are in court, they may still not understand how the algorithm works, making evidence-gathering difficult, and may remain unclear about the corporate structure behind the platform. Yet they must now acquire legal knowledge on top of all these. Each phase demands a different kind of epistemic labor: interpretive labor when facing opacity, communicative and bureaucratic labor when seeking accountability, and formal analytic labor when navigating the legal system. Few individual workers have the capacity or support to sustain all these at once. This is why many “successful” plaintiffs are those backed by unions or legal aid organizations, suggesting that collective support and third-party intervention is usually necessary to overcome these burdens. This points to a further epistemic implication of gig work's isolated and individualized structure. By keeping workers separate from one another, it deprives them of some of the collective resources that might otherwise ease their epistemic burdens.

It is also important to briefly note the method of selecting the empirical materials. This is a normative political theory article, and I did not conduct original interviews, surveys, or ethnographic fieldwork. The “cases” in the platform economy sections are drawn from publicly available materials used to illustrate the processual mechanisms theorized here. I constructed a corpus of sources via mainly two stages. First, I identified public and high-profile legal disputes and regulatory proceedings on platform work through publicly available legal databases. Second, I gathered public materials that document workers’ own experiences, including worker testimony and statements where available, as well as union and advocacy reports and investigative journalism that records workers’ accounts. Throughout, I draw on cases from a range of jurisdictions rather than confining the analysis to a single national setting, with examples from Europe, United States, South Africa, and Asia, among others. The cases were selected for analytical fit: each is publicly documented and directly illustrates one or more of the mechanisms discussed.

Algorithmic opacity and hermeneutical burdens

One of the most important epistemic hurdles for platform workers is the opacity of the algorithms that govern their work. Gig platforms rely heavily on algorithmic systems to allocate tasks, set prices, evaluate performance, and even discipline workers. An algorithm is generally “a computational formula that allows autonomous decision-making based on statistical models or decision rules, without an individual interfering in the decision-making process” (Felix et al., 2023: 2). In the platform context, algorithms perform what amount to managerial functions: they dispatch rides or deliveries, determine dynamic pay rates, monitor workers via ratings and GPS data, and even initiate suspensions or terminations (Kellogg et al., 2020; Rosenblat and Stark, 2016). Crucially, these algorithms are proprietary and operate as black boxes as their rules and decision criteria are hidden from workers, and often from regulators, under the guise of commercial secrecy (Burrell, 2016; Rahman, 2021). As a result, gig workers are systematically deprived of the information and interpretive resources needed to understand why things happen to them on the job.

This creates a form of hermeneutical injustice by design. Whereas Fricker's (2007) hermeneutical injustice involves a gap in social knowledge that arises unintentionally, in the platform economy the knowledge gap is intentionally engineered by the firm. The lack of transparency is a deliberate business strategy: algorithmic opacity ensures workers lack the information needed to confirm their suspicions about patterns like declining pay or unfair treatment (Jarrahi et al., 2021; Kellogg et al., 2020; Rahman, 2021). For workers, the epistemic burden of this opacity is immense. First, it undermines their basic sense-making capacity regarding their working conditions. Gig workers usually experience sudden changes, such as a drop in earnings, a reduction in ride requests, or an unexplained deactivation warning, but they are given no clear explanation for these events (Rosenblat and Stark, 2016; Vasudevan and Chan, 2022). They may suspect, for example, that an algorithm has altered the payment formula or that a hidden penalty was applied to their account, yet because the system is opaque, these suspicions remain unconfirmed and incommunicable. Workers are left “cognitively handicapped” in Fricker's (2007: 151) sense: they know something is wrong but cannot fully articulate or prove it, because of lacking access to the relevant data or rules.

Indeed, platform firms actively promote a narrative that algorithmic decisions are neutral and error-free. For example, an Uber driver was told by Uber support when he sought an explanation for a false fraud accusation, “[t]he system can’t be wrong, what have you done?” (Murgia, 2024: 114). This techno-positivist stance preemptively casts doubt on workers’ testimony. Thus, opacity coupled with the myth of algorithmic neutrality produces a twofold epistemic injury: workers lack knowledge of the algorithm, and their credibility is undermined when they attempt to complain. In terms of epistemic injustice, this is a blend of hermeneutical injustice (i.e., no shared resources to interpret their experience) and testimonial injustice (i.e., workers’ reports are discredited because the machine is deemed more objective and trustworthy). This resonates with critiques from feminist science and technology studies. As D’Ignazio and Klein (2020) argue in Data Feminism, ostensibly objective data-driven systems often reflect and reinforce existing power inequalities. A lack of transparency in algorithms is a form of power in visibility in that those who control the technology can impose their will without being accountable, while those subject to it are kept in the dark (Kwok and Chan, 2024a, 2024b). The case here exemplifies such dynamic. The intentional invisibility of platform algorithms serves to entrench the power of platform companies and marginalize workers’ knowledge of their own conditions.

Confronted with these obstacles, gig workers must engage in significant interpretive labor to make sense of their situation. Ellie Anderson's (2023) notion of “hermeneutic labor” helps to sharpen what is at stake in workers’ ongoing efforts to make sense of opaque algorithmic management. Hermeneutic labor is the “burdensome activity” of rendering one's own feelings, intentions, and motivations intelligible, discerning those of others, and devising responses to tensions that arise when communicative cues are partial, ambiguous, or strategically withheld (Anderson, 2023: 178). It is, in her account, a distinct kind of interpretive work that is cognitively demanding and disproportionately borne by the less powerful party in an asymmetric relationship. Although she develops the concept primarily within the micropolitics of intimate relationships, its core claim is not essentially about intimacy; it is about the distribution of interpretive work under power inequalities. Where one party controls the terms of intelligibility, such as what can be known, what is acknowledged, what counts as a reason, and where the other party is nonetheless expected to sustain coordination, the less powerful party is predictably burdened with ongoing interpretive repair.

Platform labor exhibits precisely this structure. The platform is a continuing governance relation with largely unilateral power. And while the relevant “other mind” here is not a single individual's psychology, it is still meaningful to speak of the worker's need to “discern the intentions” of the other party as well as “inventing solutions” for tensions (Anderson, 2023: 177). The algorithmic interface, after all, stands in for the firm in its dealings with workers. It gives orders, signals approval or disapproval, and imposes consequences, yet it rarely discloses the reasons behind any of these. In this setting, the worker's problem is more than informational scarcity; it is also an ongoing demand to stabilize a workable interpretation of what the platform is tracking and valuing, despite the platform's ability and incentive to keep those requirements under-specified.

Consider the everyday experience of gig work. Workers encounter, for example, fluctuations in pay, rating penalties, deactivation warnings, and sudden drops in ride offers. These events function as action-guiding signals as they effectively command behavioral adjustment, yet platforms generally do not disclose the rules that would tell what those signals mean. Workers are accordingly pushed into hermeneutic labor of their own. They must interpret sparse cues, infer what metric or threshold the system is applying, attribute a “reason” to the platform's response, and devise coping strategies to protect income, such as changing locations, accepting more rides, working different hours, altering cancellation behavior, and so on. Framed this way, algorithmic opacity is an exploitative redistribution of interpretive work. The company can remain non-responsive by presenting the system as effectively infallible, while workers bear the cognitive and affective costs of keeping the relationship functional under conditions they do not control. Anderson (2023: 190) emphasizes that hermeneutic labor is often “largely invisible” to those who benefit from it. Platform firms likewise profit from workers’ unpaid interpretative work. When workers successfully “reverse-engineer” the system, they self-discipline in line with shifting metrics the company need not publicly justify (Vasudevan and Chan, 2022). The burden is also processual: when algorithms, incentive schemes, and evaluative standards change, workers must redo this interpretive labor, repeatedly and at personal cost, simply to remain legible to a system that refuses to be legible to them.

This labor is rarely only at the individual level. Because no single worker has enough information to reliably infer the platform's decision logic, gig workers often have to exercise additional hermeneutical labor at the aggregate level by collectivizing interpretation and pooling fragments of experience into ad hoc epistemic communities. They exchange information in online forums and driver networks to piece together the logic of algorithmic systems. It is common for workers to form Facebook or WhatsApp groups, or to use online forums like Reddit, to share anecdotes and data in an attempt to collectively reverse-engineer the platform's black box (Moore and Joyce, 2020; Murgia, 2024). Such grassroots data collection and pattern recognition are time-consuming and mentally taxing. For instance, couriers for a grocery delivery platform, Shipt, only discovered that a new pay algorithm was systematically reducing their earnings after an external intervention by a worker advocacy group, which helped gather and analyze data from thousands of deliveries. The analysis revealed that about 40% of workers experienced sharp and continuous wage decreases under the supposedly “effort-based” algorithm, even though the company had assured workers the change would increase average pay (Coworker.org, 2020; Gurley, 2020; Noor, 2020). Before this collective effort, each worker could only guess why their pay was shrinking. The burden of obtaining that proof fell on the workers themselves. This illustrates how cumulative and arduous the epistemic burden becomes: it took third-party assistance and a large dataset to confirm what individual gig workers intuitively felt but could not substantiate. During that time, workers had to keep laboring under worsening conditions, perhaps doubting their own suspicions at times, and spending unpaid effort to log information, organize, and analyze, all of which is cognitive labor over and above their job of delivering goods.

Notably, these epistemic burdens compound over time. As one problem persists, workers must continually revise their interpretations. The labor is thus processual: workers are never “done” learning the system, because the system is a moving target that the company deliberately keeps opaque. Moreover, the emotional toll, such as frustration, anxiety, and a sense of powerlessness, accumulates, contributing to what we can describe as mental exhaustion. This chronic uncertainty can erode workers’ confidence in even asserting that an injustice is happening, because when pressed for evidence or explanation, they fear they cannot fully substantiate their claims. In turn, that lack of confidence is a barrier to collective action. Indeed, opacity can be weaponized to undermine dissent, especially when workers themselves are not entirely sure about the mechanisms of their exploitation, it becomes harder for them to rally around specific demands or to convince others, such as the public and regulators, of the need for change. Here we see how the epistemic burden ties directly into political agency. Without clarity and credibility, workers’ voices remain unheard of and unattended to. Platform algorithms produce outputs desired by platform firms precisely because they are opaque.

Despite these obstacles, gig workers do resist epistemic marginalization by building informal peer networks for collective sensemaking. Across messaging apps and online forums, dispersed workers compare experiences, air grievances, and piece together working knowledge of the opaque systems that govern their work (Kellogg et al., 2020; Tassinari and Maccarrone, 2020; Yu et al., 2022). But why must workers go to such lengths to acquire basic understanding of their work conditions? In traditional employment, a worker could ask a human manager for clarification or rely on legally mandated transparency, like a pay record detailing hours and rates. By contrast, platform workers have no HR department to consult, no statutory right to algorithmic transparency, and no straightforward way to question algorithmic decisions. Instead, they must rely on back-channel strategies and crowd-sourced detective work. This is thus a processual burden: an ongoing requirement that they invest time and mental effort beyond their paid labor. The injustice lies in the necessity of this extra work and in the unequal distribution of epistemic labor. The workers, already the more vulnerable party, are saddled with figuring out information that the platform already has but withholds.

Obfuscating accountability and testimonial silencing

The lack of transparency in algorithmic management is compounded by deliberate ambiguity in accountability structures on platforms. Even if a gig worker manages to pinpoint a problem, such as a sudden account deactivation or an unfair pay cut, they often face a confounding question: who can answer for this or set it right? Platforms intentionally convolute the lines of accountability, making it difficult for workers to trace causes and seek remedies for problematic and wrongful decisions (Bieber and Moggia, 2021; Halliday, 2021). Such “accountability complication” is often a corporate strategy of muddying the waters of responsibility to increase workers’ epistemic burdens in pursuing their claims and justice. In the platform economy, the complication of accountability is done through two ways: (1) through automated and impersonal decision-making and (2) through complex corporate structures and outsourcing (Kwok, 2025b; Muldoon and Raekstad, 2023; Zhang, 2024; Muldoon, 2022). This section examines how such complication not only frustrates workers’ attempts at seeking redress but also effectively silences their testimony, leaving it with nowhere to go, or ensuring that when it does reach someone, it is simply denied.

First, consider disciplinary decisions made by algorithms via opaque processes. On many platforms, terminating a worker's access to the app (deactivation) can happen automatically or with minimal human oversight, triggered by metrics or user reports that the worker usually does not fully see (Moore and Joyce, 2020; Veen et al., 2020). In a survey of 810 current and former Uber and Lyft drivers in California, two out of three respondents (66%) reported experiencing at least one deactivation; among drivers who had been deactivated, 30% reported that Uber or Lyft provided no explanation or reason for the deactivation (Rideshare Drivers United & Asian Americans Advancing Justice, 2023). Even those given a reason often find it “incorrect, unfounded, or based on incomplete information” (Rideshare Drivers United & Asian Americans Advancing Justice, 2023: 21).

For example, one driver, M. Ahmed, had both his Uber and Uber Eats accounts suddenly suspended in 2020 and then deactivated in 2021 (Worker Info Exchange, 2021: 46–48). When he inquired, Uber first cited a failed facial recognition check, then later claimed he had too many undelivered orders. Each time Ahmed tried to follow up with evidence and explanations, he hit a bureaucratic wall: at one point Uber told him he needed to log into the very account they had deleted in order to resolve the issue. After several rounds of fruitless emails, the company simply stopped responding. Ahmed's case exemplifies a burdensome process of epistemic navigation. He was forced to play detective, attempting to figure out what infraction, if any, he actually committed, and to plead his case through impersonal channels that provided no accountable interlocutor. The shifting explanations and eventual silence nullified his testimony. No matter what he said or proved, which, ironically, he had clear evidence of his compliance, there was no one responsible who actually heard him out. This is a form of testimonial silencing: even though Ahmed was actively speaking, via sending emails, asking questions, providing documentation, his voice was absorbed into a void where no responsible hearer took ownership of addressing his claims.

Another driver, Alexandru Iftimie, described a similarly Kafkaesque scenario of algorithmic suspicion. Uber's system sent him an automated warning that he had been “flagged for fraudulent activity”; two weeks later, a second warning arrived, and he understood that “three strikes” would mean termination (Murgia, 2024: 114). These decisions were communicated entirely through automated messages, with no clear channel for appeal and clarification. Unsure what conduct could be “construed as deceptive,” he found himself “second-guessing an inscrutable computer system” (Murgia, 2024: 114). When he called driver support, the agent offered no case-specific explanation and repeated: “The system can’t be wrong, what have you done?” (Murgia, 2024: 114). Facing the prospect of an account shut-down, he stopped taking Uber jobs because he was “too afraid to lose access permanently” (Murgia, 2024: 115). In his own analogy, it was like finding “a warning left on your desk by your boss,” who remains inaccessible and will not say what the alleged wrongdoing is, and the experience left him feeling “targeted, discriminated against” (Murgia, 2024: 115–116).

This analogy reveals a core epistemic burden. In a normal workplace, one could ask a boss for clarification or appeal to HR, but in the gig economy there is effectively no person to talk to. The worker is left to guess and even self-censor their actions in hopes of not triggering unknown rules. This is both a hermeneutical burden (i.e., the interpretive void regarding what rule did I break? what actually went wrong?) and a communicative burden (i.e., how do I protest or correct an error when I can’t get a human audience?). Iftimie's story also illustrates how testimonial injustice can be baked into these systems: the presumption that “the system can’t be wrong” implies that any challenge by the worker is automatically seen as lacking credibility. Both Ahmed's and Alexandru's experiences show how automated and impersonal management practice effectively silences workers: not by preventing them from speaking at all, but by ensuring their speech never actually reaches a responsive and responsible hearer who might address the issue.

The second way platforms complicate accountability is through complex corporate structuring and fissuring employment relation, which obscures who the employer is and where employer responsibilities should attach (International Labour Organization, 2021). Many platform companies operate an ecosystem of subsidiaries, shell companies, and third-party contractors (International Labour Organization, 2021). This means that when workers seek to assert rights, they encounter a hall-of-mirrors effect: the platform points to the subcontractor, the subcontractor points back to the platform or to another entity, and so on. For example, Just Eat, a UK food delivery platform, advertises steady jobs with benefits, but some couriers find themselves technically employed by third-party agencies, like Randstad or Stuart, which then deny them some of their benefits and worsen their pay conditions (Meaker, 2022). Couriers report that if they raise an issue, “Just Eat would refer you to Randstad, Randstad would refer you back to Just Eat” (Meaker, 2022). This deliberate buck-passing creates epistemic confusion: the worker must figure out who actually holds decision-making power over different aspects of their work. The burden of simply identifying the responsible party falls on the worker. The International Labour Organization (2021) has noted that such complex contracting is a growing problem. In many cases, the platform's convoluted contractual and corporate architecture becomes most visible when workers try to enforce their rights: at that point they confront what lawyers call an “impenetrable legal armoury” that pushes them into protracted litigation (International Lawyers Assisting Workers Network, 2021: 10).

A clear illustration is Uber South Africa Technology Services (Pty) Ltd v National Union of Public Service and Allied Workers (NUPSAW) and Others (2018). After Uber drivers were deactivated, they tried to bring an unfair dismissal claim in South Africa. Uber's corporate setup, however, split functions across entities: drivers’ contracts pointed to Uber BV (based in the Netherlands), while Uber SA (the local company) presented itself as merely providing support services. The court accepted this division, holding that Uber SA “did no more than provide administrative and marketing support to Uber BV” (Uber South Africa Technology Services (Pty) Ltd v NUPSAW, 2018: 3) and therefore was not the drivers’ employer. Crucially, the court also did not decide whether Uber BV was the employer, leaving the workers with no clear target for accountability in that proceeding. This case exemplifies how multi-layered corporate arrangements can make it highly difficult for workers to pin down the subject of their right-claiming. If a worker wants to sue or demand contractual rights, they first need to spend substantial effort to understand corporate law and discern the network of entities involved. The epistemic burden here is the burden of sorting out complex legal fictions, which is an onerous task for anyone, let alone a gig worker who might not have the legal literacy and resources to trace business registries and contracts. From the worker's viewpoint, this complexity results in delays, frustration, and often the abandonment of claims. The process can be so draining that many do not pursue complaints at all. In this sense, accountability complexity is not just a legal shield for companies, but a form of epistemic dilution of workers’ testimonies.

Having explored how platform workers are burdened in making sense of their exploitation and in finding someone accountable to hear their claims, we now turn to the final, and often last-resort, arena of this process: the legal system. If a worker presses on despite opacity and obfuscation, they may attempt to vindicate their rights through litigation and regulatory action. As the article shows next, even here the odds are stacked in ways that drain the epistemic and material resources of workers, adding yet another layer to the protracted struggle.

Legal struggles and epistemic draining

For many gig workers, the ultimate step in claiming justice is through the courts or other legal processes, challenging their misclassification, suing for unpaid wages or injury compensation, and seeking regulatory enforcement. Unfortunately, the legal avenue, which should in theory provide clarity and remedy, often becomes a venue for prolonging and intensified epistemic burdens. Platform companies have taken advantage of legal grey zones and procedural complexities to engage in what I term “epistemic draining” (Chan and Kwok, 2021). Epistemic draining is a process by which firms drag workers into protracted legal battles, exploiting uncertainty and asymmetries of information to exhaust the workers’ capacity to make and sustain their claims. It is “draining” both intellectually and emotionally, as workers must learn new legal concepts, gather extensive evidence, and endure lengthy and multi-stage proceedings, all of which require significant epistemic effort over time (Ahmed, 2021; Garden, 2018; International Lawyers Assisting Workers Network, 2021; Stone, 2016). In many cases, platform companies also use these legal fights as learning opportunities to further insulate themselves. They adapt their practices when workers find a gap, thereby forcing workers to constantly chase a moving target (Garden, 2018; Lei, 2021; Stone, 2016). When combined with legal processes that can take years and involve multiple preliminary hearings, this iterative redesign predictably deters sustained rights-claiming.

The legal context imposes processual epistemic burdens on gig workers in three main ways: (1) the burden of proof in establishing employment status and rights; (2) the uncertainty and inconsistency of legal standards; and (3) the extreme information and resource asymmetry between individual workers and platform corporations. There is one further caveat. Jurisdictional context shapes how processual epistemic burdens play out in practice. The burdens I describe in the “legal draining” phase are not uniform across legal orders. They are mediated by national, and sometimes sub-national, legal infrastructures that determine what counts as a legally cognizable claim, what evidence is required, whether workers can compel disclosure of platform data, and whether dispute resolution is public and precedent-generating or privatized and fragmented. The same platform architecture can therefore produce markedly different epistemic journeys depending on the legal system in which a claim is pursued. UK litigation, for instance, is shaped by the availability of an intermediate “worker” status and by employment-tribunal routes, as illustrated by Uber BV and others v Aslam and others (). In the United States, by contrast, standard contracts often require individualized confidential arbitration and waive collective procedures, which suppresses claims and prevents employment-status disputes from being resolved publicly and on a precedential basis (National Employment Lawyers Association, 2021; Stone, 2016). My aim here is not to offer a comparative labor law analysis, but to make a narrower point: the allocation of epistemic labor is partly a function of legal design. Given this contextual sensitivity, in developing my arguments, I therefore draw on cases and examples from a range of jurisdictions to show that the epistemic burdens I identify recur across different institutional contexts.

First, in many legal systems, when a worker files a claim, the burden of proof lies on the plaintiff, meaning the worker, to prove the facts of their employment relationship and the alleged wrongdoing (Heyes and Hastings, 2017). For pseudo-gig workers, this burden is especially heavy. It is not enough to show that the platform exerted substantial control over their work; they often must prove multiple criteria of employment status, such as mutual obligation, lack of entrepreneurial independence, integration into the company's business, and so on (International Labour Organization, 2016; Risak, 2018), all while the company will contest through different legal strategies that they are independent contractors. Many of the doctrinal indicators used to classify platform work turn on facts that are difficult for workers to observe and document. For example, courts often treat as key indicators whether the putative employer retains a “right to control the manner and means” of performance, and whether the worker is effectively obligated to accept work when “on duty” (International Lawyers Assisting Workers Network, 2021: 72, 78).

In gig work, the contract might say the worker can choose when to work, but in practice the algorithm might nudge and penalize workers for not working at peak times (Vasudevan and Chan, 2022; Veen et al., 2020). To prove the reality of control versus the paper independence, a worker needs access to data about how the algorithm schedules tasks or what the company communicates to workers, which is information typically in the company's sole possession. Platforms hold digital records, and they often deny workers access to their own data. For instance, once deactivated, workers may be locked out of their accounts, therefore losing all their work history evidence. This creates a vicious epistemic circle: the worker needs evidence from the platform to prove the platform is an employer, but the platform, while contesting that it is an employer, has no legal duty in many jurisdictions to hand over evidence unless a court orders it, which typically happens only after the worker somehow makes a prima facie case. Even in discovery, companies can claim trade secrets or simply bury the process in paperwork given their greater resources. In short, platform workers often enter legal proceedings at a structural epistemic disadvantage: the key evidence needed to establish employee status and enforce rights lies in a complex “web of contracts” and in the platform's “internal workings,” which workers cannot access, making proof in court extremely difficult and sometimes effectively impossible (Risak, 2018: 10).

A recent example from Zeek in Hong Kong is revealing. Six couriers brought Labour Tribunal claims for wage arrears and statutory entitlements, and the Tribunal affirmed that they were employees rather than contractors (Lee, 2023a). Yet the status victory did not translate into an effective remedy: Zeek had already ceased operations and the company's side claimed bankruptcy, pushing workers toward the government established protection fund (Lee, 2023a, 2023b). The case also exposes a second-order epistemic obstacle. Riders lacked written employment agreements and described being exhausted by the need to assemble “documentation and proof” across multiple stages (Lee, 2023b). Meanwhile, compensation eligibility depends on demonstrating “services rendered” within a limited period prior to the “last day of service,” a marker that is itself ambiguous in platform work. In this sense, even after winning the employment-status argument, couriers can still be forced to “prove the remainder,” such as concrete wage proof and eligibility, under conditions where the relevant evidential trail is fragile and may be practically inaccessible once the platform has shut down. As a worker who launched the legal battle forcefully put it: “We’re tired … The whole process took a lot of time” (Lee, 2023b).

Second, beyond specific proof issues, gig workers face broader uncertainty in the law itself. The legal status of gig workers has been a moving target in many jurisdictions. Courts and regulators have not reached a consensus: different tests and approaches apply in different places, and even within one jurisdiction, landmark rulings do not always settle the issue, as lower courts can diverge and jurisprudence remains contested (European Commission, 2021). From the worker's standpoint, this means navigating legal ambiguities, which is an epistemic burden of having to become, in effect, an amateur labor lawyer to gauge their chances. The European Union, for instance, has noted that “[t]here is no standard definition of the term ‘worker’ in EU law, which makes it difficult to delineate and classify the grey zone between traditional employment and self-employment” (European Parliament, 2025: 3). Gig workers pressing for rights often find that judicial outcomes are inconsistent: while some courts have ruled in their favor, as in the landmark UK Supreme Court case Uber BV and others v Aslam and others (2021), which held that Uber drivers qualify as “workers” entitled to minimum wage and other protections, other courts and subsequent cases do not necessarily follow such precedents. Later, the UK Supreme Court in another case concerning Deliveroo riders, Independent Workers Union of Great Britain v Central Arbitration Committee and another (2023), reached an opposite conclusion, holding that the Deliveroo riders were not in an employment relationship. This time the court emphasizes riders’ flexibility to reject work and use substitutes as evidence of independent contractor status. The same legal system, even the same court, thus made starkly different conclusions based on nuanced factual differences that may not seem significant to the workers themselves.

The same phenomenon also exists in other common law jurisdictions. In Australia, recent decisions have often moved in the opposite methodological direction by adopting a more “contract-centered” orientation. This is visible both in the High Court's approach to employment status in ZG Operations Australia Pty Ltd v Jamsek (2022) and in gig-economy related cases such as Deliveroo Australia Pty Ltd v Diego Franco (2022), where tribunals prioritized contractual structure over the broader “economic realities” of platform dependency. Thus, a gig worker contemplating legal action must interpret multifactor legal tests, predict how a given judge might weigh them, and keep informed of new decisions that might influence their case law support (European Commission, 2021). This is a heavy epistemic demand, which is essentially requiring workers to engage in ongoing legal hermeneutics just to understand their own rights.

Moreover, because of these uncertainties, even a win at trial may require appeals and re-appeals. A company that loses in one court might appeal to a higher court with a different view. Workers may have to endure multiple rounds of litigation: for instance, an initial labor tribunal, then an appellate court, maybe up to a supreme court. Each round involves new legal arguments, delays, and various costs. Dufresne and Leterme (2021) report that in their dataset of 59 European court decisions on platform-work status (2016–2020), about 40% of court decisions still did not favor reclassifying platform workers. The prospect of a long fight with uncertain payoff is itself an epistemic deterrent. For rational workers who are aware of these odds, they may decide it is not worth pursuing a claim at all. The effort to just figure out “What does the law actually say? What evidence do I need? What are my real chances?” can deter filing a case before it even begins. This is a kind of chilling effect inflicted by complexity and unpredictability.

Third, when workers do pursue formal complaints or legal action, platform firms frequently use contractual mechanisms to limit their exposure. A widespread strategy is to require workers to sign arbitration agreements as a condition of employment, which mandate that disputes be resolved through private arbitration, while arbitration proceedings are confidential and tend to produce outcomes less favorable to workers (National Employment Lawyers Association, 2021; Stone, 2016). This is an asymmetry of legal knowledge and resources. Workers must navigate complex contractual terms drafted entirely by the firm, often without realizing what rights they have forfeited, which is again an epistemic game of legal technicalities where the company has significantly more advantages.

Delay is another common tactic. Platforms often stretch out proceedings in the hope that workers will exhaust their resources before reaching a resolution (Feiner, 2020; International Lawyers Assisting Workers Network, 2021; Mazur and Serafin, 2023; Muldoon and Sun, 2024). Each postponement requires the worker to remain psychologically committed to the dispute, often while forgoing income due to deactivation and diverting time away from work. As cases drag on, workers put life plans on hold, adding to the emotional toll of the dispute. Prolonged proceedings also impose epistemic costs: memories may fade, digital evidence may be lost or become harder to retrieve, and workers must continually update their understanding of legal developments that could affect the outcome of their case.

Further reflections

A natural question is that the preceding discussion seems to generalize from platform work that is governed through algorithmic management. Yet algorithmic management is not unique to gig work. Software-mediated scheduling, evaluation, and discipline have spread widely across standard employment. Moreover, some platforms function more like “passive” marketplaces that match customers and providers without exercising employer-like control. If so, why treat the epistemic burdens described in the preceding discussion as distinctive of gig work than as a more general feature of algorithmically managed work or even of ordinary market opacity?

First, although the article's focal point is dependent pseudo-gig work, its core mechanisms can extend to other analytically similar labor settings, including non-gig workplaces, insofar as they reproduce the same structural features: unilateral informational control, weak contestation routes, and fragmented responsibility. I focus on dependent pseudo-gig work because platform labor is a rapidly expanding domain of work and a central site of contemporary struggles over employment status and labor rights, as illustrated by high-profile conflicts involving firms such as Uber and Deliveroo and the wider organizing and litigation ecology around them (Dufresne and Leterme, 2021; Vallas and Schor, 2020). This setting makes especially visible how firms can exploit regulatory uncertainties and invent novel contracting arrangements that significantly raise the cost of the epistemic work involved in workers’ claim-making process (Bieber and Moggia, 2021; Chan and Kwok, 2021).

Muldoon and Raekstad's (2023) republican framework helps sharpen the scope claim. They define algorithmic domination as subjection to a dominating power whose operations are determined directly by an algorithm, such that instructions, judgments, and communications can reach workers without intermediate human agency. While their focus is the gig economy, they are explicit that the concept applies in principle wherever algorithms are used to manage workers, including in standard employment (Muldoon and Raekstad, 2023). This article could be read as following their general spirit. A central strand of their analysis concerns the informational and computational asymmetries that reduce workers’ capacity to understand and contest the decisions made over them. This provides support to treating platform work as a diagnostic amplifier, and their important work opens up a further question which the article centers on: at what point does algorithmic domination become epistemically draining, and through what mechanisms?

Second, the three epistemic burdens developed above are distinct in their origin and in how they accumulate over time. The hermeneutical burden originates in unilateral information control: proprietary and shifting rules make it costly for workers to interpret pay fluctuations, ratings, allocation patterns, or deactivations, thereby shifting ongoing interpretive work onto workers (Anderson, 2023; Rahman, 2021). The testimonial burden originates in fractured accountability: when decisions are attributed to “the system” and responsibility is distributed across corporate and contractual layers, workers struggle to identify an accountable hearer for complaints, turning even well-grounded grievances into failed attempts at being taken seriously (Dotson, 2011, 2014). The burden of epistemic draining originates in misclassification-driven legal battles: because employment status is disputed and key data remain platform-controlled, workers must repeatedly translate lived experience into legally cognizable claims under asymmetric evidential burdens. What is distinctive of dependent pseudo-gig work is not any one mechanism in isolation, but their convergence into a predictable sequence: opacity generates the need for interpretation, obfuscation undermines meaningful voice, and legal contestation makes the remaining path to recognition a war of attrition.

These clarifications also limit the thesis of the article. A platform that, for example, primarily facilitates matching without unilateral control over price-setting, visibility, and sanctions, may still be opaque, but it will not typically generate the full sequence described here. In contrast, standard employees subject to far-reaching algorithmic management can face comparable burdens where voice and contestability are weak. The platform gig economy is therefore treated here as a particularly vivid, not exclusive, site in which algorithmic domination intersects with dependency, misclassification, and structurally diffused responsibility, producing intensified processual epistemic burdens. That being said, to what extent these processual epistemic burdens arise in other contexts is a question that lies beyond the scope of this article. But the arguments offered here can be read as a first step toward articulating an epistemic dimension of labor injustice that remains underexplored, and as an invitation for future research to extend this lens to other settings.

Conclusion

This article introduces processual epistemic burdens: the epistemic work of making one's experience intelligible, credible, and actionable under conditions adversarial to institutional recognition. Using platform labor as a case, it shows how algorithmic opacity forces workers into continual interpretation, fragmented accountability turns complaints into credibility struggles, and misclassification-driven legal contestation drains time and resources through evidential asymmetries and delay. The contribution of the article is thus both to catalogue harms and to reframe labor injustice as epistemically mediated. Platforms secure control by offloading investigative, interpretive, and justificatory labor onto those they govern while shielding themselves from accountability. Treating these burdens as an additional locus of injustice highlights the significance of redistributing epistemic labor through rights to explanation and data access, clearer responsibility chains, and procedures that lower the cost of contestation. Recognizing and redistributing this labor is thus central to epistemically just societies.