The contingencies of platform power and risk management in the gig economy

Introduction

How and where does power take shape in the platform-mediated gig economy? To date, the prevailing answer among gig economy scholars is that platform companies exert an excessive amount of power by using algorithmic management systems that enable workforce control at scale (Griesbach et al., 2019; Wood et al., 2019; Stark & Pais, 2020). Indeed, critics have argued that algorithmic technologies are “ushering in a novel form of rational control that is distinct from the technical and bureaucratic control used by employers for the past century” (Kellogg et al., 2020, p. 383). This development has been subject to extensive scholarly scrutiny, as has its dual counter-movement: individual and collective forms of worker resistance to the power of algorithms on the one hand (Anwar & Graham, 2020; Cameron & Rahman, 2022; Maccarone et al., 2023) and regulatory or policy initiatives to curb this power on the other (Ponce & Naranjo, 2022; Veale et al., 2023). Yet what do we miss about gig workers’ everyday negotiations of platform power across specific settings and situations when prioritising the study of algorithmic labour control? While the implementation of algorithmic management systems has certainly afforded platform companies distinct means of control, it is equally clear that such control does not materialise identically across platforms or industries (Griesbach et al., 2019) or will have the same impact on all gig workers (Schor et al., 2020).

Drawing on a five-year research project investigating platform-based food delivery and domestic cleaning in Amsterdam, Berlin, and New York City, this paper advances the argument that understanding the situated and contingent nature of platform power in the gig economy requires an analytical shift away from platform-governed algorithmic management and toward the risk management practices of gig workers — practices that exceed the point of production and are caught up in various kinds of power. To be sure, algorithmic management is essentially a form of risk management for platform companies, whose primary operational risk follows from their practice of hiring workers as independent contractors rather than employees, which significantly reduces labour costs but also limits managerial control over their workforce. Importantly, however, algorithms do not only support labour process control. Approached more expansively, they also help to “stabilise complex and ongoing social relations” (Thomas et al., 2018, p. 2) predicated on a neoliberal consensus that everyone — companies and workers alike — must accept responsibility for managing the risk associated with being an entrepreneurial market participant (Trnka & Trundle, 2014). In doing so, their implementation can reinforce and even exacerbate capitalism’s unequal distribution of the benefits and costs associated with risk management, which structurally favours corporations over workers (Snider & Bittle, 2022). Besides their algorithms, platform firms have at their disposal a range of financial and legal resources for managing business risk and reproducing their power across value and wealth chains (Grasten et al., 2023).

It is clear that, in comparison, gig workers have significantly fewer and less powerful means to manage risk. While they are confronted with different types of risk, they lack the capacity to strategically disembed from risky spaces or jump through legal loopholes in ways that allow platform companies to benefit from risk financially (Grasten et al., 2023). Nevertheless, a handful of studies (e.g. Schor et al., 2023; Maffie, 2023; Lefcoe et al., 2023; Gregory, 2021) have shown how gig workers in different sectors and settings use every available resource to navigate or even exploit platform-mediated risk to the best of their abilities, often under precarious circumstances. Yet we still know little about how the socioeconomic conditions and trajectories of specific gig workers shape their vulnerability to various kinds of risk or their access to resources that support the management of these risks, both of which can be expected to mediate their susceptibility to platform power — as will be explained below.

In response, this paper takes what Aaron Shapiro and I call a “platform-adjacent” approach to the study of platform power (Van Doorn & Shapiro, 2023), which situates gig work within people’s broader work and life trajectories. A platform-adjacent approach includes in its analytical scope biographies, social arrangements, and forms of institutional (dis)embeddedness usually omitted from gig economy research that trains its critical eye on labour process control. In this way, it seeks to generate a more nuanced and comprehensive understanding of how platforms become integrated into people’s everyday lives and how platform power is articulated and negotiated over time in conjunction with other power relations. As I will show, gig platforms can feature both as a resource for risk management and a new source of risk, depending on the complex interaction between a platform’s labour management strategies on the one hand and the mix of support structures, dependencies, and responsibilities in a worker’s life on the other. Besides expanding and refining the scope of scholarly research, a platform-adjacent approach also urges regulators and policymakers to rethink the parameter of their response to platform power in the gig economy, highlighting how it is more than a labour or tech issue. Beyond the point of production, platform policy should be thought of as social policy (Van Doorn, 2022) since curbing the power of gig platforms will require comprehensive reforms across policy areas such as low-wage labour, migration, welfare, mental health, and housing. The viability of such reforms will, however, depend on state actors’ willingness to collectively examine how their own powers are entangled with — and may even shore up — platform power.

In the next section, I discuss scholarship that has examined the power of gig platforms in relation to their capacity to shapeshift between being a workplace and a marketplace, which matters for how we conceive of risk in the gig economy. The following section then switches perspective from the platform company to the gig worker, critically engaging especially with Juliet Schor’s contributions (Schor et al., 2020; 2023) to (re)theorise the relationship between workforce heterogeneity, platform dependency, worker approaches to risk, and the articulation of platform power. This sets the scene for section 3, elaborating on the study’s methodology, which in turn contextualises the two biographical worker narratives that are presented in section 4. Section 5 discusses these narratives in light of the key conceptual frameworks derived from the literature, before concluding with a reflection on the policy implications of taking a platform-adjacent approach to the study of platform power in the gig economy.

1. Platform-mediated gig work: Between workplace and marketplace risk

To compensate for the lack of formal managerial control over their freelance workforce, gig platforms have introduced technical and legal innovations that intensify labour control and worker precarity while rendering these changes legible to workers and courts as part of the entrepreneurial risk associated with market-based freedom (Maffie, 2023). These blurred boundaries between gig platforms as a workplace and a marketplace are a prevalent theme in gig economy research. Richardson (2023), for example, proposes a shift from approaching platforms as sites of work to understanding them as infrastructures for work. As infrastructures, platforms “contingently structure work” through “the socio-technical practices of arrangement and coordination” that allow for both standardisation and flexibility (p. 2, emphasis in original). Notably, however, an earlier study by Richardson (2020) theorised platforms as market-makers rather than infrastructures for work, which, I argue, points to more than a shift in critical perspective. Instead, it reflects broader scholarly efforts to grapple with what platforms are and do. As Viljoen et al. (2021) have noted, the highly automated ways in which principles and methods from the field of mechanism design are implemented on digital platforms “carries forward the normative justification of markets into settings that appear like markets but operate more like control infrastructures” (p. 2). Keeping up this appearance is vital to the survival of gig economy platforms since, as mentioned above, it exempts these companies from costly employer status.

In an incisive attempt to bridge, or rather circumnavigate, the workplace-marketplace divide, Shapiro (2023) introduces the term “platform sabotage” to describe how “gig platforms actively engineer inefficient routines into and beyond the service transactions they broker” (p. 2). The introduction of such “strategic inefficiencies” (Shapiro, 2023) in effect raises the transaction costs that these platforms are supposed to lower, exposing gig workers to various risks such as having to perform unpaid work, dealing with suddenly cancelled jobs, getting into accidents, and negotiating income fluctuations (Gregory, 2021). But are these workplace or marketplace risks? On the one hand, platform sabotage generates the kinds of “degraded work” identified by Doussard ten years earlier, by updating “the strategies, tactics, and market-making practices employers use to maximise the benefits of low-wage work” (Doussard, 2013, p. 29). On the other hand, however, platform companies’ pioneering use of boilerplate Terms of Service agreements in a labour market context has sought to shift gig work from the realm of employment law to that of contract law (Cohen, 2019). Given this unprecedented strategic combination of technical and legal means by which these companies have created market-like architectures to govern their workforce, some scholars identify a profound transformation in which the platform presents a new “institutional” (Bratton, 2015; Kenney et al., 2021) or “economic” (Vallas & Schor, 2020) form that “incorporates many of the features of prior economic structures — markets, hierarchies, and networks” (Vallas & Schor, 2020, p. 282).

Since Vallas and Schor’s contribution to the debate is widely cited and offers one of the most sophisticated theorisations of platform power in the gig economy to date, I review it more closely here. They introduce the concept of “permissive potentate” to frame their argument “that platforms govern economic transactions not by expanding their control over participants but by relinquishing important dimensions of control and delegating them to the other two parties to the exchange”, namely workers and customers (Vallas & Schor, 2020, p. 282). Platform companies retain control over critical functions and decisions (e.g. task allocation, service pricing, data collection/analysis, and operational optimisation strategies) while distributing control over elements such as work methods, scheduling, or performance evaluation. Consequently, the authors assert, “the extraction of value rests on a new structural form [or “economic architecture”, as they also call it] in which platforms remain powerful even as they cede control over aspects of the labor process” (Vallas & Schor, 2020, p. 282). I would add here, following Viljoen et al. (2021), that this distribution of control among different “sides” of their platform market is precisely what enables gig companies to dress up labour control as market-based risk. But what about the gig workers who are expected to manage such risk? The next section shifts the analytical vantage point from the platform to the people doing platform work.

2. Theorising platform power’s differentiated impacts

One benefit of Vallas and Schor’s model of platform power is its pliability: it is able to accommodate differences with respect to a platform’s relative permissiveness, its specific distribution of labour control, and the impact of these decisions on gig workers — all of which will depend on the industry or type of work being governed, the regulatory context, and the chosen business model. As the authors admit, evidence from ride-hailing and food delivery suggests that permissive potentates can become “permissive predators” when they leverage the engineered freedom of platform governance to enhance the exploitation of their workforce (Vallas & Schor, 2020, p. 282; Shapiro, 2023). However, while the “permissive potentate” model of platform power accounts for heterogeneity among gig platforms, it has less capacity to explain diversity within the gig economy workforce and how such diversity mediates platform power. After all, although platforms clearly exert power, this will not affect all workers in the same way or to the same degree — as has been compellingly demonstrated in research on gendered and racialised experiences of gig work (Webster & Zhang, 2020; Milkman et al., 2021; Gebrial, 2022; James, 2024), and in the growing body of literature on gig economies in the Global South and China (e.g. Arriagada et al., 2023; Tandon & Rathi, 2022; Qadri & D’ignazio, 2022; Sun & Zhao, 2022; Pollio, 2019).

Vallas and Schor do note that, due to their low entry barriers, “platforms foster greater variation in the work orientations, labor market positions, and sociodemographic composition of their workers” compared to traditional low-wage industries (Vallas & Schor, 2020, p. 283), resulting in varying levels of workers’ dependency on platform earnings. But how to theorise the link between the former and the latter and thereby forge a more complex and relational notion of platform power? To help advance such a theorisation, I turn to a more recent contribution from the same authors, which builds on the “permissive potentate” model as well as on Schor’s earlier work on workforce heterogeneity and platform dependence (Schor et al., 2020).

Notably, rather than theorising the nuances of platform power directly, the authors’ objective in this study is to explain how the gig economy’s “structurally induced workforce heterogeneity” results in different orientations to risk among workers (Schor et al., 2023, p. 2). They propose a theory of “differential embeddedness, in which the structural positions that gig workers hold in the economy condition how they perceive and experience the risks that platform work entails” (Schor et al., 2023, p. 5, emphasis in original). Generally, the more a gig worker is embedded in the “conventional economy” and the greater their “distance to necessity”, or financial security (Schor et al., 2023), the more likely they are to perceive gig-related risks optimistically. To represent and schematize these variations, the authors introduce a typology of risk orientations ranging from consent to contestation. Gig workers who contest platform- and customer-induced risk, according to the authors’ findings, tend to be more economically disembedded and thus more dependent on platform earnings. What emerges from their study, then, is a structural(ist) model in which differential embeddedness determines platform dependency which in turn shapes risk orientation.

While this is an important contribution to the gig economy literature, and one that has substantially advanced my own thinking on the topic, there are a few issues that should be addressed before their model can help me rethink how power operates in the gig economy. First, I argue that the authors’ theory of “differential embeddedness”, while pertinent, is not sufficiently holistic and agile to fully account for the diverse and dynamic nature of gig workers’ socioeconomic situations, work motivations, and attachments to platforms. In their model, gig workers are statically positioned in an economic order, where being embedded means being financially secure. Yet this economistic perspective misses other relevant forms of embeddedness, such as membership of social networks (including familial and diasporic ties), which may afford modes of security and wellbeing that constitute a buffer against risk outside of the formal economy (Ray & Sam, 2023). Moreover, the means by which people establish a measure of “distance from necessity” are diverse and may entail various (intimate) power relations and inequalities that are likely to inform motivations for doing platform work and navigating its risks (Van Doorn & Vijay, 2021).

This is to say that it matters whether one’s financial security derives from “a secure ‘main job’, spousal or family support, savings, or other buffers” (Schor et al., 2023, p. 3), since some generate dependencies while others may offer more autonomy (Van Doorn, 2023; Ray & Sam, 2023). Finally, people take up gig work at various points in their lives, leaving and returning to it as their personal and professional life trajectories unfold (Van Doorn, 2023) and as platforms evolve in ways that (dis)advantage some workers over others. This means that workers’ position in the economic order is dynamic: rather than thinking in terms of static groups of more or less embedded gig workers and mapping their experiences on a typology of risk orientations, it is more productive to consider how different modes and shifting degrees of embeddedness correlate to variable, sometimes contradictory, understandings and experiences of risk.

This brings me to the second issue: the proposed theory doesn’t quite capture the ambivalence that is central to gig workers’ perceptions and experiences of risk. According to the authors, their findings can sensitise scholars “to a largely unrecognized duality of labor platforms: while for many, platform work constitutes a cause of the risks, for others it mitigates exposure to the risks they might otherwise confront” (Schor et al., 2023, p. 3). Rather than constituting a division between two groups of gig workers, however, my own research shows that this duality is better thought of as a fundamental ambivalence experienced by nearly everyone doing gig work: platforms mitigate some types of risk while simultaneously introducing new risks (Van Doorn, 2023). As gig workers, people face labour-related “market risk” (Maffie, 2023) induced by platforms and customers, which can be physical, financial, and epistemic in nature (Gregory, 2021). Yet as (undocumented) migrants, parents, welfare recipients, tenants, creditors, and/or debtors, they also navigate other kinds of risk — often in relation to various state institutions — that platform work can help them deal with, albeit in often provisional ways (Metawala et al., 2021; Wolf, 2022).

Risk, as I will demonstrate below, appears in multiple forms and comes from various directions and sources that frequently exceed the platform-governed point of production, emerging from or bleeding into spheres of social reproduction. Accordingly, studying how gig workers’ “differential embeddedness” influences their susceptibility to and negotiation of risk would benefit from a platform-adjacent approach that seeks to foreground “what has so far featured in the background of gig economy research, by scrutinizing practices and relationships located at the edges of the field’s conceptual and methodological boundaries” (Van Doorn & Shapiro, 2023, p. 3). These edges are where gig work intersects with people’s broader work and life trajectories: where social ties, income, assets, (care) responsibilities, rights, and obligations are gained, maintained, and lost or discontinued. Risk, as one can imagine, marks these edges.

Third, what does the above discussion have to do with platform power? If, as was argued in the introduction, the neoliberal consensus stipulates that businesses, workers and households alike are expected to engage in risk management practices to secure their survival, then power can be defined as the capacity to manage — i.e. to absorb, relocate, avoid, and/or monetize — risk. Accordingly, it is clear that power is unevenly distributed, both between gig workers and platform companies and between differently positioned or embedded gig workers. A platform-adjacent approach, adopting an expansive and dynamic perspective on differential embeddedness as well as risk navigation (rather than cultural logics of risk “orientation”), locates this uneven distribution of power in the extent to which gig workers are able to leverage platforms to manage risk in their everyday lives.

To be sure, platform dependency is a pivotal variable here, since the more dependent gig workers are on platform earnings, the more platforms tend to become a source of risk rather than a resource for risk management (Schor et al., 2020). The more power a platform has to determine a worker’s livelihood, the more disempowered this worker is likely to be. But, again, this is a dynamic situation, since people’s dependency on platform earnings will vary over time as they negotiate other opportunities and obstacles that fall within, as well as outside, the formal economic realm. Finally, platform dependency can itself not be neatly captured in economic terms, since people develop different kinds of (affective) attachments to gig platforms that exceed economic rationality (Bissel, 2022).

3. Methodology: Illuminating the “platform-adjacent” through biographical interviews

Building on the basic model proposed by Schor et al. (2023), the platform-adjacent approach to studying platform power expands the key notions of embeddedness, dependency, and risk beyond the authors’ economistic and structuralist interpretation. This expansion follows from the basic notion that gig workers are people and as such they are never solely concerned with risk minimization and profit maximisation. While these motivations certainly play a critical role, they are always embedded in larger, more complex and layered social fields where other interests, responsibilities, and desires inevitably intersect with, inflect, and counteract purely economic objectives. To state the obvious and say that gig workers are people thus serves as a reminder that they, like other workers, only spend part of their time working and that whatever else they have going on in their life will influence the decisions and risks they take at work, as well as the nature of their relationship to work and the provider of work. It also advocates for qualitative research on gig work that looks “beyond the gig” (Van Doorn & Shapiro, 2023), insofar as such projects encompass the larger social environments that gig workers navigate and attempt to grasp their activities and predicaments at work from a biographical vantage point that exceeds the platform-governed point of production.

So far, most qualitative gig economy research has interviewed people about their work experiences and strategies, focusing on negotiations of algorithmic management, variable payment structures, and precarious working conditions more generally. My own multi-sited research project likewise concentrated on these themes when I started my fieldwork in New York, in February 2018. Yet I gradually came to realise, through formal semi-structured interviews and informal conversations with food delivery workers and domestic cleaners, that a deeper understanding of gig work’s complexities and contradictions across sectors could only be gained by situating narratives about work experiences within people’s dynamic life worlds. As my fieldwork progressed, the orientations and points of departure of my interviews changed, conversations grew longer, and what had started out as worker interviews increasingly developed into life history interviews that explored (the entanglements of) professional and personal trajectories.

To be sure, I apply this term retroactively, since at the time I wasn’t aware that what I was doing closely resembled the life history or biographical interview method. Accordingly, my improvisational approach was not informed by the methodological debates and protocols that have shaped this mode of interviewing (Barabasch & Merrill, 2014). Nevertheless, my approach aligns with the key premise of the biographical and life history interview methods, namely that life trajectories are “more than the background carpet” of people’s experiences and identifications at work (Dybbroe, 2013, p. 111). Indeed, documenting life histories can offer nuanced insights into livelihood situations and states of wellbeing that change over time (Singh, 2019). What further associates my way of doing interviews with these methods is its dialogical elicitation of “rich stories”, ideally enabled by the cultivation of a personal connection between researcher and interviewee (Barabasch & Merrill, 2014).

Clearly, both are time consuming endeavours that in no small part depend on favourable institutional conditions and available resources. Being able to spend eight months in each field site (New York, Berlin, and Amsterdam) afforded me the necessary time to explore the local food delivery and domestic cleaning markets, experiment with participant recruitment methods, get acquainted with food delivery workers I met during participant and non-participant observation, and establish a sufficient level of familiarity and trust to secure formal interviews.¹ Being able to offer $15/€15 gift cards as a small compensation for participants’ time also helped in some cases. After 24 months of fieldwork across three cities, I conducted a total of 151 interviews that lasted between 42 and 175 minutes, with a median interview length of 118 minutes. While it proved to be difficult to get people to do follow-up interviews, I regularly communicated with participants in each city, mostly via Facebook Messenger and Whatsapp. When my fieldwork ended on 1 February 2020, right before the start of the Covid-19 pandemic, I learned via these app groups that many of my contacts (temporarily) stopped doing gig work and/or left town, especially those doing domestic cleaning. As I have discussed elsewhere, the majority of research participants in Berlin and Amsterdam were migrants (for more details on sample demographics, see Van Doorn & Vijay, 2021).

Finally, how to represent the biographical narratives collected through the interviews, in a way that conveys how participants’ platform-based and platform-adjacent experiences intersect over time? Given the space constraints imposed by most social science journals, gig economy scholars often choose to extract a collection of decontextualized quotes from their interviews, which serve to illustrate an analytical category or overarching theme instead of being used to flesh out participants’ life/work trajectories. By prioritising the quantity of participant voices over the depth and integrity of individual narratives, this way of presenting research findings tends to obfuscate how gig platforms become (dis)embedded in people’s everyday life. While having elsewhere opted for a thematically ordered presentation, in this paper I offer two longer-form biographical narratives that illuminate the lived varieties of “embeddedness”, “platform dependency”, and “risk management”.

The two participants profiled in these narratives depend on platform earnings, yet, as we will see, the extent of their platform dependency varies over time. In this sense, they are “representative” of this study’s participant sample (N=151), in which platform dependent gig workers are strongly over-represented, yet they were not selected to represent a particular gig worker category. The objective of the paper is not to make generalised claims about the experiences or attitudes of specific groups of gig workers, but rather to show the complexity, ambivalence, and change that marks people’s engagement with gig platforms and to examine the attendant power dynamics. What I am interested in is idiosyncrasy, not generalizability, even though the resonances between the two narratives also reveal common struggles. By documenting how these play out in the lives of a male-identified Berlin-based food delivery worker and a female-identified New York-based domestic cleaner, the biographical narratives can teach us about the situated, dynamic, and contingent nature of platform power in the gig economy.

4. Biographical narratives: Pablo and Tish²

Pablo

Pablo, a 38-year old divorced father originally from Gran Canaria, has worked in different restaurants, bars, and hotels for most of his life. While he is unsure about the exact number of jobs he’s had over the years, he knows that he got fired from each of them. His mental health problems and self-described “really difficult character” repeatedly got him into trouble with managers and customers, after which he would just move on to the next job. This trajectory took him to various EU countries, before he decided to move in with his German girlfriend in Berlin, where they later married and had two children (she later filed for a divorce). One day in 2016, after getting fired from another restaurant job, he felt so depressed about fighting with his wife and not being able to properly support his kids that he seriously considered ending his life. He felt like a failure, alienated from German society due to the language barrier and cultural differences, which also limited the jobs he could find. What saved him that day was a video of his children playing together, which he watched on his phone as he sat on the ledge of a bridge on the outskirts of Berlin. He realised he could not abandon them like this. What subsequently changed the course of his work trajectory was the conversation he had with a police officer who came to check in on him, as he was still sitting on the bridge. The officer asked him if there was anything that made him happy, to which Pablo responded that he’s at his happiest when riding his bike, listening to music. When the officer suggested he should try to find a job as a bike messenger, something inside his head “clicked” and he decided to follow up on the advice.

Via a friend, Pablo learned about the existence of food delivery apps like Deliveroo and Foodora. When he first signed up with Deliveroo his wife had to help him with the application because the paperwork was all in German. He then did a “trial run” with a team captain and, to his surprise and regret, he ended up being rejected. This was still in 2016, when the company’s recruitment process wasn’t as easy, fast, and accommodating as it would later become. Fortunately, his second attempt was successful and Pablo could then choose a contract type: employee or freelancer. After going through the contracts with his wife, he chose the former, since it was winter in Berlin and he was still feeling uneasy on the job. Because it was a so-called “Mini Job” contract,³ the main drawback was that he was not allowed to work more than 22 hours per week, but it did mean that he could save money by using his wife’s public health insurance. Besides, he was still receiving financial support from the Jobcenter (a German welfare-to-work agency), which complemented his income.

When Pablo started doing deliveries, Deliveroo’s market was still very small and there were few orders. The work involved a lot of waiting around in the cold and, despite getting paid an hourly wage for basically doing nothing, he wasn’t enjoying his new job. Things got better once more riders joined, since at least they could now hang out together and Pablo remembers how they used to smoke weed in his garage. For the first time since moving to Berlin, he felt a sense of belonging and community among fellow riders, many of whom were social outsiders like him. Those good times didn’t last very long, however. As the delivery market was growing and more orders started to roll in, Pablo noticed how freelancers — who were making 5 euro per delivery — were sometimes making more money than employees. So when his contract expired, he switched to freelancer status and this ended up changing his situation dramatically. Initially, he received fewer orders than anticipated and he was glad he could still claim Jobcenter support because, with 5 euro per delivery, he wasn’t earning enough to pay his bills. While as a Mini Job employee, Deliveroo had covered his accident insurance and he was covered by his wife’s health insurance; now that he turned freelancer he had to purchase both insurances at a considerable cost. When things picked up and he was completing more deliveries, this at first had a positive effect on his earnings, but once he passed a certain income threshold he lost his Jobcenter support, decreasing his net income. At this point, Pablo had radically changed his approach to delivery work, constantly calculating how many orders — now presented to him as offers — he has to accept to meet his income goals:

In the beginning I don’t have pay nothing, but then I calculate and I say "okay, so if I have to pay my Krankenversicherung [health insurance] and the BG Verkehr [accident insurance], I have to make like six or seven orders per hour, and at least three, with three orders I pay the insurance, whatever."

Pushing himself to work harder, faster, and smarter so he can manage the risk Deliveroo offloaded to him, Pablo decided to buy a scooter. Although he loves biking, on his scooter he can do up to seven deliveries per hour and make enough money to be comfortable. He’s also getting older, smokes a lot, and has a bad knee these days. He knows that, on a scooter, he can still compete with other riders, many of whom are younger immigrants. Since Deliveroo started working with freelancers and stopped hiring riders as employees, things have become more competitive and impersonal, and Pablo lost the sense of community he appreciated so much when starting out. There are just so many new riders these days. His current strategy is to work as many hours as possible and to accept nearly all incoming offers, since he believes Deliveroo punishes repeated rejections by throttling offers: “If you are accepting more, you will receive more. If you are rejecting many ones, your rates go down”.⁴ Although Deliveroo claims that offer acceptance rate isn’t used to (de)prioritise riders, Pablo says his own experiences suggest otherwise.

He’s also sceptical about the new “distance-based fees”, Deliveroo’s variable pay system. While he likes seeing the delivery destination before accepting an offer (this information was previously hidden), this system has made it even more difficult to anticipate his earnings and he’s not sure about how the platform calculates the fees. Overall, work is becoming more game-like, he says, especially since the introduction of the Weekend Bonuses: “I always try to make the bonus. On the weekend is more work and then you can make 120 euros. On the month it’s 240 [euro]. It’s a good money”. However, this high-intensity work routine also comes with physical and financial risks: I saw Pablo drive his scooter recklessly on multiple occasions and he had been in a few accidents.

When I bring this up, Pablo is cavalier about his driving style and, although he would like Deliveroo to cover his accident insurance, he seems to gladly accept what the company throws at him. Compared to what he’s been through in previous jobs, this arrangement suits him perfectly since it doesn’t come with the same stress, abuse, and exploitation that he never managed to get used to working in the hospitality sector. He claims that Deliveroo is the best job he’s had, at least in Germany. Living in a relatively cheap part of Berlin, Pablo’s expenses are modest and Deliveroo makes him “good money”. But, more importantly, doing delivery work makes him happy. For him, the job has been therapeutic: “the thing is, I cannot find a fucking psychologic that speak Spanish in Berlin. So when you are on the street ten hours, on the bike, you have time a lot for thinking, a lot of time for thinking many things. It's like therapy!” Although the work is intense, Deliveroo also offered him a space of reprieve from his daily mental struggles, most prominently during the period leading up to and following his divorce. As strange as it may sound to some, Pablo is thankful to have Deliveroo in his life. The platform both grinds and grounds him, the latter in both the mental and geographical sense. He tells me he hopes to be working for Deliveroo for the foreseeable future, just as he intends to stay in Berlin to see his children grow up. The platform and his kids are what embeds him in Berlin. He no longer feels the need to rebel, run off and try elsewhere.

This is why it pained me so much to hear that, only a few months after our interview in the spring of 2019, Pablo’s account was deactivated after an altercation at a restaurant. Apparently, it wasn’t the first time he got into trouble with a restaurant owner and this time Deliveroo decided to terminate their user agreement. It was devastating for him, since losing access to the platform meant not only losing his income but also being deprived of one of his main life anchors. He made repeated attempts to get reinstated, supported by pleas from fellow riders, but to no avail. When he realised he wasn’t going to be allowed back, he signed up with Lime, an electric scooter rental startup, for which he roamed around the city in the middle of the night charging batteries. His nocturnal lifestyle as a “juicer” was trying, mainly because it kept him from being around his children as much as he would like since he had to sleep during the day. But Pablo has always been tough so he managed to persevere until, right before the Covid-19 pandemic hit Germany, he signed up with Lieferando — a subsidiary of Just Eat Takeaway. In an ironic twist of fate, Pablo had returned to doing food delivery not very long after Deliveroo retreated from Berlin and other German markets in August 2019 (Altenried, 2021).

Tish

Tish, who is African American and recently turned 27, got her first cleaning job at the age of 18 through New York City’s Back to Work workforce re-entry programme. The summer before turning 18, she had already taken on a summer job at a women’s shelter through the City’s youth employment program, after her dad was incarcerated and she had to basically take care of herself. Once she moved in with her grandmother, she also helped her pay the bills. After her birthday, she turned to Back to Work because she needed the money and, having been on welfare through her parents, she preferred having a job and receiving a wage. Although she didn’t want to do cleaning work at first, she decided to give it a try. The first company she worked for paid $9 an hour and quickly went out of business, after which she found another cleaning job via Craigslist – off the books for $10 an hour. However, she quit this position after she found out that the woman she worked for scammed her out of nearly $700 in unpaid wages. By that time she was also in school for an associate degree in business management and tried to find a position that better matched her skills, but to no avail. The pressures of mounting bills and student loans then pushed her toward another cleaning company, which paid $11 on the books and promoted her to supervision after a few months. While she was “climbing up the pay ladder little by little”, her salary didn’t match the enormous amount of work she was now responsible for and she grew increasingly impatient. It was around this time that she saw a street table promoting the Handy platform, but it eventually took her another year to finally quit her job and sign up, in 2013.

Once she did, she loved the “easy approval” process. She only had to attend a group orientation session (a feature of Handy’s “onboarding” routine, since discontinued) where they checked basic cleaning knowledge and handed everyone a big blue bag containing a vacuum cleaner and other supplies for which they subtracted $150 from the first paycheck. Tish was fine with this: “I mean, Handy was saying ‘oh, we're hiring cleaners for $15 an hour’ and I'm like ‘oh my god that's the most I've ever made!’” As she was used to doing three to four cleanings per day, she claimed a lot of offers during her first months — as many as she could — and quickly moved from the $15 tier to $17 and then even to $22 for a while, raking in money. In those early days she would even get paid daily, which made her “feel like a celebrity, like I would just wake up and be like ‘am I dreaming?’” Pulling 50-hour work weeks was exhausting, especially with all the travelling, but the experience of financial independence was intoxicating. Handy’s tiered wage system further stimulated her working pace: to retain their tiered hourly rate, “Pros” (as the company calls its cleaners) need to keep up their ratings and meet the tier’s monthly job target. Once you sink below one of the set thresholds, you are demoted.

After taking a two-month break because she got sick from working too much and not taking proper care of herself, Tish found out that she was back at the bottom tier making $15 an hour. This made her feel a little betrayed: “I don’t like it because […] you work hard for it, and it's like all the work that I put in these last months doesn't mean anything right now […] You just feel forgotten about.” She had no option but to start all over again and climb the payment tiers. This time, however, she never made it back to the $22 tier because Handy made it harder to reach. The jobs were getting worse too, as she was often confronted with angry clients who took their frustrations with Handy or previous Pros out on her. Moreover, some clients would try to scam Handy by submitting a complaint so they wouldn’t have to pay for the cleaning, which means Tish wouldn’t get paid either or would even get fined. Fines are an important “stick” that forms a counterpart to Handy’s tiered system of “carrots” and seem to have grown into a source of revenue in and by itself (Van Doorn, 2018). One time she travelled to a distant neighbourhood in Queens only to be stood up by the client and then found out that Handy charged her a $50 fee for ostensibly failing to show up without prior notice. These kinds of incidents triggered a sense of indignation: “I've put a lot of work in, I had great ratings, you know, customers gave me great feedback, like why would I ever do a no show?” Whereas she used to be able to call a number and talk to someone, by then she had to email her objections and it took a long time to get a response. Her appeal wasn’t successful and Tish’s account was even deactivated for a few weeks.

When Tish started with Handy there were no fees yet, but now she had to be much more careful about arriving late for a job or leaving earlier, because clients could — and did — report her and she would be charged a $15 fee. Yet some things were beyond her control, such as when her grandmother got sick and she had to cancel a number of jobs less than 48 hours in advance, for which Handy charged her between $10 and $40 each. Instead of making money she was now losing income, which meant that she needed to accept more jobs to make up for her losses. If it would have only been possible to reschedule none of this would have happened, but Handy only shares the contact information of new clients four hours ahead of the job and erases it afterwards (for one-off cleanings). Importantly, this is not the only information the company erases or restricts: after being deactivated a second time, Tish permanently lost access to her account:

It was an obstacle because I had no proof of all the work that I did. Or the recognition. You know once you get deactivated, you can't log back into the app anymore […] Especially because Handy shows how many jobs I did. And I had like almost 700 jobs. So I would love to show people how long I've worked with them, how many jobs I've completed, you know.

Deactivation doesn’t only pose an immediate livelihood crisis but is also a threat to future job opportunities because you essentially lose your resume, which turns out never to have been truly yours in the first place. At the time of our interview, Tish was looking to set up her own cleaning company, expanding her customer base via the social network of the small group of clients she retained after her deactivation. Her biggest assets, she knows, are her extensive professional experience and her Associate Degree in Business Management. All she still needs is a business licence, for which she is saving up, and her own website. Since she’s been working for a local upscale cleaning service to earn extra money, she has received many good ratings and personalised reviews on Yelp and she’s thinking about copy-pasting these to her future website as evidence of her skills. Unfortunately, she won’t be able to do the same with the reputational data accumulated via Handy.

What worries her, moving forward with her business plans, is the chance that she might get audited by the IRS. Tish did not pay any income tax during the four years she worked through the platform: “Honestly, at that time, I didn't really care because it was just the money, and I didn't really understand it.” She believes the company should have done more to inform its Pros, considering that “the majority of people that work with Handy are minorities, you know, so we don't really know that much about taxes. Like, a lot of people that work there was low-income.” Still, she holds no hard feelings toward Handy and, overall, she looks back on that period as a “wonderful time”. In an almost cruel way, the platform set her on this path toward starting her own business, thereby making available something it could never offer: a potential career. With Handy, “it’s no progression, no benefits, no growth, you know?”, she reflects at the end of a long conversation. Her mood and perspective suddenly shift: “in that sense I wasted my time”. Her father always pleaded with her to get a job with benefits, but she continues to be on Medicaid despite officially exceeding the maximum income cap. That’s another risk she’s willing to take, for now.

5. Discussion: The contingencies of platform power

What can these two biographical narratives tell us about the differentiated and evolving nature of platform power in the gig economy? To start with a platform perspective, it is clear that both Deliveroo and Handy are attempting to minimise the operational risk/cost associated with gig worker autonomy. Handy’s approach has grown more disciplinary and punitive over time, tying its tiered wage system to a set of customer-evaluated performance metrics that became increasingly demanding, while introducing its NYC-based cleaners to a broad range of fees to deter them from breaking the platform’s rules (Van Doorn, 2018). If Handy can be characterised as a “permissive potentate” (Vallas & Schor, 2020), its power to dress up control as market-based risk ultimately derives as much from strategic and persistent acts of sabotage as from the outsourcing of discipline and punishment to customers. Deliveroo, meanwhile, has moved in the opposite direction. Whereas it employed its riders when starting in Berlin so that it could train and steer them directly, it switched to a freelance workforce whose newfound autonomy had to be curbed in ways that aren’t legible as labour control. Eventually, the firm introduced a “free login” system that further loosened its platform-based control infrastructure, making it appear like a food delivery market where rider “partners” are free to enter and exit at any time but are also expected to manage job-related risks (Viljoen et al., 2021). Accordingly, Deliveroo has grown into a “permissive predator” (Vallas & Schor, 2020), insofar as it deploys engineered freedoms to deepen the exploitation of riders who face increasing competition due to the company’s over-hiring strategy.

Despite developing in opposite directions, both Deliveroo and Handy negotiate the generative tension between openness and constraint — i.e. between being a marketplace and a workplace — that is constitutive of platform power (Vallas & Schor, 2020; Bratton, 2015; Williams, 2015). While they “set the terms of participation according to fixed protocols” (Bratton, 2015, p. 44), they also allow for a great variety of participants to enter the platform, resulting in gig workforce heterogeneity (Schor et al., 2023). What happens next will depend on the complex interaction between a platform’s operations, the particular industry or sector, the local regulatory setting, and the socio-economic situation of individual gig workers. In this paper, I have focused on the latter by building on Schor et al.'s (2023) theoretical model in which “differential embeddedness” explains the varying extents to which gig workers are dependent on platform earnings. Yet instead of asking how this shapes cultural logics of “risk orientation”, I have examined how it impacts people’s ability to empower themselves by using gig platforms to manage risk in their everyday lives, which are also shaped by other kinds of power relations.

Deliveroo showed up in Pablo’s life when he felt particularly low and alienated from German society, where he was trying to root himself after a long period of institutional disembeddedness while doing hospitality work across Europe. The company’s early “Mini Job” model felt relatively risk-free, especially since he was still covered by his then-wife’s health insurance and could claim welfare support. To be sure, this social safety net had an important embedding function, insofar as it provided Pablo a measure of financial protection. His then-wife also formed a social anchor, helping him to apply with Deliveroo and checking the contracts. In the early days of his tenure with Deliveroo, moreover, Pablo relished in the community he found among riders, which created a sense of professional embeddedness that made the negative aspects of the job more palatable. Upon switching contracts, his relatively embedded position started to disintegrate: Pablo lost his social safety net as well as his rider community, becoming increasingly dependent on a platform that was suddenly paying him per delivery and had offloaded most responsibilities for sustaining him off and on the job. While he was making a good amount of money by working very hard, and in this sense managed to keep some “distance from necessity”, he was also absorbing a higher level of platform-induced risk. The reason he did not just quit and try his luck elsewhere, however, is because he was considering another, more important kind of risk: losing his job and thereby letting down or, worse, losing contact with his children. In a psychologically complex way, the two had become intimately entangled and the thought of losing either one of his anchors in life made him exceedingly willing to go the extra mile for Deliveroo. Accordingly, the platform’s power over him — mental as much as financial — grew more entrenched.

In Tish’s case, her previous work experiences had created a certain path-dependency that made Handy an appealing option. Facing bills and student debt, the platform promised both higher earnings and more autonomy — and indeed this is what it offered for a while. In financial terms, Handy increased her “distance from necessity” and thereby embedded her economically to an extent that was previously inaccessible. Importantly, living with her grandmother and having access to Medicaid also helped in this regard, by lowering her expenses. Yet once first Tish and then her grandmother fell ill, Handy’s increasingly punitive labour management system turned against her and she found herself having to manage much more platform- and customer-induced risk than before. Still, although the costs of working through Handy grew significantly, Tish did not quit because she continued to see the opportunities it offered. She had also invested much time and effort into cultivating her reputation on the platform, which by then had entrenched its gatekeeping power. Once it permanently closed the gate on Tish, the outcome felt ambivalent: on the one hand, she mourned the loss of her work history, yet on the other hand it may have been the push she needed to start her own cleaning business, re-embedding herself while taking some of Handy’s customers with her.

Tish and Pablo are both weakly embedded in conventional economic terms, especially when judged by their labour market positions, and they have been dependent on platform earnings for large periods of time. Yet their attitude toward platform-induced risk cannot be uniformly characterised as antagonistic, resigned, or acquiescing, as Schor et al. (2023) would suggest. How they approached the risks associated with platform-based work was always related to how they estimated its opportunities in the context of other options, resources, and forms of risk at certain moments in time. In other words, the navigation of platform-induced risk is a deeply relational, dynamic, and pragmatic affair whose moral economy is informed by often affectively driven and value-laden assessments of what risks are worth taking. A sense of agency in a web of dependencies is critical in both biographies; sometimes risk taking was entrepreneurial in nature, while at other times it just felt potentially rewarding, dignifying, or simply the right thing to do. Yet risk — particularly platform-induced risk — was also experienced as agency obstructing and demoralising, especially when it was judged to be unfair or undue, which led to feelings of ambivalence.

Ambivalence and change are key themes in both biographical narratives. Yet these tend to be overlooked in gig economy scholarship that prioritises what happens at the point of production over what happens to people in other areas of their lives. Taking into account platform-adjacent work and life trajectories expands this narrow critical scope and shows how “embeddedness”, “platform-dependency”, and “risk” are mutating and multifaceted phenomena whose interaction will co-determine how much and what kind of power a platform can exercise. In both biographies, we have seen how platform power can gradually grow more entrenched, as changes in other relations of power and dependence — often involving state institutions administering welfare and taxation — rendered Pablo and Tish more vulnerable. Finally, despite the field’s preoccupation with (algorithmic) labour discipline and control, the sudden deactivation of both Pablo’s and Tish’s accounts should remind us that platforms routinely wield a type of “sovereign” power that acutely jeopardises workers’ livelihoods. Yet what the narratives also demonstrate is that, even then, work and life trajectories continue. There is a life before and after platform work, even as a platform can change one’s life and inform decisions that shape work trajectories.

6. Conclusion: Toward a platform-adjacent policy response

Given its emphasis on ambivalence, complexity, and change, it may be hard to imagine how an analysis of these biographical narratives could inform regulatory and policy initiatives seeking to improve labour conditions in the gig economy. As I have learned over the years, such initiatives usually prefer unambiguous research resulting in concrete and ideally scalable “solutions” that can be implemented across the board. Like previous work that emphasises workforce heterogeneity (Schor et al., 2020; 2023), a platform-adjacent analysis complicates this predisposition insofar as it pushes back against silver bullet approaches that too narrowly focus on platform culpability and misclassification while conceiving of platform power in universal and “algorithmic” terms. In this sense, scholarly limitations in understanding the risks and vulnerabilities that gig workers experience are partly responsible for the prevalence of shortsighted regulatory interventions.

So what does a platform-adjacent policy response look like? First, it urges regulators and policymakers to adopt a multidimensional, holistic approach that treats gig workers not just as workers but as people living complex lives full of dynamic relationships with other people as well as a variety of public and private institutions that can both empower them and deprive them of agency. Since gig platforms are often just one institutional actor among others, policymakers should start by forging a comprehensive image of the socioeconomic circumstances — i.e. the particular needs, problems, work/life trajectories, and aspirations — of those who seek out gig work, before they consider the power of platforms or algorithms. While I realise that this will require much time and effort, not to mention political will, only tailored approaches that take into account the heterogeneity of gig workforces and recognize the complexity of their struggles have a chance at lasting success.

Second, a platform-adjacent policy response should not only be holistic with respect to how it approaches the problem space of gig work; it should also be self-reflexive. This means that policymakers and regulators should acknowledge the multifarious and ambivalent role of the state in the lives of many gig workers. It should be noted here that the legal scholarship on platform-mediated gig work largely mirrors the self-conception of those making and debating new platform work legislation, insofar as both see the state as custodian of the public good entrusted with the responsibility to redress injustice and injury (Brown, 1995). However, many of my research participants, including Pablo and Tish, have more diverse and contradictory experiences with state power, as different state agencies — at various scales of government — provide sources of care and support while also inducing vulnerability by exercising forms of punitive and disciplinary power. While the nature and impacts of state power have always been manifold, contemporary neoliberal states are increasingly composed of “fragmented governance architectures” and their agendas “facilitate scattered policy interventions” co-shaped by private actors serving market interests at the expense of the poor and working class (Taşan-Kok & Özogul, 2021, p. 1318). In this environment, it is not just critical to ensure that platform policymakers are protected against corporate influence but also to create frameworks in which they can engage in collaborative and inclusive policy reforms across areas of governance.

Third and finally, a platform-adjacent policy response thus encourages a relational and integrative approach to policy-making, seeking to dismantle policy silos and involving more institutional actors that do not represent the interests of capital. As long as gig workers and their households are asked to shoulder more risk, gig platforms will continue to be attractive options for those lacking robust resources for income generation and daily risk management. Recognising this, regulators and policymakers should work together with advocacy groups and community-based organisations to find ways of ensuring that we do not allow these platforms to become people’s main source of income, since this creates dependencies that result in the entrenchment of platform power — notwithstanding the variegated impacts of such power. Serious efforts to address the problems of the gig economy should not and cannot be disconnected from broader policy reforms that address the gender pay gap and racial wealth inequalities, improve job opportunities and working conditions for labour market outsiders, and make welfare, (health)care, and educational services more widely accessible and less punitive for marginalised communities.