1 Introduction

There is growing interest in research regarding the role of migration in platform capitalism (Lata et al., 2023). Empirical investigations from across the world show that migrant labour constitutes a crucial part of the labour force working in and around platforms (Gebrial, 2022; McDonald et al., 2019; Van Doorn et al., 2020; Collins, 2020; Zhou, 2022). However, there is still little analytical engagement with questions about the role of migration in the extraction of value from social cooperation happening through and in platforms (Altenried, 2021; Gebrial, 2022; Schaupp, 2021; Dubal, 2020).

In this chapter, I interrogate the empirical data of the qualitative research conducted at the project Platform Labour in Urban Spaces (PLUS), to develop a nuanced understanding of how migration, mobilities, racism, and platforms relate to one another. What is the relationship between platforms and migration regimes? What type of labour relations do platforms shape, and how are they related to processes of racialization of the labour force? Under which forms does racism emerge in the platform economy, and what can we learn about racism, by studying platform labour? To tackle these questions, I adopt three theoretical viewpoints.

Firstly, I critically engage with the infrastructural turn (Wiig et al., 2022) and the mobility turn (Faist, 2013) and ask about the processes of “infrastructuring” (Star, 1999; Lin et al., 2014) of mobilities and migration which platforms trigger and preserve. While platforms have been considered as infrastructures of migration (Altenried, 2021; Van Doorn, 2020), I ask about the mobilities which they generate on more scales than the one concerning international migration.

Secondly, I embrace the Labour Process Theory (Gandini, 2018; Smith, 2015) to delve into the complex system of labour relations designed by platforms. I identify processes of multiplication of forms of oppression and sketch out how these relate to racism. Here, I am less interested in the politically contested question of whether platform workers should be understood as employees or as freelancers, but rather focus on the promiscuity of labour in the platform economy.

Thirdly, I investigate the impact of platforms on racism. Although it is now widely recognized that digital technologies are far from representing a “colour-blind utopia” (Hamilton, 2020), the interaction of the platformization of labour with the proliferation of racism in contemporary societies is still under-researched. While there is growing literature on the circulation of racism across social media platforms, this is less the case for other types of platforms. There is yet little engagement with a broader theorization of racism in platform capitalism (Gebrial, 2022). In this chapter, I develop the concept of infrastructural racism, to emphasize the capacity of platforms to make racism and anti-racism circulate across society in specific ways and circuits.

For reasons of length, this chapter will not focus on forms of resistance and practices of autonomy developed by workers, strategies operated by workers, although they are a fundamental component of the way platform capitalism is to be understood in relation to migration, mobility, and racism. Here, I focus on “what platforms do” (Vallas & Schor, 2020), and on the systems of domination and oppression which they tap into and/or generate.

The chapter is organized as follows: I start the analysis with a methodological note aimed at tackling the question of whether the majority of platform workers can be considered as migrant. Thereafter, I proceed to delineate the main arguments about the infrastructuring of migration and mobility, about the investigation on processes of racialization connected to platforms, and to the conceptualization of infrastructural racism. In the conclusions, I wrap up the results of the analysis and sketch out some possible trends for future research.

2 Are All Platform Workers Migrant Workers? Methodological Notes from the PLUS Project

The question if platform workers are mostly migrants is a highly challenging one. While international literature on platform work points to a large component of migrants among the workforce of platforms across the globe (McDonald et al., 2019; Anwar & Graham, 2020; Van Doorn et al., 2020; Altenried, 2021; Gebrial, 2022; Zhou, 2022), researchers face several challenges when collecting data on the socio-demographics of platform workers. The fragmentation of labour relations across and within platforms, the high turnover of the workforce, and the complexity and diversity in national statistical categorizations of migrants jeopardize the reliability of data on platform workers.

Moreover, since their onset, platform economies across the world have developed at a very fast pace, with their workforce changing as well very rapidly both in numbers and in their socio-demographic composition. The Covid pandemic sparked a phase of dramatic expansion of some sectors such as the last-mile delivery and of crisis of other sectors, such as transportation and domestic services, while we are now witnessing a phase of financial trouble and massive layoffs. At the PLUS project, interviews were mostly conducted in 2020, thus shortly before and after the outbreak of the Covid pandemic in Europe. Our research provides thus an important insight into the effects of the pandemic shock on platform economies in European markets. In the current phase of squeezing of the labour force of platforms, however, it is legitimate to expect that also the composition of the labour force might have further changed.

Nevertheless, in our data, migrant workers compose a large part of our interviewees. We carried out 230 interviews with platform workers of Deliveroo, Uber, Airbnb, and Helpling. Ca. 1/3 of the interviewees were classified as owning a nationality other than the national one, while in 7.5% of the interviews no nationality could be assigned. Noteworthy, the ratio of non-migrant platform workers was not surprisingly disproportionately high among Airbnb. Airbnb represented a liminal platform in our sample, since Airbnb hosts do not participate formally in a labour relation with the platform. In the case of Helpling, Deliveroo, and Deliveroo, in some cities migrants made up half of the interviewed workers. Moreover, migrant workers might be underrepresented in our sample for various reasons. Generally speaking, reaching out to migrant workers was often challenging. The local teams faced language barriers when trying to reach out to the migrant communities, and of course reported that those workers with the most precarious status, such as illegalized workers and asylum seekers, might avoid participating in interviews to protect themselves. However, all evidence pointed to a large presence of these groups in the local platform economies. While it is not possible to determine statistically whether the majority of platform workers are migrants, we can assert that they do constitute a central part of the labour force of platforms. Our data suggest that migration and more broadly speaking mobility are quintessential for the extraction of value enforced by platforms, as I will show in the next paragraphs.

3 Platforms as Infrastructures of Mobility

This chapter engages with the concept of infrastructure to make sense of the ubiquity of mobility that we detected in the investigation of platform labour. Infrastructures are “socio-technical platforms for mobility” (Larkin, 2013 in Lin et al., 2017); they are “matter that enable the movement of other matter” (Larkin, 2013: 329). Their analysis is therefore crucial for the understanding of the circulation and mobility of people, capital, things, and ideas. The paths and networks along which mobilities take place are not pre-given, they are constantly re-figured by transformations in societal modes of production and reproduction, technological fixes, and ideological constructs. Spanning from Critical Urban Studies to Studies on Socio-Technical Systems and Critical Migration Research, research on infrastructures has helped to disentangle these transformations. The infrastructural, logistical, and mobility turns have contributed to open the black boxes in which phenomena such as migration, logistics, extractivism, and other forms of extraction of labour value are placed, in the context of variegated neoliberalism marked by pervasive digitalization.

Platforms rely on distributed systems. They exist as data which circulate along cables, through electronic devices, and across data centres, but also exist as headquarters and offices, in which managers ideate business strategies and tech workers program the algorithms which make the data flow. And of course, platforms are also present as the beep of a smartphone which makes human bodies move, ride, and drive; they organize social interactions between workers and clients, in other words they regulate human labour. Platforms are difficult to territorialize, there is no delimited physical space in which a platform can be contained. As research has suggested, platforms can be understood as digital infrastructures (Plantin & Punathambekar, 2019; Ferrari & Graham, 2021). Due to their infrastructural power, they are able to stimulate, organize, mediate, mould, channel, and stop mobilities on various scales (Altenried et al., 2020; Stehlin et al., 2020). They are deeply interested in the extraction of value from mobile work (Gibbings et al. 2022); the mobility of things, information, money, and people is at the core of the labour process of platform labour. I argue therefore that platforms are involved in operations of “infrastructuring” of mobility across multiple scales. The concept of “infrastructuring” (Star, 1999; Lin et al., 2014; Simone, 2022) highlights the open, contradiction-ridden, generative, and processual nature of the making of infrastructure.

In this chapter, I focus on the processes of “infrastructuring” which platforms operate on the mobility of human labour across national borders (i.e., migration) and across platforms borders.

3.1 Migration and Mobility

Firstly, I want to direct the attention at the international scale of mobility, which is of course crucial if we want to make clarity on the relationships between platforms, migration, and racism. Platforms need disposable and cheap labour force. Mobile labour throughout history has largely provided such characteristics (Altenried, 2021). Enmeshed in processes of differential inclusion, migrant labour in late capitalism is often employed in the most precarious sectors of the labour markets. Migration regimes work to make migrants vulnerable and forced into working conditions which are below the common conditions of acceptability.

On their part, platforms thrive particularly in those sectors where migrant labour was already crucial before their arrival, such as gastronomy, logistics, taxi industry, and cleaning services. These sectors of urban economies on their turn perform essential functions for the reproduction of cities and their inhabitants. Platforms, however, broaden the composition of the labour force in these sectors, extending access to labour to populations exposed to very diverse migration regimes in terms of temporality and of spatiality.

On the one hand, platforms allow newly arrived migrants to work without knowing the language and local context. Hereby, there are important distinctions to be made concerning the contemporary dynamics of differential inclusion at play, as newly arrived migrants are faced with different border regimes according to their nationality. Within the EU migration space, migrants from EU countries might be better able to work for platforms for a shorter time, while they look for better jobs which match with their qualifications. Instead, migrants from third countries have to stick longer to platform labour, either because their qualifications might take longer to be recognized or because they cannot afford being unemployed for economic reasons as well as for securing their visa renewal.

On the other hand, platforms also attract into their labour pool workers belonging to second and third generations and previously employed in activities characterized by minimum wage, informality, and precariousness. In this case, platforms exploit and tap into existing processes of racialization. They portray themselves as comparatively better alternatives to non-platformized labour relations and leverage on the discourse of “freedom” and “entrepreneurship” (see the case of Uber drivers in Tallinn for a compelling example of such neoliberal appellation). This dynamic is very clear if we think of the recruiting campaigns which Uber carries out in the peripheries of European metropolises, as we have observed in Paris, or in London. In both cities, the platform openly targeted racialized and underemployed youth, with the goal of attracting them into their pool of labour force.

Out of these preliminary observations on the composition of the labour force of the investigated platforms, I start sketching the hypothesis that the labour relations designed by platforms rest on racialization processes generated outside the platform and/or prior to its arrival in the local labour market. Platforms do benefit from the disposability of labour made cheaper than the average by the combination of migration regimes and previous processes of segmentation of labour markets. They are thus able to leverage on a variety of conditions created by the interplay of specific regimes of management of flows of migration located in the present as well as in the past.

It is thus legitimate to state that migrants constitute a central component of platform labour, although it must be clear that such an argument is not to be considered as universal and not subject to change. The question of whether platforms can prosper also in labour markets where migrant labour is not accessible or not present should be tackled by empirical research. In the research carried out at PLUS, we observed that platforms attract a more heterogenous labour force, in which also non-migrant groups such as students and retired people play a non-marginal role. In times of increasing precarity and of worsening conditions for large parts of societies, other groups might find platform labour attractive. Both students and retired workers, however, are themselves mobile on the scale of the local labour markets, either because they are about to access them or to leave them. For this reason, I specify the argument, by stating that platforms attract mobile labour which in most cases tends to be also migrant labour.

In fact, platforms seem to be interested in labour which is mobile on more scales. Since they are highly dependent on extremely volatile venture capital, they must be able to expand and shrink on a very fast pace. While during the pandemic there has been a phase of impressive expansion, we are now witnessing one of financial squeezing and massive rounds of layoffs throughout the platform economies across the world. To face such volatility, platforms rest on agile business models and an extremely flexible workforce. The first distinction to be made concerns the one which separated the managers, tech, sales, and customer care workers in the offices from the vast mass of mobile workers operating on the streets. On the one hand, the tech workers producing the algorithms benefit from their scarce skills and thus enjoy relatively secure positions, but they are by far not spared from sudden layoffs. More research is needed to investigate these workers, as they are key to the prosperity of platforms, but are still part of the (elite) of the working class in these economies. On the other, the platform workers on which our project focuses constitute the most expendable component of the workforce. They are made as surplus populations within the platforms. As we could see in many cities, the platforms started hiring huge masses of workers to realize their expansion, at the same time lowering the possibilities for income of workers, who found themselves in crowds on the streets competing for a commission or a ride. To cope with such accelerations and downturns, platforms need labour prone to perform unqualified and low paid activities in exchange for a quick fix to their need of income.

3.2 Platforms as Infrastructures of Migration

In literature, a newly recurring argument states that platforms have become migration infrastructures (Altenried, 2021; Van Doorn et al., 2020). This idea rests on literature on migration brokers, intermediary agencies, and transnational networks of migration (Xiang & Lindquist, 2014; Lin et al., 2017; Meeus et al., 2019), according to which these infrastructures, while they mediate the migration of people, also actively shape these mobility flows. The research carried out at PLUS can help to clarify and specify this argument.

As the interviews with Helpling workers in Berlin show, young Latin Americans preparing to move to Europe with a Work and Holiday visa already plan to register at Helpling or Deliveroo before moving. Work and Holiday programmes allow young people between the ages of 18 and 30 to reside and work in a foreign country for the duration of 12 months, based on bilateral agreements concluded by the two countries involved. Germany, for instance, has concluded such agreements with countries such as Chile, Uruguay, Brazil, and Argentina. Many young people from these countries have taken advantage of this possibility in recent years. In our research in Berlin, we met a couple of young people from Argentina who had moved together to the city. Tomás holds Italian citizenship, Roxana only the Argentinian one. The two had married before leaving so that the girl could enjoy a more secure status, although they did not exactly know what kind of rights Roxana would enjoy in the European Union after marriage. Before leaving, they had collected information about the platforms where they would be able to work in Europe. In a first move, they had moved to France. As the calculations made on the visa to be obtained on French soil turned out to be wrong, the two then managed to obtain a Work & Holiday visa with Germany, and moved eventually to Berlin, where she started working for Helpling and he for Lieferando (a delivery platform).

People willing to migrate of course prepare for the move by exploring the possibilities for income, the expenses for housing and living in the place they might move to. They operate calculations based on uncertain variables and generalized precariousness (Gago, 2017). Very often, they use the knowledge resources circulating along the transnational networks they have access to and are thus able to make forecasts about which job possibilities are to be found. With their cosmopolitanism, multilingualism, and interchangeability, platforms appear as a more secure infrastructure to make these calculations as correct as they can be. Platforms, in other words, seem reliable to those who are about to become migrants. In fact, they contribute to making migration possible, as they regulate the valorization of mobile labour force and operate on the porosity of borders. Moreover, platforms are potentially generative of migration flows, as they open spaces of access in the places of arrival, triggering the “word of mouth” from those who already migrated to their peers still based in the homeland. They lower the entry barriers located at the borders of local market and make migration smoother. Of course, platforms do not only trigger the mobility of newcomers, but they also attract all sorts of migrant workers, and represent a last resort for those who are in urgent need of a formalized labour relationship, for instance because their visa period is about to expire. Therefore, they can operate also as infrastructures for staying “in migration”.

By acting as infrastructures of migration, platforms can create relationships of exploitation at a faster pace, including a zone of informality into their boundaries. Since the bureaucratic procedures to register for the apps are simplified, and external controls are made more difficult by the opacity of algorithmic management, workers do not have to “waste time” waiting for permissions to work but can start working straight away after they arrive. In many cities, Deliveroo, Uber, and Helpling allow for a grey zone which has been defined of “selective formalization” (Lanamäki & Tuvikene, 2022). While this type of formalization aims to codify standards to increase customer confidence and guarantee the availability of a reservoir of customers, thus securing the demand for the service, it does not necessarily formalize the rights of workers, who remain in a state of insecurity. At the same time, the relative detachment of platforms from local systems (tax, social security, but also migration regimes) leaves open doors to spaces of grey formalization, where workers can access the platform despite not having the formal requirements to work (Uber is also worth mentioning in this respect). Platform workers thus access the source of income before they can actually use it. However, the digitized spaces that exist between account creation, login, the acceptance of the first commission and the actual realization of the first gig, are black-boxed. While these informal spaces might be subjected to regulation after some time, platforms exploit the initial phase to initiate an expansion that allows them to increase their monopolistic power and to establish themselves as infrastructures. Platforms use the disruption moment after their arrival in a new city to operate in such a shady zone of non-legibility by state institutions and use this window of opportunity to scale up very quickly their pool of migrant labour. Finally, the production of working subjects which challenge the antagonism between formality and informality is a key aspect of the platform model.

3.3 Labour Mobilities Across Platform Borders: Mobility as Turnover

Another way of expanding the concept of mobility emerges if we think of workplace turnover (Andrijasevic & Sacchetto, 2016). From this perspective, migration can be thought of as a counter-practice that people enact to benefit from differentials of exploitation, and workplace turnover as well a practice of resistance against the employer. The ability to make use of the exit option has historically been connected to the Marxian concept of free labour. Authors such as Moulier Boutang have highlighted how forms of unfree labour have always co-existed to wage labour (1998). The bridling of labour by employers stands in antagonism to the autonomous practices of flight enacted by workers. On its counterpart, immobility in relation to the workplace relates to the Fordist claim against delocalization and outsourcing and for permanent employment. Under late neoliberal conditions, however, with increasing precarization and fragmentation of labour, immobility in the sense of a permanent labour relationship moves to the background, as a remainder from a past mode of production. What comes to the foreground is immobility as confinement into a labour relationship which workers would like to exit but are not able to, because of lack of alternatives.

From this perspective on labour mobility, our research has highlighted recurring patterns across platforms and cities, according to which the socio-demographics of the labour force of platforms change over time. We argue that these patterns relate to the way how platforms, acting as infrastructures of mobility, incorporate and profit from the racialized segmentation of labouring populations, redrawing the boundaries around them.

As mentioned earlier, the composition of the labour force of platforms is heterogenous. In many cases of our sample, such as Deliveroo in Bologna and Uber and Deliveroo in Paris, we observed that such composition also changed over time along a recurring pattern. While after their arrival in the city, the two platforms offered bonuses and perks to attract a pool as wide as possible of workers, after some time they would start to “tighten the belt”, for instance by switching the pay mode from hourly to piece pay or by introducing a ranking system, so that only highly performing workers might get the most profitable shifts. As platforms kept hiring workers, to cope with their expansion, commissions became scarce, thus exposing workers to increased competition among themselves and compelling them to work longer hours in order to reach a living wage. Due to the worsening of conditions, those workers able to find other jobs in other platforms or at other workplaces left the platform. As the PLUS researchers in Bologna and Paris observed, some workers, especially the ones who were working as a side-job, abandoned the platform, often starting to work as employees in the same sector. In Paris, many riders, once they left Deliveroo, started working for fairer employers, such as cooperatives. As rare as it is, they could leverage on their experience at Deliveroo and use the acquired skills in a labour relationship with better social security and higher pay. By contrast, those workers with uncertain residency status or more precarious living conditions maintained their occupation at Deliveroo and Uber, having to accept the intensification of working hours and heightened pressure in their everyday work life.

In Paris and Bologna, such differential exclusion from the platform translated into a “migrantization” of the labour force, with nationals moving to better opportunities outside of the boundaries of the platforms. When platforms started “tightening the belt”, processes of stratification along racialized lines emerged, impacting not only on the labour force within, but also outside the platforms. Here it becomes clear how processes of racialization of the labour force cannot be contained within the boundaries of the platforms but connect the insides and the outsides of platforms. As the mobilities of labour out of and into the platforms themselves are inherently connected to the immobilization of labour due to the interplay of migration and labour regimes.

4 Platformization of Labour Relations and Racialization Processes

Our research, in line with literature on platform work, reveals that platform workers in European cities are exposed to several forms of oppression related to the labour process they engage with. These include the exploitation of their labour force (Vallas & Schor, 2020), increased control of the mobility of bodies through algorithmic management (Animento et al., 2017; Muldoon & Raekstad, 2022), and high vulnerability in their relationship with clients because of the non-liability of the latter regarding ratings (Rosenblat et al., 2017). Labour process refers to the “conversion movement that transform labour power in a commodity” (Gandini, 2018: 1040). In the case of platform labour, the conversion can take place through the labour contract or the terms & conditions which respectively platform employees and freelancers sign when they enter in relation with the platforms. Regardless of the status of platform workers, the labour process theory can be fruitfully applied to platform labour to “unpack the relationship between the employer, who owns the means of production, and the worker as the possessor of labour power” (ebd. 1043). This relationship is understood as antagonistic, with the employer trying to maximize profit out of labour force, and the worker aiming at reducing her/his efforts in the labour process. The service economy, however, which the platforms in our sample belong to, is typically characterized by a more complex, often triadic power relation (Lopez, 2010), as it rests on relations of power between workers, managers, and customers. The relationship between workers and managers might resemble the classical relation between labour and capital in the context of industrial capitalism, while the relationship between workers and clients entails specific post-Fordist traits, for instance regarding the centrality of emotional labour (Hochschild, 1983).

In platform work, the labour relations are even more complicated, as platforms are not the only intermediaries. In fact, our research shows a proliferation of actors and stakeholders in the local platform economies. From the viewpoint of Labour Process Theory, most workers of the gig economy are faced with multiple forms of oppression, beyond the triadic relation with managers (the platform) and customers. Our research revealed that in cities such as Berlin, Lisbon, and Paris, in the local sectors of ride-hailing the triadic power relationship is further complicated. Due to the local regulatory frameworks, subcontracting companies must mediate between Uber and drivers. These companies, called “Uber partners”, employ drivers through formal labour contracts. This business model, which is becoming predominant across various sectors of the platform economy (see Niebler et al. forthcoming), guarantees the workers with some social benefits attached to the labour contracts, but at the same time puts them in an augmented relationship of exploitation (Animento et al. forthcoming). Finally, a large part of platform workers adopts strategies of “multi-apping”, i.e., uses simultaneously various apps, in order to maximize income. Given the very strict control and command of the labour process via algorithmic management, it can be argued that from the point of view of labour process, platform drivers doing “multi-apping” are substantially facing several employers at one time. The multiplication of bosses (including their own self, of course, see Purcell & Brook, 2022) weakens the capacity of organizing and puts platform workers in a structurally weak bargaining position.

The majority of workers involved in these complex systems of labour relations are migrants. From the restaurants where food to be delivered is cooked, to most of the subcontractors interviewed, from the companies providing cleaning services to professional Airbnb hosts to the traditional taxi drivers occasionally using apps, platforms mobilize whole ecosystems in which migrant and mobile labour are pervasive. However, workers performing different functions within these ecosystems are often stratified along their status and background.

In our research, many of the employers of subcontracting companies working for Uber were migrants themselves. However, in comparison to Uber drivers they enjoy an economically more secure position, also since they often belong to the second or third generation. Thanks to earlier processes of (albeit precarious) upward social mobility, these migrant entrepreneurs could accumulate resources to establish a company. However, small-scale entrepreneurs working as “partners” with Uber and other platforms—recently also delivery platforms have started to adopt the subcontracting model—reported that their businesses are far from being stable and secured, as they face high fixed costs (such as the leasing of the cars, fuel, etc.) and increasing competition and saturation of the market, due to the proliferation of platforms. Thus, “partner”-employers find themselves in an intermediate position, having to deal with economic actors which are far more powerful, and with platform workers, who certainly are in a relatively more precarious position, often as migrants of first generation, asylum seekers as well as migrants with uncertain status.

A further aspect related to how platforms affect labour relations is connected to the topic of skills and more specifically professionalism. As the research shows, platform workers mostly do not have any possibilities to scale up their positions, accumulate skills, and push for upward mobility. Their jobs are not integrated into a corporate ladder; in fact, such a ladder simply does not exist. The chances for bettering their condition within the companies are connected to internal ranking systems, which based on untransparent algorithmic calculations allocate gigs differently to workers. At Airbnb, hosts can achieve the badge of Super Hosts, which is endowed with some extra perks. However, their conditions do not change qualitatively; they are confined to their position.

If we consider such processes of deskilling in relation to the sectors in which the platforms operate, however, we can add a further specification. The case of the taxi industry is crucial here, as across our sample we observe that drivers working with apps such as Uber are considered less skilful, unable to drive, and not prepared for the job, in comparison to “proper” taxi drivers. Often, the lack of knowledge of the local language (and of the English language in some cases) and of the topography of the city are mentioned by actors of the industry to disqualify Uber drivers as not suitable for the job. Of course, these processes of differentiation are enmeshed with regulatory frameworks which have allowed for a segmentation of the sector, in which in many cases taxi drivers are required to fulfil a set of standards while app drivers are not. The app, which is multilingual and facilitates the use of GPS systems, allows the newly arrived migrants to start straight away to work.

On the one hand, the platform acts as infrastructure of migration and of mobility, providing a chance to work to those who are otherwise marginalized in the local labour markets. On the other hand, however, it operates through differential inclusion, integrating the workers into a niche of the labour market which offers no chances for betterment. These processes of extraction of value from labour force artificially made expendable can be defined as enacting “predatory inclusion” (McMillan Cottom, 2020: 443). As demonstrated by the recruiting campaigns of Uber in the Parisian banlieues, platforms portray themselves as actors able to activate those populations failed by state policies of integration, at the same time as they extract surplus from processes of racialization.

Further, platforms such as Uber separate the labour force, drawing in those workers who end up doing the same job as workers outside of the platform, albeit under much worse conditions. In all cities of our sample, Uber drivers are structurally located below the taxi drivers, in terms of security, income, and social prestige. However, as many interviews with taxi drivers have shown, app and non-app drivers do carry out the same work tasks; their labour processes have much more similarities than differences. The traditional taxi driver industry has undergone processes of digitalization which are not so different from the ones imposed by platforms.

Summarizing, platforms enter labour markets triggering two movements: expanding the labour force by including under-labouring populations, and then dividing these populations, allocating them to different positions in terms of power, prestige, income, and security. By doing these, platforms allow racialization processes located outside their boundaries to enter at the core of their operations. The acknowledgement of the complexities and intricacies of platform labour allows to understand processes of stratification of labour within and at the borders of platforms. Platforms seem to operate as multiplicators of inconsistencies regarding labour relations, dethroning wage labour as the form of labour relations most ambitioned by workers (Gago, 2017). Platforms make working conditions heterogeneous and more diverse. At the same time, such diversity is not casually distributed across workers, but rather follows patterns of racialized (and gendered) divisions of labour. Often, the physical activities involved in the labour process remain just the same as before the platformization, but the workers performing these same activities are exposed to different labour regimes. In other words, platforms redraw boundaries around working populations, shifting the lines between these and surplus populations, shifting working subjects from one sector to the other, and from one type of labour relation to the other.

5 Infrastructural Racism

After demonstrating that platforms can act as infrastructure of mobility on several scales, as well as operate on labour markets by expanding and dividing the labour force, thereby incorporating external processes of racialization, we ask now directly how platforms interact with racism.

We showed that forms of racism intersect with trajectories of labour into, within, and out of platforms, but we now want to ask for the specificities of the forms of racism which platforms mobilize and if they potentially generate new ones. To do this, it seems necessary to locate the analysis in relation to the lively debates about racial capitalism, which have in the past years brought racism back to the fore of critiques of capitalism (Fraser, 2016; Gebrial, 2022; Rana, 2016; Melamed, 2015). As a system of social relations based on hierarchies and differences, racism feeds capitalism by providing legitimation for differential degrees of exploitation, laying the ground for expropriation, dispossession, and other forms of extraordinary extractivism. While the question about the contingency or necessity of racism in capitalism will not be asked here (see Conroy, 2022), I intend to provide an empirically grounded proposal for conceptualizing racism within the platform economy. Racism is a persistent aspect across the data material produced by our research, whether in the forms which are usually defined as systemic, structural, institutional, or interpersonal. Across platforms and cities, we witness an “everywhereness” of racism, which cannot be left unaccounted for.

In the previous paragraphs I already mentioned several examples. For instance, we notice structural and institutional racism when migrant workers are relegated to the semi-formalized sphere of platform labour, because they have very limited possibilities of working in more regulated niches of the labour market. They are prevented access to standard labour because of lack of knowledge of national language, or because they have qualifications which are not recognized. But even within the non-standard labour relationships generated by platforms, they tend to be stratified along racial lines. Further, our research discovered also forms of institutional racism, although they are rarer, such as in London, where Uber singularly introduced a language test for drivers.

Secondly, our research revealed the emergence of black markets in which accounts for various platforms are shared, sold, and exchanged. This hints to the proliferation of illegalized economies around the market for accounts/avatars. Here, the unpresentable workers (illegalized migrants, migrants at the margins, sans-papier) are made disposable for work via a further downward stratification, in which they also run the risks of being policed and sanctioned, while they must buy their access to a “safe” account. Again, structural racism is at work here, legitimizing the practices of hyper-exploitation towards groups made vulnerable by a migration regime based on non-recognition and deportability. Further, the stratification along generations of migrants which we described earlier can be understood as a process by which the platforms benefit and leverage on the sedimentation of different migration regimes located in different times.

Finally, in most cities of our platforms, the interviewees reported that they had faced verbal aggressions, physical attacks, and discriminations because of their skin colour or origin. At Deliveroo, riders mentioned that they had been targeted by clients or restaurant personnel treating them badly. At Uber, drivers reported to be assaulted or harassed by clients, as well as by taxi drivers. In some cases, the latter have organized public campaigns against Uber drivers, including campaigns portraying them as perpetrators of sexualized violence, as in London (Gebrial, 2022), and car demonstrations during which Uber drivers were physically assaulted, as in Berlin. Platform workers are thus exposed to interpersonal forms of racism, but what is even more crucial is that they have very low chances to protect themselves from such assaults, except for calling the police. The platforms do not provide them with a security network which they can mobilize in case of emergency; the hotlines are not designed to cope with such cases. Further, most apps do not provide the workers with a rating system for the clients.

Recent literature has focused on the relationships between digital technologies and racism (Hamilton, 2020; McMillan Cottom, 2020; Nakamura, 2009), and in particular between platform capitalism and racism (Gebrial, 2022; Matamoros-Fernández, 2017). Many authors have empirically demonstrated that the idea that technology could allow for the rise of a “colour-blind utopia” is flawed (Hamilton, 2020). Digital technologies reproduce existent racialized inequalities even if they propagate an image of themselves as neutral and benevolent (Benjamin, 2019; Noble, 2018). Algorithms are designed and fed by data generated by human beings, and as such they reproduce the patterns of discrimination and inequality to be found in society.

Platforms operate on labour markets and societies already profoundly shaped by race relations; they profit from racism as a crucial organizing principle of social relations. However, there might be more to be explained when addressing the question about how platforms relate to racism. Here I want to ask whether not only labour relations in platforms are organized by racism, but also whether platforms themselves are generative of new forms of racism, and how they organize them.

As infrastructures, platforms allow for the circulation of people, things, and money, and of ideas. If platforms engage in processes of “infrastructuring” of migration and mobility, then it is legitimate to ask whether and how they are also enmeshed in the “infrastructuring” of racism. I propose to use the concept of infrastructural racism, to explain those forms of racism which are mediated by platforms and digital infrastructures. Structural and systemic racism refers to wider societal structures, institutional racism is placed at the level of institutions, and interpersonal racism is to be found in the vis-à-vis relations between people. Additionally, infrastructural racism allows to grasp the circulation and mobility of “racist scripts” (Molina, 2014) across socio-technical platforms, at the nexus between digitalized and offline social relations.

The example which I want to bring here relates to the rating system, which clients use to rate workers after the commission is completed. Of course, literature has already shown that platform rating is open for discrimination along race, gender, and many other markers (Rosenblat et al., 2017; Vallas & Schor, 2020). Clients are not required any evidence for justifying a bad rate, a verification is not integrated in the app. On the other part, instead, workers can rate clients on some platforms, notably Airbnb, but bad ratings do not impact in the access to income of platform users, it may simply make them less desirable as clients. As commissions are made scarce, of course, rating clients is not sufficient to prevent violent or discriminating behaviour by clients. Interestingly, Helpling workers in Berlin managed to create an external non-algorithmic system of rating in the form of blacklists which they share in WhatsApp groups. In many other cases, however, this is not possible, given the huge number of users.

At Uber, many drivers across the cities of our sample reported having received bad ratings which in their views were not related to their performance as drivers. Even if they felt that the ratings might have to do with their migrant origin, the colour of their skin or their gender (and of course with all these markers together), they had no possibilities to prove their intuition, and were left with a sense of non-commensurability. Rating systems are designed to build trust into the relationship between the platform users and the platform, they are not meant to bring equity into the relationship between users and workers, or between workers and platform. Nevertheless, they constitute an unappealable verdict which has the power to lower the income of workers with no intermediation other than the algorithmic one, which is not accountable, and which automatically located workers into a worse position when their rate average decreases. To cope with this, drivers develop very refined techniques to guess how they have to relate to any specific client. For instance, they become sensitive to whether the clients are in a good or bad mood, whether it is better to chat or to keep quiet. Bad ratings are often the result of a failed guess about the client, or the inability to stick to the intuition about the emotional level to be addressed.

An interesting case stemming from our interviews with Uber drivers in Berlin helps to conceptualize infrastructural racism in our research. A driver with Afghan background, who came to Berlin as a refugee, reported that he had received a bad rate by a client whom he identified as a Mexican businessman. During the ride, the driver had made a negative comment about Arabic people and about refugees, and the client gave a bad rate and even filed a complaint at the platform hotline. In the sample, we had another example of a German female driver, who also was given a bad rate for a racist comment that she had dropped during a ride. What these examples show is that rating systems can be used to police racist behaviour or speech on the side of platform workers, but it cannot work the other way around. In other words, rating systems can be turned into asymmetric sanctioning systems against interactional racism. Workers cannot do anything to defend themselves against racism, but they are exposed to randomize ratings. What is more important, rating has direct consequences on their living condition, it can affect their capacity to make ends meet.

These examples show how racism and anti-racism circulate through the platforms via rating, as well as via the other mechanisms which we have mentioned above. Thanks to their capacity to connect, platforms contribute to make racism infrastructural, with knowledges and practices circulating along the system of labour relations and circuits of value which they generate. They enable the encounter between disparate populations through their algorithmic management; tape into various forms of racism taking place in society and re-organize them. As we have seen, platforms are designed to allow this, as they structurally put their workers into the weakest and most oppressed position of the labour relations which they design.

Platforms do organize and mediate these forms of racisms, they let some forms take place and do nothing, but they act on other forms of racism, so they perform a function of policing, which unequally affects their workers. By doing this, they can affirm themselves as colour-blind moral authorities. In fact, in many interviews, workers formulated antithetical arguments towards the platforms they worked for. On the one hand, they have practical knowledge of the multiple oppressions that they are faced with because of their work, but on the other hand they recursively referred to the platforms as being colour-blind and neutral employers who had whatsoever interest in perpetrating racism or discrimination. In fact, the approach of platforms towards diversity and inclusion can be appreciated as even anti-racist, while at the same time this is an open strategy to attract those marginalized groups who are impeded access to less precarious areas of the local labour markets, as it is so well documented by the recruiting campaigns by Uber and Deliveroo in the Parisian banlieues.

6 Conclusion

This chapter has taken mobility, migration, and racism as viewpoints on platform capitalism. Based on the data produced by our qualitative research on platform workers of Uber, Deliveroo, Helpling, and Airbnb in seven European cities, the analysis asked about the intertwining between platformized extraction of labour value and migration regimes, management of mobility, and racialization of the labour market.

Firstly, I focused on the concepts of infrastructure and “infrastructuring”, to make sense of the ubiquity of mobility in the platform economy. Platforms operate on mobilities on several scales. My analysis focuses the attention on flows of labour across national borders and across platform boundaries. Platforms offer themselves as reliable employers on which migrants can count for a smoother arrival; by opening zones of semi-formality at their borders, they allow migrant labour otherwise underutilized by capital to be put to work. If we think of workplace turnover as a form of mobility, platforms adjust their business models to create a pool of mobile labour force, which they can increase or reduce at a very fast pace. As the comparative analysis showed, platforms initially offer better working conditions than those which can be found outside of their boundaries. They provide perks and bonuses. However, after the initial phase, they start cutting benefits and reducing wages, leading the better workers pour out of their boundaries, while those disadvantaged workers, often with precarious residency status and targeted by multiple precarity (Birke, 2022), must linger on.

Secondly, I asked about the types of labour relations shaped by platforms. Taking the perspective of Labour Process Theory, I showed how the multiplication of labour triggered by platforms also leads to promiscuous and augmented labour relations, as platform workers face several forms of oppression in which the wage relation is only a component. Their relationship with customers is regulated by the rating system, while their labour process is controlled and coordinated by multiple forms of algorithmic management, especially if they engage in “multi-apping”. Platforms thus complicate the relationships of power and control in which workers are placed. As wage labour is "contaminated” by the juxtaposition of forms of unfree labour, informal exchange of accounts, fake self-employment, entrepreneurialism of the working poor, the composition of the labour force across the whole platform economy becomes increasingly migrant. Platforms expand the labour force by including racialized under-labouring populations, and then allocate them to different positions in terms of power, prestige, income, and security, leveraging on societal processes of racialization.

Finally, the chapter explores the impact of platform capitalism on racism. After demonstrating that platforms benefit from already existing forms of racialization and hierarchization of the labour force, I passed to examine whether platforms themselves are generative of new forms of racism. Aiming the attention at the rating systems, I suggest that platforms contribute to make racism infrastructural. They allow the circulation of “racist scripts” (Molina, 2014) along their infrastructures, they enable interactions between socially distant populations, and police racist behaviour in ways which affect platform workers differently than platform users. At the same time, platforms portray themselves as colour-blind advocates and strongholds of diversity. They offer themselves as moral authorities, as infrastructures which can impact the offline world and make it more equal.

The chapter offers an insight into the empirical data collected with platform workers in seven European cities. The interviews were carried out shortly before and after the beginning of the Covid pandemic in Europe, during a phase of enormous expansion of platforms and of last-mile logistics. As capitalism approaches a new phase of withdrawal of capital from unstable sectors, we witness massive rounds of layoffs of both tech and platform workers across platform economies. Having set the ground for a more systematic and nuanced analysis of how platform capitalism relates to migration, mobilities, and racism, this chapter calls for more critical research engaged in locating the analysis of platforms within the broader context of capitalist re-configurations. The current phase of the platform economy will require new investigations, aimed at studying the ways how such restructuring of the sector might affect labour, race, and gender relations, as well as migration paths and flows of mobility out of and into the boundaries of platforms.