‘Misbehaving for Deliveroo. How couriers’ digital manipulation boosts the platform’s business’

Abstract

Workplace misbehaviour has always been troublesome for employers. Workers have often exhibited conducts, such as absenteeism, pilferage, soldiering, sabotage and vandalism, that are foreign to companies’ norms and detrimental to their profits. Hence, various managerial theories have been developed over time with the aim of eliminating them. However, misbehaviour is still observed in the contemporary workplace. This study shows that, in platform work, where workers are controlled remotely by algorithms, platforms no longer seek to eliminate it, but even encourage its proliferation. Notably, it is argued that algorithmic control stirs the generation of specific forms of misbehaviour that are consistent with companies’ interests. Taking Deliveroo in Ireland and Italy as a case study, the paper illustrates how four types of manipulation of the platform’s digital infrastructure performed by couriers (i.e. the utilization of multiple Deliveroo accounts, the utilization of bots, the rental of other people’s accounts, the tricking of Deliveroo’s accounts system) are central to the expansion of its network effects. Building on labour process theory, these acts are referred to as ‘compliant misbehaviour’ – encompassing a set of worker misconducts that violate company-specific norms, but whose effects are fully beneficial to platforms’ economic interests. To develop the compliant misbehaviour concept, the author has carried out qualitative research fieldwork (indepth interviews, daily observation of couriers’ activities, document analysis) in both Ireland and Italy between 2022 and 2023. The theoretical elaboration of this concept is the main contribution of the paper.

Keywords

Algorithmic control compliant misbehaviour Deliveroo digital manipulation food delivery couriers labour process theory (LPT)network effects platforms’ business

Introduction

People do many things at work (for personal gain) that they should not do. Scholarly literature calls these deviations from expected standards of behaviour ‘organizational misbehaviour’, namely, ‘anything that you do or think at work that you are not supposed to’ (Ackroyd and Thompson, 2022: 4). Employees often exhibit conducts, such as absenteeism, pilferage, soldiering, sabotage and vandalism, that are foreign to appropriate organizational norms and, more importantly, detrimental to the firm’s profits. Both organizational behaviour (OB) specialists and line managers have continually tried to correct these behaviours.

However – as documented by labour process theory (LPT) authors – misbehaviour is a permanent feature of the organizational life of any company, as it derives from the problem of labour indeterminacy and employers’ related incapacity to fully control their employees (Braverman, 1974). Workers perceive that their interests do not coincide with those of their employers. For these authors, misbehaviour is thus a rational, not a dysfunctional, response given by employees to management to cope with supervisory issues related to their work. Not by chance, it is now widely acknowledged that misbehaviour may only be tractable and not corrigible (Collinson and Ackroyd, 2006), meaning that some change can be induced but no permanent correction can be achieved.

This study shows that, in platform work, where workers are controlled remotely by algorithms, various forms of misconduct are not only neither treated nor corrected, but widely tolerated by platforms, especially if the misconduct does not challenge their core business – namely, expanding their network effects (Cini, 2023; Rosenblat and Stark, 2016; Shapiro, 2018) – although violating some of their policies. The existence of this tolerance is not, of course, completely new, nor is its analysis a totally unknown phenomenon in the scholarly literature. Indeed, classical labour studies had already observed how not all employees’ (mis)behaviour is anti-business or anti-company profits (Roy, 1959): misconducts in the form of (unofficial) work-game practices, interpreted as essential to generate and ensure worker consent to the factory system, have been identified as a factor of productivity increase (Burawoy, 1979).

However, in platform work, where algorithms operate as the new form of managerial direction, this process has gone a step further. Not only do algorithms – it is argued – not aim to eliminate workforce misbehaviour, but they also lead workers to invent new and ever-changing forms, whose unfolding is consistent with companies’ business. This feature makes platforms not only tolerate, but even encourage, misbehaviour. Algorithms and their decision mechanisms are perceived as opaque and continuously evolving (Heiland, 2023), rendering them completely inscrutable to the user side. To cope with this opaqueness, workers perform several acts of manipulation of platforms’ digital infrastructure – namely, acts performed ‘to increase their chances of gaining and maintaining access, visibility and reputation on the platform’ (Bucher et al., 2021: 52) – that breach specific company norms but are nonetheless essential for workers to earn a decent income. More importantly, workers therefore spend more time working and sharing key information (e.g. location, mobility) on the apps (Chen and Sun, 2020), with the main consequence of boosting platforms’ network effects. Ironically, the wider the spectrum of such ‘productive’ misbehaviours, the more workers are compliant with platforms’ business.

The study accomplishes two research objectives. It seeks to (1) identify these forms of misbehaviour and (2) illustrate how they contribute to boosting platforms’ economic interests. The focus is on the digital sector of food delivery – and specifically on Deliveroo – which can be considered as a crucial case to empirically prove this argument, given its full algorithmic control over its workforce (Cant, 2020; Heiland, 2023). Accordingly, if misbehaving for such company is a relatively widespread practice, then the proliferation of these misconducts can be assumed to derive from algorithmic management.

The paper explores Deliveroo couriers’ misbehaviour in Ireland and Italy. In both countries, it is shown how various acts of manipulation of Deliveroo’s digital infrastructure, such as utilizing multiple Deliveroo accounts, using boost applications (bots), renting other people’s accounts, and tricking Deliveroo’s accounts system, defy company rules but also expand its network effects. Given their similar beneficial repercussions on the platform, they have been grouped under the newly formed concept of ‘compliant misbehaviour’. Notably, this concept encompasses a set of acts of digital manipulation that are carried out by couriers, infringing Deliveroo’s terms of service, but whose effects align with Deliveroo’s economic interests. To build it, the study draws on classical LPT work (Ackroyd and Thompson, 1999, 2022; Burawoy, 1979) and on later labour and management studies that examine platform work from various approaches (Bucher et al., 2021; Heiland, 2021, 2023; Meijerink and Bondarouk, 2023; Veen et al., 2019).

By introducing the compliant misbehaviour concept to analyse platforms’ workforce control, this paper makes two contributions to labour studies, one to the resistance debate in platform work (Woodcock, 2021), the other to the classical theory of organizational misbehaviour (Ackroyd and Thompson, 1999, 2022). First, this concept helps delimit the notion of algorithmic resistance. Not all forms of manipulation violating platforms’ norms can be considered acts of defiance: some of them can be beneficial to workers’ individual interests (income earning) and platforms’ economic interests (network effects expansion) simultaneously. The acts of resistance fall into the (small) subset of practices that do not directly foster the latter. Hence, the compliant misbehaviour concept aims to enrich the resistance debate in platform work by providing a narrower definition of what algorithmic contestation means. Its adoption helps counter some naïve interpretations proliferated in recent scholarly debate, for which workers’ digital manipulation is by default (micro) resistance (Newlands, 2021; Treré and Bonini, 2022).

Secondly, this reading of (compliant) misbehaviour contributes to broadening and refining Ackroyd and Thompson’s (1999, 2022) concept of organizational misbehaviour, which views misbehaviour as the set of misconducts that employees enact at work to increase their autonomy from managerial control and obtain some relief from the work effort. In contrast, the paper shows that, under algorithmic pressure, workers are prone to misbehave vis-à-vis the platform not to increase their autonomy from algorithmic control and obtain some relief, but rather to further attract it and intensify their work commitment. In brief, such misbehaviour seems fully compliant with employers’ interests.

Literature review and theoretical framework

The persistence of organizational misbehaviour

Employees’ misbehaviour has always been troublesome for human resource management (HRM) professionals and OB scholars. Assuming (normatively) misbehaviour’s costly dysfunctionality for companies’ interests, various managerial theories, accompanied by either organizational or technological innovation in the workplace, have been developed with the aim of correcting, or even eliminating, such behaviour (for a review, see Ackroyd and Thompson, 2022: Ch. 2). In these accounts, the root of misconduct was associated primarily with employees’ poor motivation or little understanding of the labour process: once the problem was ‘properly’ treated – the argument held – misbehaviour would fade away and the organization could regain its functional efficiency (Collinson and Ackroyd, 2006).

However – as LPT scholars have convincingly shown – misbehaviour still occurs in the contemporary workplace, as its manifestation is intrinsic to the capitalist labour process. For Ackroyd and Thompson (1999, 2022), misbehaviour is endemic to this process, as it derives from a permanent mismatch between (managerial) direction and (worker) response.

Only when there is a precise correspondence between directive and responsive aspects of organizational behaviour [Ackroyd and Thompson argue (1999: 11)] will there be no perceived problems concerning behaviour. When direction—what is expected of people—precisely corresponds with response—what people are willing to do—there is no perception of organizational misbehaviour. Since there is often not a precise correspondence, misbehaviour is properly regarded as endemic to organizations.

Such correspondence is a structurally impossible condition to achieve within the capitalist organization – their argument goes – because of the problem of labour indeterminacy (Braverman, 1974): the purchase of workers’ labour-power (‘what is expected of people’) and control over workers’ executed labour (‘what people are willing to do’) are two distinct, and not overlapping, management prerogatives. Notably, people’s generic capacity to work (what employers buy) cannot be equated to their actual work effort (how this capacity is concretely realized at work); the latter can never be fully controlled by management. Hence, misbehaviour should be considered a persistent feature of the capitalist workplace.

Understanding misbehaviour in platform work

Unsurprisingly, recent studies have accounted for misbehaviour’s proliferation in the platform work context (Mendonça et al., 2022; Polkowska and Mika, 2022; Reid-Musson et al., 2020). In line with the LPT tradition, these studies interpret the rise of such behaviour as workers’ rational response to their companies to cope with the new algorithm-led management. These workers are in fact incentivized to adopt various forms of illicit, if not illegal, ‘anticipatory compliance’ (Bucher et al., 2021), expressed in individual acts of digital manipulation, to pacify the algorithm and its opaque decisions.

Algorithms operate as a control mechanism enabling companies to monitor workers’ activity and performance remotely to increase operational efficiency (Griesbach et al., 2019). They enable platforms to track everything that workers do, but their perceived opacity limits workers’ understanding of platforms’ strategies (Heiland, 2023). Workers experience algorithmic management as particularly uncertain: managerial rules are ambiguous and change continually (Grohmann et al., 2022; Meijerink and Bondarouk, 2023). Because of their incapacity to foresee these transformations, workers are often subject to blocking, not receiving payment, deactivation, among other issues. Mendonça and Kougiannou (2022: 17) have, for instance, observed how, in the food delivery sector, workers ‘experience the difficulties with opaque intraplatform algorithmic changes’: platforms may suddenly change the norms concerning the work shift allocation or login system without informing their users. Workers are left to themselves to meet what they perceive as the algorithm’s constantly changing wishes.

Most platforms rely on economic nudges (Rosenblat and Stark, 2016), in the form of algorithmically determined surge-pricing (Gandini, 2019), to entice workers to areas of high demand. Algorithms incentivize specific gamic elements and institutionalized nudges to exercise soft control on workers’ motivation and behaviour, impacting their decisions and ultimately their performance outcomes (Wood et al., 2019). However, platforms’ performance systems are black boxes to workers, with many under the impression that accounts will be automatically deactivated if performance falls below certain thresholds (Veen et al., 2019). Therefore, as Vieira (2023: 13) observed in his study on Glovo couriers in Spain, ‘to please the algorithm, one needs to prove to be better than everyone else in working all the peak hours, pleasing partners and customers, showing up on time . . . and delivering more orders than all others’. Workers end up receiving pre-determined incentives for achieving targets curated by these platforms in the first place (McDonnell et al., 2021). Yet, in striving to meet such targets, they perform a wide range of acts of manipulation of platforms’ digital infrastructure that also expand the latter’s network effects (Reid-Musson et al., 2020).

In the present study, it is argued that the dynamic interaction between platforms and workers, spurred, on the one hand, by the opacity of the algorithm, but enacted, on the other, by workers’ coping tactics, triggers the proliferation of various forms of digital manipulation that align with platforms’ economic interests. More notably, platform workers’ individual agency appears to be shaped by an algorithmic organization purposely designed to spur their visibility on the app. In striving to enhance such visibility, workers end up imbricated in a technologically sophisticated form of management enabling them to play with the algorithm but preventing them from challenging platforms’ overall operational logic (Meijerink and Bondarouk, 2023).

Building on Ackroyd and Thompson’s concept

To fully grasp this phenomenon, Ackroyd and Thompson’s (1999), (2022 misbehaviour concept seems to offer the most suitable framework, as it features two characteristics central to these digital manipulation acts: (i) its origin in the management/worker dialectic; (ii) its non-identification with resistance. First, these authors understand misbehaviour as inherent in the dialectic of control between managers and workers, arguing that workplace misconduct arises from ‘a lack of correspondence between [managerial] direction and [worker] responses’ (Ackroyd and Thompson, 1999: 12). In brief, misbehaviour is everything that should not be done at work. Second, although deriving from the management/workers dialectic, it is not to be conflated with resistance, which refers ‘to intentional, upwardly directed action against employers or their agents (actual or inferred) in pursuit of perceived interests or identities’ (Ackroyd and Thompson, 2022: 251). In sum, misbehaviour is – for them – any act executed outside the managerial radar or not compliant with workplace discipline, which is not resistance.

Platform workers’ acts of digital manipulation seem particularly aligned with this concept, as they similarly share its two key characteristics. First, these acts arise out of the algorithm/workers interplay (dialectic of control); second, they do not directly contest platforms or their logic of operation (non-resistance). Hence, Ackroyd and Thompson’s framework appears particularly suited to account for their manifestation.

The two authors operationalize misbehaviour by distinguishing four analytically distinct areas of contention between management and workers, from which various practices of misbehaviour are assumed to arise. Implied in this framework is the proposition that management and workers constitute the major groups of actors in the workplace and that they are engaged in contention over specific material and symbolic resources. Such practices may derive from contention over (i) time, (ii) work, (iii) product and (iv) identity. In conceptualizing this contention, Ackroyd and Thompson use the term appropriation as a way of signalling workers’ active non-compliance.

Appropriation of time means the appropriation of the time spent or devoted to work: for example, absenteeism.

Appropriation of work means the appropriation of the work effort: for example, sabotage.

Appropriation of product means the appropriation of the materials and resources used and produced: for example, pilferage.

Appropriation of identity means the appropriation of employee identity: for example, sarcasm, identity exchange.

In sum, adopting Ackroyd and Thompson’s framework helps provide a detailed account of couriers’ acts of digital manipulation, viewing them as the outcome of the dialectic between algorithmic direction and worker response over four areas of contention (i.e. time, work, product, and identity), not leading to any resistance.

Data and methodology

This study’s primary aim, consistent with a build-theory-from-case-study type of approach (Eisenhardt, 1989), is to make sense of platform workers’ acts of digital manipulation in the food delivery sector. An abductive approach seems particularly suited for cases like this, in which an ex-ante knowledge of the phenomenon investigated is missing; here, extant theories are not probed; rather, specific theoretical propositions are iteratively built from examination of the case. In this approach, ‘hypotheses are constantly revised during the research until they hold true for all the evidence concerning the phenomena under study’ (Corbin and Strauss, 1990: 11). This iterative process enabled to develop a category of misbehaviour capable of identifying various acts of digital manipulation, which – because of their alignment with platforms’ business – have been called compliant misbehaviour. The hypothesis is that the inducement of such acts is a key feature of digital platforms’ algorithm-based labour process.

Case selection

The food delivery platform considered here is the British corporation Deliveroo, which – according to the scholarly literature (Cant, 2020) – is a case in point of a company adopting full algorithmic control in all its operating countries. According to these accounts, Deliveroo ‘uses a super smart algorithm made up of machine-learning technology and a powerful predictive technology that is able to efficiently distribute orders in a self-learning manner’ (Heiland, 2023: 3). Thus, investigating this company allows to better focus on the relevance of algorithm-based management in the co-creation of platform workers’ acts of digital manipulation. Put otherwise, examining Deliveroo enables to assess the extent to which this phenomenon is inherently related to its algorithm-based labour process, operating as a transnational workforce control system (Cant, 2020).

Given the transnational scope of this process, a country case comparison based upon the traditional features of industrial relations systems seemed particularly inappropriate. This gave me a certain discretion in their selection. Hence, I decided to use these cases as ‘vantage points’ (Bieler, 2021), namely, I considered those countries with which I was most familiar in relation to various critical aspects of the phenomenon investigated, such as the social context, Deliveroo’s presence on the labour market, workforce composition, couriers’ working practices, and their mobilization dynamics. This choice allowed me to leverage best my research expertise and maximize my knowledge on the conditions leading to the rise and proliferation of couriers’ misbehaviour. Consequently, I focused on Ireland and Italy, the countries in which I conducted a 2021/2023 European research project – ‘COntesting GOvernance by NUmbers: The Mobilizations of Food Delivery Couriers across Europe in Time of the Pandemic (COGONU)’ – on the impact of algorithmic control on food delivery couriers’ working experience and resistance capacity.

Ireland and Italy are also worth exploring because of their diversity in terms of institutional and political initiatives on platform work. Whereas in Ireland no political measures have been adopted, in Italy several initiatives have been implemented at both local and national level. Among the latter, it is worth mentioning various labour ministers’ attempts to repeatedly encourage formalized bargaining between the courier collectives, food delivery platforms, and trade unions to delineate a regulatory agenda. Moreover, in Ireland, there have not been significant experiences of collective organization involving couriers, whereas in Italy numerous episodes of collective mobilization in various cities have occurred over several years, qualifying this workforce as one of the most combative in Europe (Tassinari and Maccarrone, 2020).

Yet, none of these factors appears to have significantly affected, thus far, how Deliveroo handles its workers in any of its operating countries: its ‘super smart algorithm’ is always centrally designed (Heiland, 2023). Not by chance, various similar forms of digital manipulation have been largely detected in both countries. Hence, studying Deliveroo’s labour process in these countries may help corroborate the hypothesis on algorithmic management as a form of transnational workforce control based on the proliferation of misbehaviour.

Data collection

To collect the data, I triangulated a range of qualitative methods (in-depth interviews, daily observation of couriers’ working practices, and document analysis) deployed in fieldwork undertaken between May 2022 and March 2023 in four cities: Dublin in Ireland; Milan, Bologna, and Florence in Italy.

I first collected and analysed the relevant scholarly material addressing the topic of algorithmic management and couriers’ working experience. Guided by these theoretical insights, I then started my fieldwork research in Ireland (May–December 2022) and Italy (January–March 2023) by daily observing couriers’ working practices, focusing especially on their gatherings in their waiting-time areas, where I chatted sporadically with various workers. Informal and prolonged contact and interaction allowed me to build a trust relation with them and gave me the opportunity to interview them, some of their fellow couriers, and closer stakeholders. Between Ireland and Italy, I conducted 25 semi-structured interviews with various respondents, including couriers, union representatives and delegates, activists, and platform managers. Specifically, 12 interviews were conducted in Ireland (10 couriers, one political activist, one platform manager) and 13 in Italy (six couriers, three couriers/union delegates, four union officials; see Appendix for respondents’ sociodemographic composition).

This sample was built according to the purposeful-sampling criterion (Corbin and Strauss, 1990). The aim was not to provide statistically representative evidence of the phenomenon investigated, but rather ‘to build a theoretical explanation by specifying phenomena in terms of conditions that give rise to them’ (Corbin and Strauss, 1990: 9). My purpose was to gather all the relevant data concerning the circumstances giving rise to couriers’ various acts of digital manipulation. I ended my sample-building process once I reached what I perceived as my knowledge saturation on the topic – meaning, the recognition that couriers shared similar motivations to adopt such acts and the awareness that any additional piece of information on those motivations was becoming irrelevant (see Appendix for the sampling process details).

Data analysis

To advance my interpretation of compliant misbehaviour, I analysed all the material collected, following Grodal et al.’s (2021) active categorization approach, which is an abductive process where theoretical concepts and hypotheses are generated, refined and stabilized to produce a novel understanding of a specific phenomenon. It comprises three analytical stages: (1) generating initial categories, (2) refining tentative categories, (3) stabilizing categories.

More specifically, in the generating-initial-categories stage, I first manually transcribed all the respondents’ recording hours (38 hours) by coding this material in line with the theoretical insights acquired and developed in my preliminary document analysis. This led me to identify misbehaviour as a relevant (initial) category to explain couriers’ working practices and advance my first explorative hypothesis: ‘these workers continuously perform illicit conducts during their working time’. The array of misbehaviours spotted at this initial stage was relatively various, ranging from traditional acts of (excess) speeding and running a red light to those specifically related to the platform’s digital labour process, such as the duplication of Deliveroo’s profiles and the tricking of its account system.

Then, building on this hypothesis, I started my refining-tentative-categories stage by scrutinizing all the codified material in the literature on LPT organizational misbehaviour and in the labour and HRM studies on platform work. This analysis enabled me to develop a relatively more specific hypothesis on couriers’ misbehaviour: ‘most of the misconducts that couriers deploy are not detrimental to the company’s business’. In this research stage, I was able to identify all the relevant forms of couriers’ misbehaviour that contributed to the expansion of Deliveroo’s network effects. This scrutiny allowed me to discard those practices, such as the adoption of applications scamming Deliveroo’s GPS system or food delivery theft, which were not conducive to increasing such effects, and which in fact the platform itself contributed to actively wrestling back. Deliveroo’s ability to effectively distinguish between ‘good’ and ‘bad’ misbehaviour made me realize the extent to which the company was aware of the existence of this set of (good) misbehaviours and how it was able to actively leverage them to its own advantage.

Finally, in the stabilizing-categories stage, I re-analysed all the material collected considering this new theoretical insight. This led me to advance a new and definitive hypothesis on Deliveroo couriers’ individual working practices: ‘couriers are incentivized to perform acts of digital manipulation that violate company norms but boost Deliveroo’s network effects’. The stabilization stage enabled me to further narrow down the set of misconducts by excluding those remaining practices not fully consistent with such categorization, namely, those that neither breach Deliveroo’s terms of service nor boost its network effects: for example, self-building electronic bikes utilizing batteries above the limit established by the European regulation (250 W) was excluded, as this practice does not directly infringe a Deliveroo norm but rather another organization’s rule (European Union). Moreover, using a vehicle with a self-produced battery has no direct influence on the platform’s network effects. Likewise, traffic infringements, such as excess speeding or cycling on footpaths, were discarded, as they neither explicitly violate any platform policy nor increment Deliveroo’s network effects.

More importantly, the formulation of my definitive hypothesis helped me to further differentiate my concept from other similar, but not overlapping, theorizations, such as those of reworking, resilience, and coping, applied to platform work by other cognate disciplines (i.e. labour geography: see, for instance, Anwar and Graham, 2020). These practices are enacted by workers to enhance their earnings, without infringing any company regulation. Therefore, they cannot be considered (compliant) misbehaviour. The latter’s peculiarity is that, while circumventing some features of Deliveroo’s digital infrastructure, they directly expand the platform’s network effects. Following this last scrutiny, four practices of digital manipulation were identified as relevant to my study.

As all these misconducts caused a similar effect, I unified these separate yet related practices under my newly created category, compliant misbehaviour. This third and final stage of the active categorization process is similar to what Corbin and Strauss (1990: 14) qualified as selective coding: ‘selective coding is the process by which all categories are unified around a “core” category’. In such coding, the core category represents the phenomenon central to the study. Compliant misbehaviour is, therefore, the core category of my study (for a visual representation of my analytical process, see Figure 1).

Figure 1.

Compliant misbehaviour process coding.

Findings

Identifying compliant misbehaviour

This section addresses the study’s first research objective. The analysis shows how couriers are able to actively reappropriate, or creatively reutilize, part of their time, work and identity during their daily working activities for Deliveroo. Hence, of the four areas of contention highlighted by Ackroyd and Thompson (2022), the only one not spotted here is misbehaviour as appropriation of product, which is usually associated with the use (and abuse) of materials related to the means of production. Given couriers’ independent contractor status – they work with their own means of delivery only (i.e. mobile phones, pedal bikes, electric bikes, motorbikes and cars) – the absence of this misbehaviour may be expected.

Notably, the scrutiny identified four practices of manipulation of Deliveroo’s digital infrastructure that couriers recursively perform in both Ireland and Italy. For analytical ease, these are presented and discussed separately, although they are partially interwoven.

Utilization of multiple Deliveroo accounts

Signing up to the platform’s terms of service through its digital account system as a Deliveroo user equates to signing up to a labour contract with its terms and conditions. Understandably, Deliveroo permits its couriers to have only one digital account to work for the company. The account provides each courier with his/her own Deliveroo identity, which is the only profile with which she/he can legitimately work: ‘if you have previously had a Deliveroo account [one can read on the New Customer Offer T&Cs] you will not be eligible’ (Deliveroo, 2023). Indeed, each account can be associated with only one mobile number.

Yet, a considerable number of couriers are able to work with more than one Deliveroo account, as reported in various interviews conducted with couriers (IT2, IT3, IT4, IR1, IR2, IR6, IR10). For instance, one respondent (IR10) in Dublin confided that he knew ‘many couriers who also have two e-bike accounts for Deliveroo’. This misconduct is also adopted in Italy, as depicted in the words of a respondent in Florence (IT2), for whom:

there are so many of these foxy people. There are people who have two profiles for Deliveroo at the same time. They get two mobile phones and two accounts [with which they are associated] and thus they can earn much more. I don’t like this thing. With two accounts, one works more. Among those, I know of a dozen people who use two accounts.

As observed in both Ireland and Italy, this practice is especially widespread among undocumented migrants, who get various Deliveroo accounts and work simultaneously for all of them to get more deliveries. As confided by an undocumented migrant courier in Dublin (IR6) who works with two accounts: ‘so, you earn with two different profiles. [To do that] I use two mobile phones. . .. And the company never realized it’. However, this conduct is not viewed particularly well by official accountholders, often locals, who consider the utilization of multiple accounts as a form of illegal competition between colleagues. As noted by a Florentine courier interviewed (IT4): ‘recently, the multi-account phenomenon among Pakistanis is something that has created some tensions with the Italian component, which does not use them. In some places, this is very noticeable’.

As mentioned, couriers act over three areas – identity, time, and work – of Ackroyd and Thompson’s (2022) framework, which are actively reappropriated with the violation of a specific article of Deliveroo’s terms of service. The reappropriation of identity stems from workers’ ability to get and use contemporaneously several Deliveroo profiles (work identity multiplication). The management of multiple accounts represents a peculiar form of time appropriation, as it leads couriers to double their time expended on the app during their working day (working-time extensification). In doing so, they are able to get more orders and therefore to obtain more work at other couriers’ expense (delivery order increase).

Utilization of boost applications

Until November 2020, Deliveroo operated a self-booking system, a method of working-shift allocation still used by Glovo and Just Eat. In this system, couriers needed to connect with the company app twice a week (normally Monday and Thursday afternoon) to book working slots for the next few days. However, these windows did not open for all couriers at the same time; this depended on the courier’s ranking over the previous weeks, generally calculated by his/her availability and delivery reliability. The higher the ranking, the greater the chances of booking the best shifts (around the weekends) and therefore earning more money.

However, this system can be easily circumvented by using various bots, namely, software available on the web whose purpose is to boost couriers’ ranking in the company’s booking system. These apps provide a permanent refresh function for the couriers: the software constantly refreshes the site to find and immediately book any working shift as soon as it becomes available. These bots are easy to use and very effective in terms of results. Indeed, as noted by an Italian union official (IT6), one only needs to pay a monthly subscription, then: ‘by using these applications, in two months people can overtake in the ranking couriers who have been working for years’. It is not by chance that over the past years Italian trade unions have tried to counter these apps by reporting their usage to the platforms. Companies’ responses, however, have always been elusive and ambiguous. In light of this, the same union official (IT6) explicitly blamed the platforms for the proliferation of these apps, remarking that: ‘there are some urban legends saying that there are also some people inside the company, who work on developing them. It might even make sense, as after all it’s to their advantage’.

By and large, this misbehaviour, which used to seriously disrupt Deliveroo’s former method of working-shift allocation, can be considered a case in point of work appropriation in Ackroyd and Thompson’s (2022) framework. Indeed, by using these apps, couriers were more likely to increase their rankings, so that they obtained more work and better working-shifts at their colleagues’ expense (working-shift acquisition).

Renting other people’s accounts

Deliveroo explicitly condemns all illicit acts that its workers may commit during their working time as well as illegal practices related to the major issue of visa documentation to access courier work. The British company has always declared that it works in close collaboration with the relevant authorities to verify couriers’ work permits and thus prevent lawbreaking. In relation to this issue, in April 2023, a Deliveroo manager stated to the British press that: ‘Deliveroo takes a zero-tolerance approach towards any rider who fails to meet their legal obligations when working with us. If a rider is found to be without the right to work in the UK, we will stop working with them with immediate effect’ (Burrell, 2023). A similar orientation was confirmed by the Deliveroo manager interviewed (IR12).

However, a considerable chunk of this workforce has worked and continues to work for Deliveroo in both Ireland and Italy without complying with its terms of service in relation to work permits. Although it is difficult to quantify, as there are no statistics on this phenomenon, various studies have reckoned that more than half of Deliveroo couriers work globally without having their own profile, but renting other people’s accounts (see, for instance, Mendonça and Kougiannou, 2024; van Doorn and Vijay, 2024). In Ireland, for instance, the Brazilian migrants who constitute most of this workforce normally hold a student visa that does not allow them to work as couriers, but they do so by using another person’s account. All the couriers interviewed pointed out how, in Dublin, there is an informal account rental market, built into several chats and marketplaces on social media networks, where migrants without work permits can exchange and rent Deliveroo accounts for a weekly fee (IR2, IR4, IR6, IR7, IR8, IR10). This rental arrangement is reported in detail in this interview extract from IR6, a Brazilian courier working illegally:

I rented an account from the beginning. There was already this informal market of account rentals. €35 a week for a normal bike, etc. . .. My official accountholder for Deliveroo is a Brazilian with European citizenship.. . . I usually rent accounts from people who have all the legal documents to hold an official account.

In Italy, Deliveroo couriers perform similar misconducts in relation to their digital accounts. Like in Ireland, they are mostly undocumented migrants who carry out this illegal practice as the only way to earn an income, as explained by IT2:

Renting an account here is 20%. If you earn €100 you have to give €20 to whoever has sublet it to you. . .. There are many who rent profiles. Those who arrive here, have no job, do not speak Italian, can immediately work as Deliveroo riders by renting an account. This is a very good thing. Deliveroo is good for that. Anyone can work for Deliveroo. Most people work with these accounts. Not only Pakistanis, but also Bangladeshis, Indians, Africans. They all work with rented profiles.

Similar to the utilization of multiple Deliveroo accounts infringing a specific article of its terms of service, this practice is a manifestation of misbehaviour related to the appropriation of work and identity. Indeed, these people rent a Deliveroo account to have access to work, meaning that they appropriate other people’s work identity (work identity faking), and, in doing so, they appropriate other couriers’ work (delivery order capture).

Tricking Deliveroo’s accounts system

Deliveroo operates an accounts system encompassing various profiles: each of these is associated with a specific means of delivery. Couriers can choose to work with one of four transport categories: pedal bike, electric bike (e-bike), motorbike, or car. This system appears internally differentiated, as Deliveroo’s algorithm is perceived to favour specific profiles by allocating more deliveries to the accounts associated with the more performing means of delivery. For instance, motorbike or e-bike accounts seem to receive more orders than pedal bike accounts, as the former have a faster means of delivery than the latter. This insight does not derive from any company communication; it is knowledge acquired and shared among couriers based on their daily working experience. As highlighted by a Brazilian courier in Dublin (IR10): ‘Deliveroo has a very smart algorithm. It is playing with us all the time. It gives more orders to specific categories of couriers’. In particular, ‘those on motorbikes receive more orders than those on bikes’, as specified by an Italian courier based in Florence (IT4).

Knowing this rule, most couriers, in both Ireland and Italy, regardless of their legal status, trick the Deliveroo’s account system by signing up to the platform with a more performing account, but still employing a cheaper means of delivery. Couriers with an e-bike often sign up on Deliveroo to a motorbike account, with a motorbike to a car account, or with a pedal bike to an e-bike account. As recounted by a Brazilian female courier based in Dublin (IR2): ‘It’s very common to have a motorbike account even though you use an electric bike. Because the motorbike account gets more orders’.

Couriers value this tricking practice so highly that it is also widespread among the local component, who are regularly in possession of all the legal documentation to work. An Italian courier working in Florence (IT3) confided that: ‘we all do this. Even regular riders’. Or, as described more extensively in the words of an Irish courier (IR9):

I have an electric bike account even though I am using a normal bike. With the electric bike account, it’s better, because they give you priority in the orders. It’s not about distance, but about priority in receiving orders. I have several short-distance orders indeed. Most people cheat like that. Using an electric bike account, although you have a normal bike.

In Ackroyd and Thompson’s (2022) language, this misbehaviour can be considered as work and time appropriation. Indeed, in tricking Deliveroo’s accounts system in the various ways described above, couriers are able to get more orders and therefore to obtain more work at their colleagues’ expense (delivery order increase). In turn, this leads them to a work intensification dynamic, meaning, couriers work more intensively within the same timespan by exploiting it to its utmost (working-time intensification), which can be considered a perverse form of time reappropriation (for a visual representation of the subsection’s findings, see Table 1).

Table 1.

Taxonomy of Deliveroo couriers’ forms of compliant misbehaviour.

Misbehaviour	Time reappropriation	Work reappropriation	Product reappropriation	Identity reappropriation
1. Utilization of multiple Deliveroo accounts	Working-time extensification	Delivery order increase	X	Work identity multiplication
2. Utilization of boost applications	X	Working-shift acquisition	X	X
3. Renting other people’s accounts	X	Delivery order capture	X	Work identity faking
4. Tricking Deliveroo’s accounts system	Working-time intensification	Delivery order increase	X	X

Author, based on Ackroyd and Thompson (1999, 2022).

Misbehaving for Deliveroo

This section addresses the study’s second research objective. The analysis shows how the four forms of misbehaviour identified above are central to Deliveroo’s economic business. Despite manipulating some components of the platform’s digital infrastructure, they all contribute to increasing Deliveroo’s network effects. The latter are, in fact, the source of the platform’s power to redesign algorithms ‘to facilitate pricing and dispatching’ (Chen and Sun, 2020: 1566).

The expansion-of-the-network effect enables Deliveroo to extract more value in two ways: from the higher number of transactions on which companies apply their fee and from the capacity of platforms to decrease the fee accorded to workers for each transaction as a result of the market marginalization of the other platform companies and, accordingly, of the increase in their reserve army of workers (Cini, 2023).

Overall, one can observe a positive correlation between the time that workers spend on the Deliveroo app – whatever form it takes – and an increase in the company’s value (Shapiro, 2018). The longer couriers are logged on, the more network effects are created, the more value Deliveroo extracts (Chen and Sun, 2020). The four forms of digital manipulation identified contribute in various ways to increasing such value. This nexus is now separately analysed for each of them.

Utilization of multiple Deliveroo accounts

The more Deliveroo accounts are contemporaneously utilized, the more couriers are logged on, the more network effect is created, the more profit the company makes. It is not by chance that several respondents perceived tolerance from the company towards such misconduct. As explicitly highlighted by an Italian courier (IT3): ‘[Deliveroo seems to] tolerate the use of double accounts, even of three accounts’. Notably, the utilization of multiple accounts is beneficial to the company’s business, as it expands Deliveroo’s reserve army of couriers (expansion of workers’ reserve army), whose existence is profitable for the platform in two respects: (a) it increases the fleet of couriers potentially available at any time, thereby making the company more appealing for an increasing number of clients willing to use it and thus increasing the number of transactions; (b) the more accounts are online, the lower the delivery fee that Deliveroo must pay to couriers. Indeed, the presence of many accounts available ensures a greater likelihood of couriers potentially eager to carry out the delivery, so contributing to decreasing the fee for such delivery. Consequently, although explicitly illicit, the utilization of multiple accounts is, overall, a Deliveroo-friendly practice and hence in line with its economic interests.

Utilization of boost applications

Deliveroo needs a large and reliable supply of workers, especially during peak hours, which are the periods with more client requests. These periods are normally weekends, festivities and evenings on which there are particular events (such as football matches or other mass events). The bots help their users to climb up the company’s ranking system, thus giving them the opportunity to choose the most remunerative working shifts, namely, peak hours. Accordingly, the more couriers use bots, the greater the competition among couriers over these hours, the greater the likelihood that all client demands will be rapidly met. In doing so, these apps contribute to the rationalization of the company’s calendar, as they are designed to book instantaneously the working slots that are made available when other couriers renounce them. As effectively explained by one of the respondents (IT6): ‘the bots continuously monitor the opening of new slots within the calendar. As soon as a work shift becomes available, the apps book it immediately’. In this way, all the working shifts are always and constantly covered (working-shift rationalization). Put otherwise, bots allow Deliveroo to have a permanently ready and reliable workforce, especially during peak hours, whose constant presence contributes to increasing client satisfaction. Although formally prohibited, bot use represents thus a Deliveroo-friendly practice that is fully convergent with the platform’s economic interests.

Renting other people’s accounts

The more people are working for Deliveroo regardless of their status, the larger the fleet of couriers at its disposal to potentially increase its volume of transactions. Those who rent other people’s accounts make a particularly beneficial contribution to this cause. As invisible couriers, they are, in fact, more vulnerable, more obedient, and therefore more exploitable than regular couriers. As effectively observed by one of the migrant respondents (IR8), working illegally as a courier in Dublin since 2019: ‘The first actor profiting from this informal system of rental accounts is the company. For sure. They can exploit an army of invisible workers with no rights and no protection. Very vulnerable people whose only goal is to work as much as possible’. The impression among couriers is that Deliveroo is aware of this phenomenon (‘Deliveroo knows it’, IR4); yet, its existence is largely or deliberately neglected, as it is understood to provide an enormous benefit to the company’s profits. As succinctly expressed by a Brazilian respondent, who worked illegally for Deliveroo during the pandemic period in Dublin (IR7): ‘Platforms do not really care if their accounts are subcontracted. They only care about deliveries and profits’. Hence, although a fully illegal practice, subletting accounts to undocumented migrants is widely tolerated by Deliveroo, as long as the authorities do not initiate an investigation (Burrell, 2023), given that its proliferation is fully consistent with its economic interests (expansion of workers’ reserve army).

Tricking Deliveroo’s accounts system

Unlike the other three forms of misbehaviour where the connection with Deliveroo’s interest is explicitly visible, the practice of tricking its digital accounts system does not seem to be particularly consistent with the company’s interests. In fact, couriers sign up on the platform with accounts that are perceived as consigning more orders while working with delivery means that have a lower performance level (namely, pedal bikes) than those associated with the officially registered accounts (generally motorbikes or e-bikes). Seen from this angle, this misconduct seems anything but convenient for Deliveroo. However, tricking Deliveroo’s accounts system can indirectly prove beneficial to the company – and it is perceived as such by various couriers (IT3, IT4, IR2, IR8, IR9) – as the motivation that is normally behind such behaviour, namely, earning more money by executing more deliveries, is conducive to couriers’ work intensification. In other words, as long as the company allows couriers to register with accounts that provide them with more orders without their being entitled to such orders, these couriers are motivated to work for more hours more intensively and obediently for Deliveroo (work intensification). This tricking game brings about a sort of tacit agreement between the parts, as a courier working in Dublin alluded to indirectly (IR8): ‘for the courier, it is more convenient because the app gives you more deliveries. So, in the end you earn more, and the company makes more profit’. Consequently, tricking Deliveroo’s accounts system can substantially produce and result in a very company-friendly outcome (for a visual representation of this subsection’s findings, see Table 2).

Table 2.

Taxonomy of compliant misbehaviour’s Deliveroo friendly effects.

Misbehaviour	Deliveroo friendly
1. Utilization of multiple Deliveroo accounts	Expansion of workers’ reserve army
2. Utilization of boost applications	Working shift rationalization
3. Renting other people’s accounts	Expansion of workers’ reserve army
4. Tricking Deliveroo’s accounts system	Work intensification

Author.

Discussion and conclusion

This study of Deliveroo couriers’ working practices in Ireland and Italy shows that misbehaving at work is still widespread in the contemporary workplace, even in the new platform work context where the standard employment relationship is absent and workers are controlled by algorithms (Duggan et al., 2020). In line with the LPT tradition (Ackroyd and Thompson, 2022), the analysis highlights how such behaviour is to be understood as a rational response that workers deploy when faced with managerial control. In this sense, the findings rebut the classical managerialist assumption that misbehaviour is a dysfunctional feature of capitalist organizations that can be corrected with a more rational organization of labour based on the introduction of new technologies. This assumption is misleading, as people will always misbehave at work, meaning that they will always do things that they are not supposed to do, given the persistence of the labour indeterminacy problem (Braverman, 1974).

However, attending to such problem seems no longer necessary in platform work. With the introduction of the algorithm as a workforce management mechanism, workers feel compelled to adopt certain anticipatory behaviours of compliance (Bucher et al., 2021). As also shown in this study, under the algorithmic pressure, workers enact specific (mis)conducts, geared to reappropriate and reutilize part of Deliveroo’s digital work equipment in a way that is convergent with the company’s economic interests. Paradoxically, the wider the spectrum of such behaviours, the higher the level of workers’ compliance towards the platform. When couriers misbehave during their working day, they do it mainly for Deliveroo; or, more precisely, the most relevant practices of digital manipulation that Deliveroo couriers perform augment the platform’s network effects. Notably, four of these conducts were identified as critical: the utilization of multiple Deliveroo accounts, the utilization of bots, the rental of other people’s accounts, the tricking of Deliveroo’s accounts system. As their effects are simultaneously beneficial to both couriers (at least in the short term) and Deliveroo, they have been grouped under the concept of compliant misbehaviour. The theoretical elaboration of this concept represents the main contribution of the paper.

Theoretical contributions

The introduction of the compliant misbehaviour concept makes two contributions to specific debates in labour studies – one to the literature on platform work resistance (Cant, 2020; Lei, 2021; Newlands, 2021; Tassinari and Maccarrone, 2020; Treré and Bonini, 2022; Woodcock, 2021), the other to the classical theory of organizational misbehaviour (Ackroyd and Thompson, 1999, 2022).

First, its introduction counters various optimistic readings of algorithmic manipulation, which conceive of it as an expression of platform workers’ everyday resistance capacity (Newlands, 2021; Treré and Bonini, 2022). These interpretations view such manipulation, aiming overall to circumvent algorithmic surveillance to attain more visibility on the platform, as individual acts of (micro) resistance to platform power, as, for instance, Newlands (2021: 732) put it in her account of food delivery couriers’ practices of data obfuscation:

On an individualistic level, workers also perform micro acts of resistance through the use of technical measures, such as fake GPS signals. By exploiting gaps in organisational knowledge about a worker’s ‘actual’ position, a worker can obtain more profitable work for themselves. This is one form of data obfuscation, though levied for purposes of economic benefit rather than privacy enhancement. Workers do not want to avoid observation within the normal course of work and they go to great efforts to maintain the algorithmic gaze in instances where internet connections are disrupted or their smartphones run out of battery.

Likewise, Treré and Bonini (2022: 310) focus on a similar tactic of manipulation – what they call algorithmic amplification – which is an act of algorithmic circumvention that users adopt ‘to multiply and amplify their voice to acquire more visibility’ on platforms. In deploying such tactic – this argument goes – workers are able to turn the algorithmic gaze to their (economic) advantage: more visibility on the platform means increasing their likelihood of getting more work and, thus, earning more money. Yet, in doing so, they also spend more time on the app, which – as seen – augments digital companies’ network effects. Put otherwise, the acts of data obfuscation and algorithmic amplification do not disrupt platforms’ economic business, but rather reinforce it by following precisely the algorithms’ inbuilt operational logic (Meijerink and Bondarouk, 2023). Hence, these acts of individual manipulation cannot be considered resistance.

However, algorithmic resistance does occur, and it has taken place in platform work also (Cant, 2020; Lei, 2021; Tassinari and Maccarrone, 2020; Woodcock, 2021). Coordinating various forms of collective action, some of these workers have managed to challenge platforms to demand better conditions in terms of pay, labour rights and employment status (Tassinari and Maccarrone, 2020). For instance, particular categories of cloudwork, namely, workers who perform labour remotely mediated by digital platforms (e.g. Amazon Mechanical Turk and Clickworker), have created online forums in which to help one another and from which to run brand-shaming social media campaigns against their platform (Woodcock, 2021); and food delivery couriers have organized ‘digital strikes’, namely, protests combining the practice of ‘unlogging’ en masse from the app with onsite pickets and online social media campaigns against the company for which they work (Cant, 2020; Lei, 2021; Tassinari and Maccarrone, 2020).

The peculiarity of all these actions is that they presuppose workers’ overt willingness to disrupt digital companies’ economic interests: the practice of mass unlogging from the app is precisely aimed at damaging their network effects. Hence, the strategic goal of deliberately targeting platforms’ economic business is the key feature distinguishing these tactics from the acts of algorithmic manipulation. In the absence of such feature, these acts are doomed to be incorporated in the dynamics of the labour process and thus made functional to platforms’ interests.

Second, the concept of (compliant) misbehaviour contributes to refining Ackroyd and Thompson’s (1999, 2022) classical notion by revisiting it in a single but relevant respect: its assumption that workers misbehave in search of (increasing) autonomy from managerial control. In their interpretation, misbehaviour always consists of ‘non-compliant or counter-productive’ practices (Ackroyd and Thompson, 1999: 26) that employees execute vis-à-vis their company to evade work discipline. Therefore, misbehaving is assumed to be a way to resist managerial direction in order to increase worker autonomy. However, the analysis of Deliveroo couriers’ working practices in both Ireland and Italy shows that there is no automatic connection between their misbehaviour and the evasion of managerial control. On the contrary, couriers misbehave with various acts of digital manipulation to attract more algorithmic supervision. In doing so, they reappropriate part of their time, identity, and work, not to obtain some relief from their work effort, but rather to increase it.

In this respect, the concept of compliant misbehaviour seems relatively closer to Burawoy’s (1979) classical interpretation, whereby some forms of misconduct in the pattern of work games, such as the practice of making out, are an expression of competition games between employees played out of boredom, leading to an increase in their own productiveness at work. These games were informally tolerated by the management, as they were seen as a factor for the generation of workplace consent: the practice of making out was, in fact, assumed to bond workers to the factory system and, thus, strengthen their ideological commitment to it.

The character of work games seems considerably changed in platform work though. The analysis of Deliveroo’s labour process shows that, today, a techno-organizational architecture (i.e. algorithm-based control), and not an ideological condition, aligns workers’ (mis)behaviour with the company’s business. Under the algorithmic pressure, workers are stirred to play competition games out of necessity. Not by chance, the workers who are more likely to participate in these games are also those more economically dependent on the platform: the undocumented migrant component is even compelled to do so to get access to food delivery work (Mendonça and Kougiannou, 2024).

Anyhow, whatever their legal status or socioeconomic condition, all Deliveroo couriers are committed to attracting the company’s algorithmic gaze to get more work. In doing so, they deploy a varied set of misconducts, whose proliferation reinforces not only the platform’s network effects but also their own exploitation (Cini, 2023). The manifestation of compliant misbehaviour points, thus, to the deep-seated power imbalance underpinning such a working relation (Chan, 2021). The algorithm-based control seems to exacerbate, rather than pacify, the inherent antagonism between capital and labour, which results to be central to platform work also (Gandini, 2019; Tassinari and Maccarrone, 2020).

Limitations of my study

My paper has some limitations. Here, I identify what I consider the three most relevant. The first is a time limitation and has paradoxically to do with the key feature of algorithms themselves. Indeed, as workforce control tools in continuous evolution, algorithms may favour the proliferation of novel forms of (mis)behaviour that this study cannot predict. Seen from this angle, the four forms of compliant misbehaviour that Deliveroo couriers have performed so far represent only a snapshot of a specific pattern of interaction in a limited period. This holds particularly for one of the forms of misbehaviour identified above, namely, the utilization of bots. As mentioned, this misconduct is no longer practiced, as the delivery allocation system with which it was associated – the self-booking system – is no longer operational, but it is kept in the analysis to show how couriers’ digital manipulation is relevant to Deliveroo’s algorithmic management, as much before as now. Therefore, its delivery allocation system may also vary over time, but the proliferation of (new and ever-changing) forms of misbehaviour related to its digital infrastructure by which this platform controls its workforce and boosts its core business remains central.

The second limitation relates to the scarce attention that my study has devoted to the role of other employment relations actors, especially state authorities and trade unions, in the analysis of Deliveroo couriers’ misbehaviour. Although the relatively low intervention capacity that these actors have so far exhibited over the matter of workers’ digital manipulation may have played some role in its proliferation in the food delivery sector, it is likely that the attitude of these actors towards such matter will change soon. Without any predictive ambition, it seems likely that the increasing intervention of various political actors (and legislations) in platform work will force platforms to change their workforce control logic and potentially their algorithms’ content. These interventions may be particularly expected to impact on the development of food delivery’s labour market, whose segmentation along ethnic lines has been, so far, a key source for Deliveroo to impose its algorithm-based work games, leading to division and competition among its workers. What is more, the invisible status of several of these workers seems to have played a critical role in their only episodic capacity, thus far, to unite in class-based solidarity actions (Hua and Ray, 2021). The hope is that the new political measures may effectively address the issues of labour market segmentation and workers’ invisibilization by promoting a more dignified working environment in the food delivery sector and in the platform economy more in general.

The third and last limitation pertains to the low generalizability of compliant misbehaviour to other platform work sectors. My paper focuses on only one digital platform in the food delivery sector, Deliveroo, which, although relatively paradigmatic in terms of algorithmic management, cannot represent the full sector, let alone other sectors. Future studies are thus encouraged to explore whether and how the interaction dynamics between algorithms and workers lead to the proliferation of various forms of compliant misbehaviour in other sectors also.

Supplemental Material

sj-docx-1-org-10.1177_13505084251334994 – Supplemental material for ‘Misbehaving for Deliveroo. How couriers’ digital manipulation boosts the platform’s business’

Supplemental material, sj-docx-1-org-10.1177_13505084251334994 for ‘Misbehaving for Deliveroo. How couriers’ digital manipulation boosts the platform’s business’ by Lorenzo Cini in Organization

Footnotes

Acknowledgements

I would not have been able to write this article without the long and prolific discussions with several friends and colleagues of mine. In particular, I am very grateful for their valuable feedback to Lisa Dorigatti, Valeria Pulignano and all the members of her research team, Paul Thompson and all the participants at the workshop ‘New Directions in Labour Process Theory’ held in Padua in January 2023, three anonymous reviewers, and the two editors of Organization. This article is also theirs. All errors, instead, remain mine.

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The author received financial support for the research and authorship of this article from the MSCA-IF-2020 research action ‘COntesting GOvernance by NUmbers: The Mobilizations of Food Delivery Couriers across Europe in Time of the Pandemic (COGONU)’ (Grant agreement ID: 101028811).

ORCID iD

Lorenzo Cini

Supplemental Material

Supplemental material for this article is available online.

Author biography

Lorenzo Cini is a Lecturer in Employment Relations and HRM at Cork University Business School, University College Cork, Ireland. His main research interests are labour movements and conflicts in contemporary transformations of the world of work. His research has been published in leading international journals such as Economic and Industrial Democracy; Theory and Society; New Technology, Work and Employment; European Journal of Industrial Relations; and Work, Employment and Society, as well as academic presses such as Bristol University Press. Lorenzo is an editorial board member of the ABS4 journal Work, Employment & Society.

References

Ackroyd

Thompson

(1999) Organizational Misbehaviour. London: Sage.

Ackroyd

Thompson

(eds) (2022) Organizational Misbehaviour, 2nd ed. London: Sage.

Anwar

M. A.

Graham

(2020) ‘Hidden Transcripts of the Gig Economy: Labour Agency and the New Art of Resistance Among African Gig Workers’, Environment and Planning A 52: 1269–91.

Bieler

(2021) Fighting for Water: Resisting Privatization in Europe. London: Bloomsbury.

Braverman

(1974) Labor and Monopoly Capital: The Degradation of Work in the Twentieth Century. New York, NY: Monthly Review Press.

Bucher

E. L.

Schou

P. K.

Waldkirch

(2021) ‘Pacifying the Algorithm. Anticipatory Compliance in the Face of Algorithmic Management in the Gig Economy’, Organization 28: 44–67.

Burawoy

(1979) Manufacturing Consent. Changes in the Labor Process Under Monopoly Capitalism. Chicago: University of Chicago Press.

Burrell

(2023) ‘Deliveroo, UberEats riders among 60 arrested in London amid Home Office crackdown’. Evening Standard, 24 April. https://www.standard.co.uk/news/london/deliveroo-uber-eats-just-eats-driver-arrest-home-office-immigration-b1076211.html

Cant

(2020) Riding for Deliveroo: Resistance in the New Economy. Cambridge: Polity Press.

10.

Chan

(2021) ‘Hunger for Profit: How Food Delivery Platforms Manage Couriers in China’, Sociologias 23: 58–82.

11.

Chen

J. Y.

Sun

(2020) ‘Temporal Arbitrage, Fragmented Rush, and Opportunistic Behaviors: The Labor Politics of Time in the Platform Economy’, New Media & Society 22: 1561–79.

12.

Cini

(2023) ‘How Algorithms Are Reshaping the Exploitation of Labour-Power: Insights Into the Process of Labour Invisibilization in the Platform Economy’, Theory and Society 52: 885–911.

13.

Collinson

Ackroyd

(2006) ‘Resistance, Misbehaviour, and Dissent’, in Ackroyd

Batt

Thompson

, et al (eds) The Oxford Handbook of Work and Organization, pp. 305–26. Oxford: Oxford University Press.

14.

Corbin

J. M.

Strauss

(1990) ‘Grounded Theory Research, Procedures, Canons and Evaluative Criteria’, Qualitative Sociology 13: 3–21.

15.

Deliveroo (2023) New Customer Offer T&Cs. T&Cs for our promotions & restaurant deals. https://deliveroo.co.uk/terms/

16.

Duggan

Sherman

Carbery

, et al (2020) ‘Algorithmic Management and App-Work in the Gig Economy: A Research Agenda for Employment Relations and HRM’, Human Resource Management Journal 30: 114–32.

17.

Eisenhardt

K. M.

(1989) ‘Building Theories From Case Study Research’, Academy of Management Review 14: 532–50.

18.

Gandini

(2019) ‘Labour Process Theory and the Gig Economy’, Human Relations 72: 1039–56.

19.

Griesbach

Reich

Elliott-Negri

, et al (2019) ‘Algorithmic Control in Platform Food Delivery Work’, Socius: Sociological Research for a Dynamic World. Epub ahead of print August 30 2019. doi: 10.1177/2378023119870041

20.

Grodal

Anteby

Holm

A. L.

(2021) ‘Achieving Rigor in Qualitative Analysis: The Role of Active Categorization in Theory Building’, Academy of Management Review 46: 591–612.

21.

Grohmann

Pereira

Guerra

, et al (2022) ‘Brazilian Workers’ Experiences of Dishonest and Uncertain Algorithmic Management’, New Media & Society 24: 1611–31.

22.

Heiland

(2021) ‘Neither Timeless, Nor Placeless: Control of Food Delivery Gig Work via Place-Based Working Time Regimes’, Human Relations 75: 1824–48.

23.

Heiland

(2023) ‘The Social Construction of Algorithms: A Reassessment of Algorithmic Management in Food Delivery Gig Work’, New Technology Work and Employment 40: 1–19.

24.

Hua

Ray

(2021) Spent Behind the Wheel: Drivers’ Labor in the Uber Economy. Minneapolis, MN: University of Minnesota Press.

25.

Lei

Y. W.

(2021) ‘Delivering Solidarity Platform: Architecture and Collective Contention in China’s Platform Economy’, American Sociological Review 86: 279–309.

26.

McDonnell

Carbery

Burgess

, et al (2021) ‘Technologically Ediated Human Resource Management in the Ggig Eeconomy’, International Journal of Human Resource Management 32: 3995–4015.

27.

Meijerink

Bondarouk

(2023) ‘The Duality of Algorithmic Management: Toward a Research Agenda on HRM Algorithms, Autonomy and Value Creation’, Human Resource Management Review 33: 1–14.

28.

Mendonça

Kougiannou

N. K.

(2022) ‘Disconnecting Labour: The Impact of Intraplatform Algorithmic Changes on the Labour Process and Workers’ Capacity to Organize Collectively’, New Technology Work and Employment 38: 1–20.

29.

Mendonça

Kougiannou

N. K.

(2024) “We Are Not All the Same’: The Capacity of Different Groups of Food Delivery Gig Workers to Build Collective and Individual Power Resources’, Work Employment & Society. Epub ahead of print June 8 2024. doi: 10.1177/0950017024125743

30.

Mendonça

Kougiannou

N. K.

Clark

(2022) ‘Informalization in Gig Food Delivery in the UK: The Case of Hyper-Flexible and Precarious Work’, Industrial Relations 62: 60–77.

31.

Newlands

(2021) ‘Algorithmic Surveillance in the Gig Economy: The Organization of Work Through Lefebvrian Conceived Space’, Organization Studies 42: 719–37.

32.

Polkowska

Mika

(2022) ‘Is It Possible to Outsmart Uber? Individual Working Tactics Within Platform Work in Poland’, European Societies 25: 606–26.

33.

Reid-Musson

MacEachen

Bartel

(2020) “Don’t Take a Poo!’: Worker Misbehaviour in on-Demand Ride-Hail Carpooling’, New Technology Work and Employment 35: 145–61.

34.

Rosenblat

Stark

(2016) ‘Algorithmic Labour and Information Asymmetries: A Case Study of Uber’s Drivers’, Journal of International Communication 10: 3758–84.

35.

Roy

D. F.

(1959) “Banana Time’: Job Satisfaction and Informal Interaction’, Human Organization 18: 158–68.

36.

Shapiro

(2018) ‘Between Autonomy and Control: Strategies of Arbitrage in the ‘on-Demand’ Economy’, New Media & Society 20: 2954–71.

37.

Tassinari

Maccarrone

(2020) ‘Riders on the Storm. Workplace Solidarity Among Gig Economy Couriers in Italy and the UK’, Work Employment & Society 34: 35–54.

38.

Treré

Bonini

(2022) ‘Amplification, Evasion, Hijacking: Algorithms as Repertoire for Social Movements and the Struggle for Visibility’, Social Movement Studies 23: 303–19.

39.

van Doorn

Vijay

. (2024) ‘Gig Work as Migrant Work: The Platformization of Migration Infrastructure’, Environment and Planning A 56: 1129–49.

40.

Veen

Barratt

Goods

(2019) ‘Platform-Capital’s ‘App-Etite’ for Control: A Labour Process Analysis of Food-Delivery Work in Australia’, Work Employment & Society 34: 388–406.

41.

Vieira

(2023) ‘Platform Couriers’ Self-Exploitation: The Case Study of Glovo’, New Technology Work and Employment 38: 493–512.

42.

Wood

A. J.

Graham

Lehdonvirta

, et al (2019) ‘Good Gig, Bad Gig: Autonomy and Algorithmic Control in the Global Gig Economy’, Work Employment & Society 33: 56–75.

43.

Woodcock

(2021) The Fight Against Platform Capitalism: An Inquiry Into the Global Struggles of the Gig Economy. London: University of Westminster Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB