Riders in app time: Exploring the temporal experiences of food delivery platform work

Prologue

Tomás leaned up against the glass front of McDonald’s as he drew a sigh, balancing his bicycle with one hand, while updating the Deliveroo rider app with the other. The first author Katrine Duus had just finished interviewing him inside the McDonald’s in exchange for a cup of coffee. They now waited together for Tomás to receive his first order. Tomás was 19 years old, and his studies in architecture had taken him from his home country Portugal to Brussels. He had been working for Deliveroo for six months and was paid five euros per delivered order. As Tomás and Duus stepped out of the McDonald’s, Tomás told her that his shift had already started. He laughed at Duus when he saw the panic in her face. She had learned from other riders that the misdemeanour of not logging onto the rider app at the agreed time could cause riders to lose their good statistics, and she was afraid that she had caused him to delay in logging onto the app. ‘Don’t worry’, he said, while he opened the massive lock that attached his expensive looking mountain bike to the lamp post outside the McDonald’s, ‘I already connected while we were inside, wrapping up’. Tomás’ good statistics ensured that he was among the first to choose between shifts when they were made available in the Deliveroo rider app every two weeks. If Tomás’ score fell too low, he would have to fight over the shifts that were left when riders with a better score than his had already had their pick.

Tomás and Duus waited twenty minutes outside McDonald’s while he told her about how he kept his phone ‘juiced’ throughout the entire shift. He explained that he always had his phone connected to the power bank he had invested in. ‘If the phone dies, I win no money’, he added. His phone was the only connection he had to Deliveroo. If he was not connected to the Deliveroo platform through the app, he would get no notifications for new orders and would not be paid for his work. Hence, both the internet connection and battery life of his smartphone were vital for the way he earned his money.

Tomás had now been waiting for thirty minutes for a food order. A thirty-minute wait was not uncommon among the Deliveroo riders that Duus interviewed in Brussels during her ethnographic fieldwork in 2018. Tomás kept looking at his phone screen, refreshing the app every minute, hoping the red pin on the map would move. Finally, it showed ‘New Order!’, and the red pin instantly moved to the location of the restaurant requesting a delivery.

The order request that he had just received was less than one kilometre away from where they were. Even so, Tomás sighed at the sight of the location of the red pin that marked a specialised burger restaurant: ‘They are never ready when I come’. As he mounted his bike, he explained that some restaurants accepted orders through the Deliveroo app even though they did not have the capacity to finish them in time. However, Tomás did not want his acceptance rate to be affected (this rate is another performance score that is visible in the rider app). So, he accepted the order from the restaurant, said farewell to Duus, and rushed off on his bike.

The next time Duus met Tomás, he told her that he had waited for another thirty minutes for the order after arriving at the burger restaurant. Moreover, the manager of the restaurant had asked him to wait outside because he did not want the restaurant’s brand to be associated with Deliveroo riders. Tomás was very angry, but he knew that if he left he would lose the five euros he earned per successful delivery, and he feared that his acceptance rate would be affected, too. Moreover, all the waiting time and the time he had spent biking to the restaurant made it seem stupid to walk away from the restaurant, despite their rude behaviour. After all, waiting time was an integral part of working for Deliveroo.

Introduction

This article is about the time perception and temporal agency of food delivery bicycle riders working through a digital platform. Drawn to the idea of flexible work, riders engaged in delivery work mediated by the Deliveroo rider app. Digital platforms like Deliveroo promise to optimise their resources on an ongoing basis through continuous data collection and analysis, helping all users to spend their money and time in a more efficient way. At least, this is the image that Deliveroo try to convey to their users.

With the omnipresence of smartphones, the optimisation of time through technology has reached its zenith. Clock time in its original invention depended on space and distance: the world is divided into time zones, and we have an idea of how long it takes to go from one place to the other by car, train, or plane. Time mediated by information and communication technologies, or ICT time as it is sometimes called, however, promises to compress duration to zero and to deliver concerted action at once in so-called ‘real time’ (Adam, 2007; Hope, 2006). ICT time offers an image of an accelerated temporal reality in which time is pressed and diminished, and succession and duration have been replaced by seeming instantaneity and simultaneity (Adam, 2007: xi). The image of ICT time is smooth. It produces a fantasy of liberating time from space and creating spaceless, or frictionless, time. However, as our analysis shows, it also causes new frictions, ruptures, and deceleration, since ICT time cannot escape the intertwinement with other forms of time that are deeply bound to the spatial reality of human existence in a body.

Studies of time in the digitalised labour market often emphasise acceleration and speed (Wajcman, 2014; Srnicek, 2017; Rosa, 2010, 2013), but our empirical data shows that the riders’ time is partly made up of gaps and waiting time, what we will refer to as unwanted time. We will argue that instances of unwanted time are created by frictions between the data doubles produced by Deliveroo’s real-time renditions of the riders on the one hand, and the riders’ movements through the spatiotemporal food delivery economy of Brussels on the other. We also employ Michael Flaherty’s conceptualisation of time work (2003) in order to analyse the riders’ experiences of navigating the temporal structure of the app, its mediations among users, and the spatiotemporal food delivery economy of Brussels. Thinking through the riders’ time work allows us not only to address the temporal structure of the app but also to explore how the riders experience app work and how it fits with their lives and plans.

Food delivery work mediated by an app provides an exemplary lens to increase our understanding of the nature of time in the digitalised labour market. The riders work through their smartphones, which are closely connected to the instructions communicated in ICT time while simultaneously moving around the cityscape of Brussels. The riders’ work would not be possible if they did not carry a smartphone on them with a GPS signal, and if they did not have access to a stable internet connection. Studies of ICT time and digital platform work are scarce, but Julie Yujie Chen and Ping Sun have made an excellent contribution regarding the labour politics of time in the platform economy. Chen and Sun’s study highlights the extremely precarious working conditions of digital food delivery riders in Beijing, China (2020). They provide a snapshot of a temporal arbitrage regime that, via the image of real time enabled by internet communication technologies, marginalises immigrant riders in Beijing. In our study, we seek instead to examine the riders’ experiences of and motivations for taking on digital platform work in Brussels. Duus’ conversations with this particular group of young, male riders indicate that in their own conception their work offers them a specific kind of flexibility that allows them to fit work around life.

We draw on data collected by Duus through ethnographic fieldwork focusing on digital platform work in Brussels during the first half of 2018, supplemented by her continued contact with riders and the wider field of platform labour since then. However, it is vital to stress that in this article we examine the conditions of work only in the first half of 2018 where riders received a fixed price per order and worked in shifts. The working conditions have changed many times since then, with an overall decrease in earnings (Willems, 2021). Furthermore, some of the riders’ tactics of manipulating the time of the app mentioned here have in recent years become impossible to apply due to a higher likelihood of being disconnected from the platform, losing all connection to Deliveroo. This among other reasons, also means that only one of the riders that Duus met in 2018 is still working through the app at the time of writing. Overall, there has been a low retention rate of Deliveroo riders, and in the last years there has been a noticeable change in the socio-demographic of riders, where migrant workers with or without a work permit is now dominant (Willems, 2021) compared to a majority of student workers at the time of Duus’ fieldwork in Brussels. Duus’ interlocutors were male Deliveroo riders, ¹ aged 18–36. She gained contact with the riders partly by hanging out at the designated meeting points that the riders frequented during their working hours, and partly through a local Facebook group for Deliveroo riders in Brussels. In many cases, she approached the riders and conducted short, semi-structured interviews and invited them later to participate in more formalised interviews lasting approximately an hour. In this article, we introduce four riders out of the total of 27 riders interviewed during Duus’ fieldwork in 2018. The riders’ names are pseudonyms, and the four riders have been selected with the intention of representing the wide range of tactics for reducing unwanted time that Duus encountered among all her interlocutors. By using only four riders to narrate the findings, we intend to make the findings more intelligible to the reader. By focusing on individuals and their specific circumstances in life, we can gain access to a deeper phenomenological understanding of the temporal experiences of flexible digital capitalism. Furthermore, by zooming in on a few riders, the different narratives also demonstrate that not only do the tactics for reducing unwanted time differ from rider to rider, so does the overall temporal experience of working through an app.

Time and temporality in flexible digital capitalism

In the essay ‘Time, Work-Discipline and Industrial Capitalism’ (1967), E.P. Thompson describes the effects of industrial capitalism on people’s perception of time and work. The introduction of modern capitalism led to a clear demarcation of the concepts of work and life: workers who sell their labour to employers in the labour market distinguish between domains and activities that are identified as work, and those that are identified as social life. Time became something that could be spent or saved, and thrift became a virtue pertaining not only to money but also to time, as captured in the well-known phrase ‘time is money’. Work became a commodity that could be bought and sold like other commodities, and clock time became the common way of measuring productive labour.

Clock time, the kind of time that is used to measure working hours, is disembedded from social life and social bodies. It is often referred to as objective time or machine time because it is built upon modern science and technology and the idea of taming nature (e.g. Adam, 2004). In contrast, body time, or subjective time, is the time of nature, the body, and their rhythms: the sun that rises in the morning, our stomachs that growl to tell us we are hungry. This is not to say that there is something called natural, uncontaminated time that is subject to modern technologies like the clock, but that different technologies and mediations constitute different kinds of time and temporality (Stiegler, 1998). What kinds of time, and which frictions between different kinds of time, come into existence when workers sell their labour and organise their work through smartphone apps and digital platforms?

In the modern era, we understand the world both through body time and clock time, and modern capitalist production relies on representations, institutions, and technologies of time that control and mediate between these often conflicting and overlapping experiences of time (Adam, 2004; Bear, 2016). Likewise, the digital platform economy of food delivery is driven by customers’ appetite, the daily rhythms of mealtimes, and the riders’ bodily efforts as they traverse the city to transport food. It is also organised through clock time. For instance, the whole food economy is based on opening hours, and food is ordered and delivered at specific times that are registered and used for calculations and statistics, even though the riders are not paid by the hour but by completed orders. However, the new kind of time reckoning or time handling that is created by apps and digital networks also creates new experiences and categories of time and new kinds of frictions between body time, clock time, and the ICT time of the app, as well as other experiences of time among the riders (see Adam, 2004).

Since the beginning of modern industrial capitalism, one of the goals of technological developments has been to speed up production, increase efficiency and profitability, and produce more, or deliver more services, in less time. This pursuit of efficiency is also what drives digitalisation forward in contemporary societies, and new digital technologies are transforming our notions of time and work. One of the best examples of this pursuit of efficiency and transformations of time and work with digitalisation and datafication is the emergence of digital platforms that facilitate communication, interaction, and sales between different businesses, employees, and customers, all usually described as different user groups (Gillespie, 2010), as if they were structurally equal.

Digital platforms mediate and manage a range of different services and products, from social media platforms (Facebook, Twitter, etc.) and media sharing platforms (e.g. YouTube) to service-oriented platforms for transport (e.g. Uber or Lyft), accommodation (e.g. Airbnb), and food delivery (such as Deliveroo). Deliveroo offers an infrastructure for three different interconnected apps: a rider app that is used by riders to access deliveries of food from restaurants to customers; a restaurant app through which the restaurants can accept orders from the customers and communicate to the riders that the food needs delivery; and an app for the customers in which they can browse through the different restaurants and their menus and place and pay for food orders that the riders deliver to their addresses.

Information and communication technologies (ICTs) and the availability of ubiquitous computing, typically through smartphones and other digital devices, have enabled quick and cheap data sharing and direct communication and financial transactions between businesses and customers. However, ICT does not just speed up communication. It also creates an ideology of real time in which time is liberated from space, thereby promising instantaneity despite distance. The ideas of optimisation, efficiency, and profit through instantaneity permeate many other aspects of human life and productivity. The Deliveroo riders in our study also wanted to optimise their time and earnings and sought flexible work in order to be able to optimise their time.

The desire for flexibility

All the Deliveroo riders interviewed in our study shared a desire for a flexible working life, and many of them had deliberately applied for work on the Deliveroo platform with this in mind. They all mentioned the flexible scheduling as a key motivator for seeking out digital platform work.

Many studies of platform work point to the precarious working conditions (e.g. Cano et al., 2021; Van Doorn, 2017; Cant, 2019). In such studies, flexibility often turns out to be only illusory (see Piasna and Drahokoupil, 2021: 2). Instead of neglecting the precarious nature of platform work, we seek to add a more nuanced understanding of why the riders engage in platform work and how they experience such work by investigating their understanding of flexibility. The Deliveroo riders in our ethnographic study are more homogenous than those seen in previous studies in Brussels by Jan Drahokoupil and Agnieszka Piasna (2019). As an example, Duus only encountered two riders with financial responsibility for anyone other than themselves and those riders were not students. Compared to Drahokoupil and Piasna’s report, ² 15% of the interviewed student riders reported to support others than themselves financially on a regular basis (2019: 21). The report was based on Deliveroo riders’ paycheck data from September 2016 to April 2017 and two surveys in, respectively, December 2017 and January 2018. Despite the homogeneity of Duus’ interlocutors, Drahokoupil and Piasna also found that among the three most important reasons, 85% of the riders they surveyed had chosen this job because of its scheduling flexibility (2019: 27). In contrast to Drahokoupil and Piasna’s study, Duus followed her interlocutors over a longer period of time not only while they were working but also after work, using in-depth semi-structured interviews, thereby gaining a more nuanced perspective of exactly what the riders meant by ‘flexibility’ and how they experienced it.

In general, the flexibility that many of Duus’ interlocutors described was characterised by the flexibility to cancel a shift until 24 hours before it started without any negative repercussions, and to end a shift after it had begun (for instance, if it started raining heavily). They also described an experience of flexibility in terms of when they wanted to work, that is, how many and which shifts they wanted to work. However, unlike the ability to cancel a shift, the degree of this experienced flexibility depended on the individual rider’s performance scores which determined how many available shifts the rider would be able to choose from. This overall scheduling flexibility was enabled by the design of the app. Through what other studies of digital platforms have coined ‘algorithmic management’ (e.g. Lee et al., 2015), the riders experienced a specific type of scheduling flexibility to which they were not accustomed in more regular forms of employment. The algorithmic management of the scheduling system was predicated on the riders’ use of smartphones. By managing the riders through their smartphones and not in situ, it was possible to monitor a multitude of riders with very few managers.

Drawing on Manuel Castells’ theory of urban space, we see management as taking place in the space of flows rather than in the space of places (2020). Management in the space of flows was enabled by different internet communication technologies, but first and foremost by the riders’ smartphones. Managing a digital representation of the rider, or their data double, enabled Deliveroo to offer the more flexible scheduling that the riders desired. According to Kevin Haggerty and Richard Ericson, data doubles are made by ‘(…) abstracting human bodies from their territorial settings and separating them into a series of discrete flows. These flows are then reassembled into distinct “data doubles” which can be scrutinized and targeted for intervention’ (Haggerty and Ericson, 2000: 606). Gemma Newlands also draws on the notion of data doubles. She argues that if the platform only monitors the data double, it is monitoring the part of the job that riders do on their phones and not the part they do in the spatiotemporal setting of the cityscape (Newlands, 2020). The scheduling flexibility that Deliveroo offered was achieved by managing data doubles in the space of flows instead of managing the riders in the space of places.

The flexible nature of working through the rider app had persuaded the architecture student Tomás to work for Deliveroo. Coming from Portugal to study in Belgium, he had chosen to work as a Deliveroo rider because it was the only job that allowed him to always put his studies first: ‘With my studies another job would not be possible’. The flexibility of the Deliveroo system enabled Tomás to work more at the beginning of the month until he had earned what he needed to cover his expenses, and he also appreciated the ease with which he could cancel a shift without any negative repercussions.

In the following, we draw on Tomás’ experiences of working through the app to show how the app made and mediated other demands on Tomás’ time in return for the flexibility that he gained.

App time

For Tomás, the Deliveroo app first and foremost dissolved the connection between the work invested in the app and the money gained from it, a connection he was used to from other types of work. Now he earned five euros per order, and his total pay depended on his ability to secure and complete orders through the app. Because large parts of the logic behind Deliveroo’s algorithm were unknown to the riders, Tomás tried to navigate using the information he had received from Deliveroo and his own guesses of how the orders were assigned. He knew how his performance scores were calculated: good performance scores were achieved by logging onto the rider app on time when his shifts began and by not cancelling any shifts less than 24 hours before they started. He kept his performance scores as close to a hundred percent as possible. This secured him the privilege to choose shifts before other riders who had worse scores than him. Usually, he was among the riders who chose shifts first when the new shifts were released every two weeks.

While the allocation of shifts seemed pretty transparent to Tomás, the allocation of orders during a shift was a mystery to him. This made it difficult for Tomás to assess when to accept or reject an order. It did not help his decision making that the app only showed the location of the restaurant where the food was to be picked up, but not the end-destination of the trip, that is, the address where the customer lived. Tomás told Duus that one of the most annoying parts of the job was knowing that Deliveroo had this information but did not share it with him before he had accepted the order and picked up the food at the restaurant. Tomás felt it was unfair that he was paid the same for each order, when the distance he had to bike from the restaurant to the customer could vary by several kilometres.

The degree of transparency in the functioning of the app for the riders and for the other users was, of course, something that was carefully designed and a feature in algorithmic management (Lee et al., 2015). Giving out or withholding information was part of the temporal structure of the app as well as being a way to manage the riders and secure the flow of the whole system, ensuring that the orders that took longer time to deliver would still be completed.

This general opaqueness of the way in which the orders were allocated and the impact of the acceptance rate on the riders’ performance statistics made it difficult for the riders to navigate the app. Knowledge of how many orders would be offered during a shift, and how much time it would take to complete an order, was either uncertain or intentionally kept from the riders by the platform. The riders’ uncertainty regarding the final earnings from a shift can be explained by the instant remuneration per completed order and the deliberate opaqueness of the app, but cannot be reduced to this alone because the Deliveroo platform also mediates between the restaurant and customers across distances in the cityscape that the riders must then traverse.

In contemporary global capitalism, the drive towards real time plays an important role (Hope, 2006). A new wave of finance capitalism is based on ICT’s drive towards instantaneity in that finance capital is generated in a space of information flows, and uses ICT time to make an extra profit, unlike productive capital with its slower, sequential time frames which are linked to physical reality in which time and space are interlinked (Hope, 2006: 280). This compression of time in digital capitalism and its social and sociological consequences have been described and critiqued by social science scholars (Rosa, 2010, 2013; Wajcman, 2014). Instead of assuming a compression of time and the existence of real time in platform work using smartphones, we will investigate how the frictions between different forms of time occur in the app by analysing the frictions between the rider and the rider’s data double.

When customers order food through the Deliveroo app, and riders agree to deliver an order, these tasks are translated into flows of digital information that travel at lightning speed – or at least faster than one can speak or bike through the city. As soon as a customer enters a food order into her app, the restaurant receives a notification on its app, emulating a kind of instantaneity and speed that speaks to and reinforces modern ideals of efficiency. When the rider cycles through the city, his movements are automatically tracked through GPS signals that the customers and restaurants can follow, and when the rider enters information on his smartphone (accepting, rejecting, or completing an order), this information is also immediately shared. In the information flows of the app, an order is delivered, and thereby completed, not when the food is handed over to the customers in physical space, but when the rider enters the action on the screen. Thus, discrepancies (or frictions, as we call them) easily occur between the ICT time of the app and the spatiotemporal food delivery economy and cityscape of Brussels.

The spatiotemporal structure of the food delivery economy in Brussels

‘All the roads are blocked because of a bike race’, Tomás tells the Deliveroo employee on the other end of the phone, who is calling to inquire about a delayed order that Tomás, according to the app’s estimations, was supposed to deliver fifteen minutes ago. ‘I’ve told you [referring to the rider support team] many times today. I can’t say how long it will take – all the roads are blocked at that end of Brussels’, and Tomás continues: ‘Could you please instal a bonus so it is worth it to keep riding today?’ The employee informs Tomás that she is not in charge of the bonus system, but that he can write an email to Deliveroo. Tomás knows that it usually takes Deliveroo two weeks to reply to emails, so he hangs up and ends up wheeling his bicycle on the sidewalk in order to deliver the food order, which must be cold by now, dreading the meeting with the customer who has waited thirty minutes more than the estimated time of arrival in the app.

Food delivery work makes different demands on the riders’ time and movement. These demands are structured partly by the temporal structure and algorithmic management of Deliveroo’s rider app, and partly by the spatiotemporal structure of the food delivery economy of Brussels. The riders experience frictions because they are situated in the cityscape of Brussels while they are logged onto the app as well as being connected to customers and restaurants through digital information flows (as shown in Tomás’ attempt to navigate different temporal structures). The Deliveroo platform and its algorithms calculate how long it normally takes to complete a delivery from one place to the other, sending this information to the customers who have ordered food, irrespective of the actual cityscape and the rider’s current ability to meet this expectation.

Tomás quickly learned that the information he received from Deliveroo regarding how the app worked did not present the whole picture of how his time was structured. He booked his shifts based on what he knew about the usual customer demand for food at different times; and before accepting an order, he considered how long the different restaurants usually made him wait for an order after he had arrived. He tried to avoid unnecessary waiting time. Apart from the weather, holiday seasons had a big impact on order demands. There were specific weeks in which Tomás would sometimes go on holiday himself because he suspected there would be an excessive number of riders fighting over very few orders, with many people having left the Belgian capital for a holiday. Aside from the seasonal and daily rhythms of the city, Tomás’ time was also structured by the spaces involved in his shifts, both the cityscape and the spaces in and around the restaurants. During the waiting times in the restaurants, Tomás would often have to stand in a corner of the restaurant to not take up a seat, and, as shown in the prologue above, the staff sometimes even asked riders to wait outside.

It was not only the riders who had to navigate the spatiotemporal structure of the food delivery economy in Brussels. Deliveroo’s rider support system, whose primary task was to ensure that the orders were delivered on time, also had to step in when the algorithmic management and predictions did not reflect the current situation in Brussels. So, as well as navigating between the real-time mediations of the Deliveroo platform and the spatiotemporal food delivery economy in Brussels (consisting of but not limited to the traffic density, weather conditions, the customers’ appetite for fast food, and how busy the restaurants were), the riders also took into consideration the bonuses that the Deliveroo rider support system sometimes put in place. Tomás often made use of the bonus system, which meant that even though he tried to schedule shifts around his studies, he was also open to the opportunity to earn a bit more per order than usual, thereby scheduling his life around work to a certain extent. As we saw at the beginning of this section, Tomás did not only have to navigate the temporal structure of the app. He also, and more importantly, had to navigate the friction between the temporal structure of the app and the food economy of Brussels.

Unwanted time

According to Tomás and most of the other riders in this study, the worst part about time spent waiting in restaurants was that the rider app often gave false information with regard to when a food order was ready to be picked up. The restaurants had incentives to accept as many orders as possible to increase their earnings. Like the burger restaurant mentioned in the prologue, many restaurants accepted all the orders that came in through their Deliveroo order app, even though they knew that it was impossible to have the food ready at the time stated in the app. This resulted in a lot of extra waiting time for the riders.

The discrepancy between the time the order was actually ready for pick-up in the restaurant and the time that the restaurant had listed in their app reflects the discrepancy between the physical reality of, for example, a meal, and its data double in the app. When a customer sees the GPS signal of the rider at the restaurant and at the same time can see that the food order is supposed to be ready, this parallel reality creates the expectation that the food is on its way and puts pressure on the riders, as we saw when Tomás tried to complete an order while the streets were blocked due to a bike race. The riders try to fill the gap between the expectation of instantaneity relating to the space of digital information flows, and time in the space of places where bodies are required to actually prepare the meal and deliver it. These gaps, and the waiting time created by them, are what we refer to as ‘unwanted time’.

We have named it ‘unwanted’ because the riders referred frequently to the extra time imposed upon them by the Deliveroo platform. In their different categorisations of time, the waiting time induced by such glitches in the Deliveroo system was particularly unsolicited and annoying for the riders, for example, in comparison with the time spent biking through the city, which they accepted as a natural and legitimate part of their working hours. With remuneration being tied to orders and not hourly wages, there were certain activities that the riders accepted as part of the work involved in each order (time spent picking up the food, biking through the city, delivering it, and even accepting short waiting periods in between). Unwanted time, however, covers all the extra time that the riders felt was imposed on them by the malfunctioning of the app: longer waiting time between orders, glitches, slow restaurants, and the overall uncertainty mediated and partly created by the app. If the riders have bought into the ideology of efficient ‘real time’ made possible through internet communication technologies, unwanted time is all the unexpected extra time that runs counter to this ideology.

There were other glitches between the space of digital information flows and the spatiotemporal food delivery economy that also produced unwanted time. For instance, there were sometimes mistakes in the addresses that the customers’ provided in their app. Sometimes the addresses were wrong, and sometimes the map software in the riders’ app could not handle the way the address was entered into the customers’ app. At other times, the customer simply did not answer the doorbell to receive the food when the rider arrived. Hence, unwanted time occurred every time the app made a temporal prediction about the food delivery that did not work out in the spatiotemporal food delivery economy.

The riders’ time work

Faced with these frictions, challenges, and impositions of unwanted time, the riders employed different tactics to avoid them. In order to examine the riders’ temporal experiences, we use Michael Flaherty’s analytical framework of time work (2003; 2011; see also Flaherty et al., 2020) to disentangle the creativity and skilfulness with which the riders attempt to make the app time and its entanglements with the spatiotemporal food delivery economy of Brussels fit their own lives. Flaherty defines time work as the ‘(…) individual or interpersonal efforts to create or suppress particular kinds of temporal experience’ (2003: 17). Using the analytical framework of time work enables us to pay equal attention to how the app structures the riders’ time as well as the riders’ actual experiences of working within this temporal structure. Even though management through data doubles created unwanted time owing to the frictions arising between the time of the app and the spatiotemporal structure of the food delivery economy, it also created a room for navigation for the riders, precisely because of these frictions.

Whereas Flaherty’s interest focuses on the temporal experience itself, the riders presented in this article are only partly interested in changing a temporal experience (for instance, the boredom or irritation experienced when restaurants are delayed). Parallel with the experience itself, they focus on the relation between time use and financial outcome. How can they make events unfold in ways that allow them to avoid non-profitable time? They always have a pee before their shift starts, check they have the correct amount of air in their bicycle tyres, and have their phone and powerbank fully charged – eliminating all the potential non-profitable extra time relating to their bodies, bicycles, and phones, that is, the time that they can exercise control over.

Based on Duus’ interviews and ethnographic fieldwork in Brussels, we have identified two different categories of time work that were shared widely by the riders. The first type was the ability to plan their work around their lives, at least to some extent. This resembles Flaherty’s notion of allocation, and the idea that people’s self-determined allocation of time to certain activities reflects their values (Flaherty, 2011: 12). The Deliveroo riders used the temporal structures of the app to exercise the kind of flexibility that allowed them to prioritise other activities, their studies for instance, as well as enabling them to avoid any commitment to a longer employment horizon resulting in the need to prioritise loyalty to an employer over other aspects of their lives. This type of time work is related to the scheduling flexibility that was the riders’ motivation for taking on digital platform work in the first place.

The second type of time work was the riders’ different tactics for reducing unwanted time in the app. In Duus’ material, she identified different sub-tactics for reducing unwanted time. While Flaherty is mainly interested in how people manipulate their experience of time, several of the riders tried to reduce the amount of unwanted time and thereby make more money during a shift. Either by manipulating the data doubles and/or by optimising the workflows involved in navigating the spatiotemporal food economy of Brussels. However, the mere shift of attention towards reducing unwanted time in the app is an act of time work in itself. Flaherty bases his concept of time work on William James’ statement: ‘My experience is what I agree to attend to’ (quoted in Flaherty, 2003: 17). Converting unwanted time into the riders’ own time was the most prevalent act of time work among Duus’ interlocutors. This type is similar to Flaherty’s notion of ‘taking time’, which is the agency you can unfold when others demand your presence, but you yourself decide where you direct your attention (2011: 115–130). In the following, we will discuss the different kinds of time work through the experiences of Tomás and the three other Deliveroo riders Andrés, Xavier, and Barthelemeus.

Fitting work around life

Andrés was 20 years old and originally from Columbia. He had worked as a cash-in-hand bike messenger for the past three years, while he travelled through South America and Europe. Working through Deliveroo was the most formalised job he had ever had. However, he could only do this job by working through an account his Belgian friend had set up for him because Andrés did not have a work permit or residency in Belgium. Xavier was 19 and eager to leave Belgium, where he had lived his entire life. He studied permaculture part-time while working as much as he could through the Deliveroo app to save up money to go travelling. Like Andrés, he loved biking and was pleased that he could earn money doing it. He also stressed the lack of a human manager and the flexible scheduling system as part of his motivation for seeking out platform work. Barthelemeus was also a Belgian citizen. He was 22 years old and spent his time working as much as the Deliveroo app allowed him while he was studying social science. Like Xavier, Andrés, and Tomás, Barthelemeus did not see working through Deliveroo as a career path, but rather as an additional source of income in between other, more important aspects of his life.

For Tomás and many of the other riders, the ability to spend time on their studies was the most important requirement they had for a job. In comparison to regular employment, working through the Deliveroo platform gave them a sense of control and flexibility, for instance, because they could cancel shifts up to 24 hours before they began. Tomás would allocate time to work when he was not at the school of architecture, and sometimes he would even ‘take back time’ by cancelling his shifts if he had miscalculated how much time he needed for his studies.

Xavier also worked through the Deliveroo app because of the flexible shift system: He only wanted to work as much as he could for two months, and then he wanted to go travelling for a lengthy period of time. ‘I don’t want to lie to an employer’, he told Duus, when she asked if there were no other jobs he could apply for to obtain full employment. He suspected that no one would hire him for two months for other positions, unless he lied and hid his plan to quit again very soon. With Deliveroo, Xavier, like Tomás and many of the other riders, appreciated the lack of a human manager that he did not have to disappoint. The app was completely indifferent to his short temporal horizon.

The fear of being fired and the dislike of being bossed around were important reasons for many of Duus’ interlocutors to take on platform work. When working as a cash-in-hand food delivery worker, Andrés had often found himself in situations where restaurant managers dictated what hours he should work, even though they only paid him per delivery. In comparison, working through Deliveroo gave him more control over his time because he decided when he wanted to work. ‘But it is still a shitty job’, he told Duus.

These examples of time work show the riders’ experiences of how the flexible scheduling system enables them, to a certain extent, to influence the timing as well as the allocation of work, despite the overall temporal structure of the app. The riders stressed that this type of work made it possible for them to focus on what was important for them, and to fit their working lives around that. They felt able to exercise control over the timing and frequency of their working hours, able to allocate time to what was important to them to a much higher extent than they had ever imagined or experienced in previous jobs.

Andrés had even found different workarounds for the temporal structure of the app. To him, the facelessness of the app made him look for every loophole and try to hack the system. First of all, he did not care so much about his performance statistics because he had found a way to circumvent the shift system and thereby avoid the temporal demands inflicted by the rider performance score. He had discovered that he could often log onto the rider app without having a shift and work anyway. Sometimes he logged on after his shift had started or forgot to cancel a shift in due time, and if the weather suddenly changed from sunny to rainy or if there were few orders he did not hesitate to log out during a shift. Sometimes he employed the tactic of overbooking shifts for himself: when the shifts were released, he would book all the shifts that were available to him and then cancel some of them later if he did not feel like working. ‘It is when I want, it is a good job for that’, he declared after sharing some of his tactics for circumventing the temporal structure of the app. In this respect, Andrés even reduced the unwanted time imposed on the riders through the temporal structure of the app, which most of the riders still navigated within.

Reducing unwanted time

As Andrés’ different ways of hacking the app to gain more flexibility already show, he took pride in outsmarting the app’s temporal structure by manipulating the data doubles. He used his phone and his knowledge of the temporal structure of the app and its mediations to increase the number of orders during a shift. Andrés always declined certain restaurants where he had waited longer than announced by the app on more than one occasion. Compared to Tomás, he did not care about the percentage indicating how many of the orders offered he had accepted. Furthermore, Andrés had found a workaround for the lack of information regarding the address of the customers. He would go to a food market with many different food stalls and nearby restaurants that were connected to the Deliveroo app. He would then accept the first order from one of the restaurants and in the same breath indicate that he had picked up the order (even though he had not). This enabled him to see the location of the customer. If the customer lived too far away, he would cancel the order through rider support. In this way, Andrés managed to hack the temporal structure of the app that withheld information about how long it would take him to reach the customer. By letting the app know that he had received the order, even though he had not, Andrés manipulated information regarding his data double, as well as the data double of the food order. In this example, we see how Andrés used the data double strategically by providing false information, thereby taking advantage of the app’s blindness to the spatiotemporal food delivery economy of Brussels.

Instead of hacking the data doubles, Barthelemeus tried to reduce the amount of unwanted time by taking control of his interactions and movements in the food economy of Brussels. For instance, the app encouraged the riders to return to designated zone centres (location pins in the interface of the Deliveroo app) while they were waiting for new orders. However, Barthelemeus had developed a specific pattern of movement in the streets with the highest density of restaurants, which he was convinced secured him more orders during a shift, even though this has never been confirmed by Deliveroo. It did, however, provide him with the understanding that he reduced unwanted waiting time. He also knew exactly which route to bike in the different work zones to get to the customers as fast as possible; and, like Tomás, he familiarised himself with the habits of the customers and the seasonal rhythms of the city. Another tactic that Barthelemeus used to increase his number of orders involved working through the Ubereats app, a competitor to Deliveroo, at the same time as taking shifts for Deliveroo. This enabled him to receive orders from two different platforms and combine the routes through the city, sometimes accepting an Ubereats order while he was still waiting for a Deliveroo order to be prepared. Furthermore, before Barthelemeus even left his apartment to start working, he would embark on the same pre-work rituals as Tomás.

Barthelemeus also employed emotional labour by chatting with the different staff at many of the restaurants, using these friendly relationships to get his order faster and thereby minimise his waiting time. While we saw in the section describing ways of fitting work around life that emotional expectations were fairly absent between the rider and the Deliveroo, this example reveals that Barthelemeus made use of emotional relations in his work in order to optimise his time spent in the app and thereby reduce unwanted time. By actively using his body, both as a means of locomotion and to establish relations in the spatiotemporal food delivery economy, Barthelemeus found a great satisfaction in working through the Deliveroo platform.

The algorithmic management system created the unwanted, unpaid time that the riders experienced and tried to delimit as in the examples above; but it also enabled them to convert unwanted time into their own time, as we will show in the following.

One of Flaherty’s (2003) types of time work concerns people’s different efforts to influence the perception of duration. One classic example is that of students who are doodling in their notebooks to speed up a boring class and reduce their experience of lengthiness. They are still in the classroom, adhering to the norms of the situation, but they spend their time in a manner that their teacher does not intend (Flaherty 2003: 22). In very similar ways, the Deliveroo riders employed time work to manipulate their experience of waiting.

Xavier employed different tactics to change his experience of unwanted time. Since he lived within the limits of one of Deliveroo’s work zones, he would always log onto the app from home and engage in other activities, like reading a book, while he waited for the first order. This gave him the feeling that he was outsmarting the app. Even though he emphasised that it was ‘cool’ that he could relax at home while he waited for his first order, he also explained that it was difficult to immerse himself fully in other activities because he kept refreshing his phone, afraid that he would miss an order request and thereby miss out on a potential source of income. In similar fashion, he would often chat with other riders, talking about working through Deliveroo or life in general to pass the time when he was waiting at the zone centre. If there were no other riders, he would watch Netflix on his phone or call a friend, while making sure that he was keeping an eye on his phone so he would not miss an order. By using his waiting time for pleasure, Xavier diminished the annoyance that other riders like Tomás sometimes felt when they were waiting for orders. Flaherty remarks that we can attend to the moments we are in with either patience or impatience. If we do so with impatience, we become victims of circumstances, obsessed by lost time. If we do so with patience, we attend to the moments we are in (Flaherty, 2011: 131). In his most recent contribution to this journal, Flaherty notes that we are all both victims and architects of time (2022). The same can be said about the riders: whereas Tomás’ conscientious attitude towards work made it aggravating to wait for new orders because he saw other riders riding and wanted to be working too – thinking about time as money that should not be wasted – Xavier simply engaged in other activities. Chatting with other riders, playing games, or watching films on your smartphone were frequent ways to kill waiting time and transform it to meaningful time of your own. Transforming waiting time into meaningful time did not eradicate the waiting time, but it made it more bearable for the riders to work in the app’s temporal regime.

Frictions of time

Unwanted time is created by frictions between ICT time in the space of flows and the spatiotemporal food economy of Brussels. Flaherty, whose concept of time work we have used extensively in this article, looks at the perceived duration of time as it is shaped by the interplay of self and situation. Like Flaherty, we have been interested in efforts directed toward changing temporal experiences. The experiences we have focused on all form part of what we have called ‘unwanted time’. The riders do not have a single category for this, but Duus detected the presence of this unwanted time by observing certain practices which are intended to prevent it. But we have also seen that however hard the riders strive to control or manipulate the experience of being in time, the human factor is just one thread among many. Despite the riders’ efforts to move through time with competence and ease, they get caught in intersections. Spatial distances and forces like wind and rain interfere with their efforts, as do the actions of other people. The riders had to navigate within the framework of obscure and unpredictable information from Deliveroo and their experiences with customers. Guesswork and careful planning and analysing were their means of action. There is nothing new in the desire of these workers to outsmart the management; what is new is the very specific limited room of flexibility that management through data doubles offers and the fact that remuneration is tied to it. By converting unwanted time into their own time, the riders gained a sense of agency. When we analyse their involvement with time work, it becomes evident that it was the riders themselves who created the flexible experience of digital platform work that they wanted to achieve by working through the rider app. The rider app might have provided a framework for flexibility, but the specific design of the algorithmic management delimited the actual scheduling flexibility available to the riders.