THEMATIC CLUSTER: THE POLITICS OF DATA IN LATIN AMERICA: TOWARDS A TERRESTRIAL INTERNET Surveillance and the ecology of frictions in platform urbanism: the case of delivery workers in Santiago de Chile

Digital platforms have quickly become popular thanks to the algorithms that have made it possible to work from a smartphone. The potential bene ﬁ ts of job ﬂ exibility and easy complementary income for “ delivery partners ” have been highlighted. However, work through tasks precariously paid, the impossibility of organizing in a labor union and the constant monitoring of labor performance has put platforms to the test. Despite these working conditions, delivery sta ﬀ are not passively under surveillance, but rather the platforms are a space of frictions. In this article, we seek to abandon the idea of delivery platforms as objecti ﬁ ed entities that are the natural/universal result of technological progress in the city by adressing the frictions and local practices of reproduction of the platform. Through a 6-month ﬁ eldwork consisting of interviews with Uber Eats delivery workers in Santiago de Chile, this article seeks to describe and delve into practices of subversion that delivery sta ﬀ use to resist excessive surveillance at work, where indicators such as rating are essential to avoid being “ deactivated ” . Our ﬁ ndings indicate that Uber Eats ’ platform deploys various strategies that we will call friendly surveillance , which operates as veiled nudges in which the system seeks to keep delivery sta ﬀ engaged through incentives and promotions. At the same time, the platform collects data and de ﬁ nes what it means to be an e ﬃ cient delivery partner.


Introduction
The delivery company Uber Eats began operating in Santiago de Chile in November 2017. The food delivery service initially added over two hundred fast food restaurants to the platform and gave customers the opportunity to track their order in real time (Montes 2017). The platform was also accompanied by a series of discourses and promises about a "revolution" in labor, with increasingly automated and smart management protocols. The new forms of digital work flexibility would allow delivery staff to decide how to link to the platform in time and space to receive extra compensation in the amount of CLP$1400 per order. Since then, Uber Eats has expanded to several Chilean cities.
While labor flexibility is the main attraction of these platforms, the category of autonomous or independent workers (Srnicek 2017;Gandini 2019;Veen, Barratt, and Goods 2020) implies that delivery workers are located on the margin of labor rights, engaging in precarious work that is referred to as "gigs," "tasks" or "rides" (De Stefano 2015). Thus, delivery work is a full-time job without medical insurance, hyper-flexible and without schedules, and also without a pension plan or the right to unionize (Scholz 2017;Dinegro 2019, Hidalgo andValencia 2019). In this context, delivery staff have staged several protests in Santiago de Chile to improve their labor conditions. Unlike a growing number of countries, in Chile there is currently no legislation to protect workers in the platform economy. Due to the absence of legislation, couriers are not recognized as workers, confronting a complete lack of labor protection. These local disputes have demonstrated that in contrast to the experience of immediacy that clients enjoy, Uber Eats delivery workers face a series of injustices and daily frictions as they deliver an order.
Along these lines, the literature has insisted that behind the apparent smartness paired with these technologies, platforms continue to generate "black boxes" (Rosenblat and Stark 2016;Veen, Barratt, and Goods 2020). When couriers sign up for Uber Eats, they tacitly accept a set of protocols and ways of working, subjecting themselves to silent surveillance by algorithms. Uber unilaterally awards promotions and monetary incentives; controls schedules, routes, and delivery fee prices; and exercises remote digital surveillance. Delivery staff who fail to comply may have their account deactivated.
In this article, we address these problems from the notion of frictions. Tsing (2005) proposes the concept of friction to describe tensions and abrasions that emerge in connection with the interests of a global project and its insertion in a local context. In other words, frictions are "the awkward, unequal, unstable, and creative qualities of interconnection across difference" (Tsing 2005, 4). This concept "refuses the lie that global power operates as a well-oiled machine. Furthermore, difference sometimes inspires insurrection" (Tsing 2005, 6). As such, the anthropologist calls on us to look at what these frictions bring to the world in terms of forms of power. Frictions will be one of the analytical tools used to address the development of Uber Eats in Santiago de Chile. We want to discuss the importance of understanding the expansion and movement of digital platforms workers in friction. As we will demonstrate, a careful analysis of frictions implies rematerializing the idea of the digital as fluid and instantaneous, recognizing the multiple asymmetrical relationships in which platform economy is inscribed. In this article, we focus on the daily conditions in which these frictional relationships are produced and reproduced, analyzing the forms that delivery work takes to grapple with these frictions and incorporate them into the practices of delivery workers. A perspective from the Global South should problematize the supposed linearity and universality of these platforms and examine local and situated modes of resisting their parameters and objectifications (Tsing 2005).
We will explore the operations deployed by the Uber Eats platform to retain, control, and monitor its "partners." The platform deploys various strategies that we will call "friendly surveillance," which operates as veiled nudges in which the system seeks to keep delivery staff engaged through incentives and promotions. This form of "friendly surveillance" is linked to the idea of "surveillance capitalism" (Zuboff 2019) in which sophisticated digital platforms seek to maximize earnings and productivity through monitoring based on smart designs and algorithms. What we will describe is related to the set of forms of resistance deployed by delivery workers, which distort what is established by the platform. While most of the critical literature on digital platforms has focused on analyzing forms of power asymmetry and surveillance (Cant 2019;Rosenblat and Stark 2016;Rosenblat 2019;Veen, Barratt, and Goods 2020;Woodcock 2020), in this article, we contribute to the debate by analyzing a less visible and documented dimension: the emergence of forms of resistance (De Certeau 1996) in response to algorithmic surveillance. Resistance emerges as a response to the constant surveillance of data, which delivers greater power to corporate and state actors (Hintz, Dencik, and Wahl-Jorgensen 2018). The fissures in surveillance are exploited by citizens, allowing for the emergence of creative uses of the digital to demand justice (Isin and Ruppert 2020). Thus, delivery resistance takes place outside the digital space, materializing through account rental and multiapping . The analysis of these practical forms of resistance (De Certeau 1996) will allow the platform to be understood from the situated experience of delivery staff, highlighting the subversive potential of these "forms of doing" (De Certeau 1996) that delivery partners deploy, generating practices that resist the control imposed by the platform's labels and controls. Through the description of these two tactics, the article shows how the platform seeks to discipline the delivery staff. However, we show that this disciplinary effort to design the delivery staff is always placed under friction by different practices that allow delivery staff to be resilient and inhabit the platform in a variety of ways to survive.
The article is organized as follows: the first section provides a review of the theory on algorithmic surveillance and platforms. The second section describes our methodological approach based on 10 semi-structured interviews and participant observation. In the third section, we present our findings and offer an analysis of the forms of surveillance that the platform deploys and the forms of resistance that delivery staff develop to coexist with the system. Finally, we conclude with reflections on the analytical and empirical opportunities that analyzing platforms through the lens of frictional relationships offers.

Surveillance on platforms
The opportunities and promises that platforms could deliver have resonated in different economies around the world. Scholars have sought to define, delimit, and characterize the deployment of platforms, surveillance, and the futures that emerge from the creation of interconnected ecosystems. Literature has delved into the forms of asymmetry established by the platforms to maintain control. Rosenblat and Stark affirm that the "Uber [driver] is empowered via information and power asymmetries to effect conditions of soft control (…)" (2016, 3759)that is, design information that seeks to direct the behavior of workers. This is expressed through heat maps, rate changes according to demand, and rewards, among others. While these elements are visible, the platform hides information necessary for workers to make informed decisions (Rosenblat 2019), such as collecting a better paid order. There are also other indicators designed for surveillance such as customer ratings and performance measures (Griesbach et al. 2019). Applications such as Uber Eats also allows buyers to monitor and punish riders who make deliveries through low/high ratings (Hidalgo and Valencia 2019).
At the same time, asymmetry and control includes several ecological factors. Veen, Barratt, and Goods (2020) propose that the algorithmic systems of the platforms are multifaceted by establishing diverse ways to keep the workforce active. Newlands (2021) proposes the concept of multimodal surveillance assemblage, with algorithmic surveillance corresponding to one of the dimensions of the surveillance framework. In addition to algorithmic architecture (algorithmic surveillance), customer surveillance needs to be considered, which not only refers to the rating system, but also to how customers are active agents by tracking the order in real time. In this line, the use of algorithms and performance metrics on the platforms reminds us of Foucault's (2000) metaphor of the panopticon: from the top of a watchtower, guards can observe every corner of the prison, but without being detected by the prisoners. This opacity achieves the illusion of constant surveillance. In the gig economy, the panopticon has moved from the architectural form of the tower to digital platforms, cloud technologies, GPS, and smartphones. According to Woodcock (2020), food delivery apps do not only seek to engage in constant surveillance, but also look to automate management. Woodcock (2020) conceptualizes platform surveillance under the concept of "algorithmic panopticon" developed by Pasquinelli (2015): "the efficacy of this approach relies upon the social power of algorithms: there is evidence of detailed supervision in the emails to workers, and discipline is enforced with occasional deactivations" (Woodcock 2020, 86).
In theory, platform workers can go online and offline whenever they wish, but their online activity can be viewed, monitored, and recorded constantly (Ticona, Mateescu, and Rosenblat 2018). Consistent with these characteristics, the concept of algorithmic governmentality (Rouvroy and Berns 2013) has been used as platforms seek to regulate individuals' behaviors and flows through programming and coding. The opaqueness of this form of surveillance and its apparent inoffensiveness comes from its subtle mode of operation. There is no explicit obligation, but it installs a digital landscape that guides and personalizes options and encourages the sensation of autonomy and freedom (Introna 2015). The process of making the functioning of technologies invisible allows people to have the sensation of moving through the interface in a transparent and natural way that is not determined by any criterion other than algorithmic objectivity (Gillespie 2014).
Other critical views on the overlap between surveillance, algorithms, and platforms, question the capitalization of personal data and the emergence of new ways of relating due to automation and personalization. Bruno (2018) establishes the concept of psychic economy of algorithms, since current capitalism is not only interested in the traces of our actions and interactions (hours connected, likes, content viewed), but also in the emotional or psychic dimensions. Thus, based on the analysis of the data extracted from their users, platforms predict and induce certain behaviors in their users. Data is the main currency of this business model (Bruno, Bentes, and Faltay 2019). Zuboff (2019) presents the concept of surveillance capitalism to describe the omniscient expansion of spaces of classification and calculability that major corporations engage in to obtain greater earnings. This new regime unilaterally captures the human experience as free primary material for the translation of behavior into numerical data (Zuboff 2019, 14). In this surveillance system, the new business of capitalism would be intimacy, seeking to seduce and profile consumers in a friendly way using increasingly personalized options, manufacturing products that seek to anticipate behaviors.
Surveillance capitalists discovered that the most-predictive behavioral data come from intervening in the state of play to nudge, coax, tune, and herd behavior toward profitable outcomes. Competitive pressures produced this shift, in which automated machine processes not only know our behavior but also shape our behavior at scale. (Zuboff 2019, 15) Behavior modification through surveillance and data gives way to a type of power that Zuboff calls "instrumentalism," which seeks to seduce or convince people to modify their behaviors in ways that benefit the goals of technology companies. This form of friendly surveillance is designed to alter surroundings and focuses on how the conditions of the physical environments in which stakeholders circulate can be changed through digital technologies (Gabrys 2016;Zuboff 2019). However, this friendly surveillance that seeks to shape behaviors and adjust them to a certain scale allows for frictions to arise, as those being surveilled do not necessarily experience surveillance passively.

Digital frictions
According to Tsing (2005), due to globalization the flow of goods, money, and people would be omnipresent and unimpeded. In this new era there would be no friction and movement would occur freely (1). Since friction can only be represented through the materiality of encounters, we argue that encounters facilitated by digital technologies and platforms, present a new opportunity to analyze how friction is embedded in digital materialities due to the global and non-situated schemes reproduced by large technology companies.
Usually, application and platform designers and programmers devote all their efforts to produce frictionless applications and interactions. Thus, "beautifully designed apps with clever functionalities offer swift introductions and make the gig economy fun and easy to use for both parties" (Prassl 2018, 16). The literature has tended to portray friction as an inherent part of the technology because the concept tends to associate interruptions with "crashes, glitches and error" (Rose 2016, 21) and technology may crash due to infrastructure failure. It can also happen that the software "crashes" because of digital decay or viruses. Anyway, software-enabled technologies are soldand usedin the full knowledge that they are "inherently partial, provisional, porous, and open to failure" (Kitchin 2014, 11).
Nevertheless, we emphasize that human labor can be disruptive in certain ecologies (Rose 2016) and actors make policy decisions about the uses of technology. Some scholars suggest understanding friction not just as a divergence but as a grip that can be navigated by users. Friction is key to the modulation of interfaces (Ash et al. 2018), and thus the interactions between the platform, surveillance, and workers will change as frictions are addressed. Along these lines, Bates (2017) examines the concept of data friction as a social phenomenon and argues that frictions "are the result of the collective decisions of human actors who experience significantly different levels of empowerment with regard to shaping the overall outcome" (Bates 2017, 425). Thinking that data can be corrupted by human action allows to underline the collaborative interactions between digital infrastructures and people (Pink et al. 2018). Friction also occurs when two or more actors and their objectives diverge, and this divergence allows the nature of friction to be unleashed in which awkward and unstable encounters occur (Tironi and Valderrama 2021).

Methodological approach
To fully describe and understand the relationship between couriers and Uber Eats, we conducted a case study before the COVID-19 outbreak, between August 2019 and January 2020, that included participant observation and semi-structured interviews with ten delivery workers.
Participant observation was carried at sites where delivery partners gathered to rest as plazas and outside fast-food restaurants. This was our first approach to the group of couriers, and we focused on understanding the work routines and daily rhythms linked to peak and off-peak hours. We also focused on understanding how the Uber Eats interface worked. Usually, the delivery staff explained how the platform worked by guiding the conversation with their smartphones in hand. Thus, we could see different processes on the platform such as rewards, payments, orders, routes, hours worked, among others. Through participant observation and informal conversations, we were able to identify the forms of resistance that emerge in the face of algorithmic surveillance.
Moreover, ten semi-structured interviews with Uber Eats delivery partners were also conducted at these sites. The duration of the interviews ranged from 15 to 45 minutes. All participants were men due to the masculinization that has already been evidenced in other studies on the gig economy (Berg et al. 2018;Hidalgo and Valencia 2019). Usually, workers who agreed to be interviewed pointed to acquaintances as people willing to be interviewed. Thus, a snowball sampling selection was generated. The selection of participants had to be adapted to the context of delivery workers. For example, a delivery person who was waiting for orders was not a candidate for the interview because his work would be interrupted. Therefore, the interviews were conducted during off-peak hours. Even so, delivery workers were free to stop the interview whenever they wished. The objective of the interviews focused on describing and condensing the know-how that the deliveries have acquired about the operation of the platform, the system of qualifications, and promotions. Participation was voluntary, and all the names of workers have been anonymized using pseudonyms. All interviews were recorded, transcribed, and coded using Atlas.ti.

Friendly surveillance on platform work
As a constituent feature of the gig economy, the concern for constant vigilance is evident in Uber Eats delivery workers, and specifically, it materializes through the rating system. According to Ticona, Mateescu, and Rosenblat (2018), "labor platforms facilitate trust between strangers in order to enable exchanges of services for pay" (5) because they are reactive systems designed to modify behavior continuously. And without going any further, ratings are essential to ensure job continuity (De Stefano 2015) which is why customers and ratings are an important part of algorithmic surveillance.
Uber's rating system is simple: once the order is received, customers can rate the delivery workers through binary options. If the delivery was good, they will press the icon that shows a thumbs-up icon; if the performance of the delivery person was unfavorable, the customer will choose the thumbs-down option. The ratings range from 0 to 100%, though most of the delivery partners we have interviewed have similar ratings: Francisco's is 99% and Mario and Juan Carlos have a rating of 95%. They say that it is rare to find accounts with low ratings because Uber Eats just deactivates them. In addition, Uber's terms and conditions are clear. To maintain good ratings, delivery staff must follow the app's recommendations regarding the parameters that affect customer satisfaction, such as speed, selecting the right route, and delivering the order in good condition. Based on the knowledge, one delivery partner describes the ratings as follows: "If you have low ratings, they can block you. If you have the highest ones, it rains orders" (Francisco, delivery worker, interview). If someone manages to maintain a "client-focused mentality" , Uber Eats will give them higher priority, and customers will tip. However, workers not only have to overcome the friction of traveling through city streets with an order that could be spoiled, but they also have to face the rating system that includes multiple factors that break with the non-moral character of the algorithms: Well … a person can rate you good or bad, because it depends on how you deliver the product. You must bring the product in good condition and everything, but sometimes you get an incomplete order because of a restaurant problem. Sometimes the store is also late in delivering the order. (Alexander, delivery worker, interview) A lot of people give you a low rating just to give you a low rating. Because they are bad people or say you got there late, took a long time, or did not go to the door. Sometimes you have nowhere to leave the bike (locked with a chain or U-lock) and you tell them to come down and they rank you as "didn't come to the door" (…). (Juan Carlos, delivery worker, interview) Other studies suggest that ratings are only part of the surveillance assembly of platforms and that clients passively participate in this surveillance (Newlands 2021). Only through a good rating will the system reward them and allow them to receive more orders. Promotions and bonuses are also an incentive to stay online on the platform. These incentives function as "nudges" and are meant to persuade workers to service high-demand areas. The piece-rate system intensifies the work and encourages couriers to take more risks to quickly move through the city and pick up the following order (Cant 2019).
Uber also penalizes certain practices that are allowed but should not become habits, like declining orders. A delivery worker explains this situation as follows: if the order is delayed due to high demand, the delivery partner can reject it because of the long wait. But he adds, "Sometimes you take everything that comes, and you sit here [outside a fast-food restaurant] for two hours" (Francisco, delivery worker, interview). This interaction of rejecting orders is turned into a cancellation rate by Uber. According to Uber Eats' terms, a small number of cancellations will help users to continue to request the services of delivery staff and restaurant partners (Uber Eats 2021). Thus, if they cancel or reject orders, they will receive fewer travel requests until the platform suspends them temporarily or even permanently (Dinegro 2019). It is precisely these mechanisms of monitoring and account deactivation that enable friction and the exploitative character of the platform.
Immersed in this universe of surveillance, all the behaviors of "delivery partners" leave digital traces so that their actions can be objects of friendly surveillance. The process of gathering digital traces makes possible the visualization of orders delivered, kilometers traveled per week, GPS routes, earnings, etc. This surveillance allows the behaviors of the delivery staff to be anticipated based on the digital fingerprints of their past performance and conduct, andwe could observeallows for biopolitical control (Foucault 1995(Foucault , 2009) on the population of delivery people: this produces docile bodies which are susceptible to "be subjected, used, transformed and improved" (Foucault 1995, 136). Thus, the platform activates a form of biopower: a form of "driving conduct" from a distance. This form of distance "persuasion" and regulation presents an ontological performance by seeking to design a certain type of "delivery subject." Under this friendly surveillance system, certain forms of being and doing are favored, and they mold and encourage to produce "good" behavior. As Uber East's metrics will always favor those who work the hardest, the system generates forms of exploitation where digital labor disables spaces of disconnection.

Tactics for resistance under the algorithmic gaze
Due to the lack of labor protection and the low payment, delivery workers have organized several protests in Santiago de Chile. Workers have received no solutions from Uber Eats or from Chilean legislators. For this reason, they have developed different tactics that allow them to remain active on the platform despite being blocked for deficient performance or for infringing terms and conditions. According to De Certeau (1996) and his theorizing on resilience, tactics are fleeting practices that use system failures and gaps to achieve a particular purpose. Thus, delivery workers have subverted the existing rules of the application by identifying different gaps that allow them to continue working to survive in the digital economy.
The first form of resistance developed by Uber workers is account rental. This is a widespread practice in Chile. It is probable that those who are deactivated from the platform will try to log back on to the platform to work, but will do so with someone else's personal data. Uber's terms and conditions prohibit account sharing: intentionally falsifying data and assuming another person's identity are prohibited (Uber Eats 2021). This is reason enough to be permanently banned from the application. However, it is possible to find posts on Facebook groups offering weekly or monthly rentals.
Anyone who wants to join the platform must go to an Uber office to buy a backpack, activate the account, and take a profile picture that allows the person's face to be identified. Some delivery staff cannot complete this process because of their migration status.
One must have a RUT (Rol Único Identificador) or ID number to work, and only those who have residency can obtain one. This process takes several months. Account rental is an "off the platform" practice (Anwar and Graham 2020), outside of the space of algorithmic surveillance and through coordination between two parties: the person who rents the account and the person who will use it. The delivery staff must arrange with a registered user or someone who can go to the office and register. According to several interviewees, the two parties often go to the office together to open an account. The person with the RUT completes the process and takes the photograph. Then the two meet outside of the office to exchange account information. Alberto, a delivery worker who works with his brother, explains that in a few cases people who have found a steady job loan their accounts out for free.
A second tactic related to rental involves having a second account, which increases the odds that the delivery staff will receive more orders. José says that he does better with two bike accounts: "Someone rents you one and you work them both at the same time using two smartphones" others prefer to work connected in applications from different companies at the same time, which Barratt, Goods, and Veen (2020) call "multi-apping." Popular platforms include PedidosYa and Rappi. However, Uber's GPS can detect them picking up the extra order and their accounts could be deactivated. They note that it is important to know when to disconnect from one app to enter another.
The very nature of the tactic, fleeting and taking advantage of failures (De Certeau 1996), becomes readily apparent in Uber's universe of metrics. Recently, Uber Eats detected rent account/sharing practices and has implemented new surveillance tools. One involves having clients report delivery staff when their profile picture does not match their face. A more radical surveillance approach involves using facial recognition software through a photograph. The delivery staff call this the selfie. The system initially asked account owners to take a selfie for the facial recognition system. The delivery staff member would disconnect from the account and then reconnect again once the owner had authenticated it with their photo: "When you are connected, the message 'your account requires a selfie' appears and the camera opens'" (Emanuel, delivery worker, interview). However, algorithmic surveillance is continuously updated: Now they changed it. I have to connect here, and you have to take the selfie on the phone you are going to use to connect. So, I have to go wherever the account owner is to take the picture. They almost always block accounts because of the selfie because a lot of people have rented accounts. (Juan Carlos, delivery worker, interview) Thus, if the face of the dealer does not match the registration photo, the account will be permanently blocked for identity theft. During participant observation, we observed one of the delivery staff outside of McDonald's with a wait time of over 45 min. While he played Call of Duty on a different smartphone, he started to complain that he was not receiving any orders even though it was 6 p.m. After a while, he showed us his profile picture and someone said, "But you look nothing alike!" This second account has a low rating, which means he is not given priority for orders: "difference in quality of service," he said. According to delivery workers, this would be the main reason for permanent deactivation. If a courier is deactivated, he or she will only receive a notification. The company is not responsible for the dismissal due to the informality in which it keeps the couriers and they are therefore left adrift to find another job. Without a source of employment, the couriers probably resort to renting an account or signing up to deliver for another company with the same precarious labor conditions. These off-the-platform resistances are crucial, since delivery workers not only experience resistances, but also look for ways to survive and generate forms of solidarity among workers. In this line, Soriano and Cabañes (2020) propose the concept of entrepreneurial solidarities, that is, that "digital workers do not passively and simplistically accept neoliberal discourses about digital labor, what with them helping each other to game a system that they feel games on them as well" (2). Thus, the tactics are not only a mechanism of resistance, but also a solidarity response to the limited job opportunities in Chile: "There are always people who need help. At the end of the day, those of us who have documents do it to support other members of the community," Jesús explains. Without account rentals, many would be unable to work, and they can only process their permanent residency in Chile through contributions made from paid work.
Once the tactic is identified due to constant monitoring, the platform will update with a new feature that will fix the fissure. As Tsing (2005, 6) states "Friction is required to keep global power in motion. It shows us where the rubber meets the road". The digital world is a scenario that perfectly illustrates this path: Uber Eats enables a platform that allows people to find a job quickly and eliminates traditional job filters such as job interviews or resumes. The platform also benefits from friction in some ways by allowing programmers to generate iterations and technologies that are more difficult to crack. In this way, friction allows the wheel to keep spinning.

Final reflections
The perspective that we have tried to develop to address Uber Eats, and this emerging platform-mediated urban condition is meant to think through the digitalization of the city and its services in its frictional nature. Instead of approaching the platform as an objectified category or as the natural/universal result of technological progress, we address the frictions and local practices of reproduction and incorporation of Uber Eats. The endless number of data and digitized orders that gives life and form to the Uber Eats platform rest on creative practices performed by delivery staff.
Our challenge was not simply to recall that the digital information that circulates is material and situated, but to insist on the intrinsic dependence that exists between the digital and friction, exploring the type of capacities that these frictions produce between territory and the digital in terms of surveillance. In this sense, frictions serve to reveal different dimensions related to platform operation. On one hand, the constant deactivation of accounts and the need to lease a new one allow us to view the multiple relationships of power or inequality in which platform and workers are inscribed, configuring forms of exploitation and surveillance developed to maintain its "correct" operation. On the other hand, paying attention to frictions brings to the foreground the work in which delivery staff engage to survive and coexist with the difficulties that emerge in their daily practices, developing forms of astuteness and skills that challenge the platform's surveillance.
While digital platforms are promoted as a space for entrepreneurship and selfemployment, exalting the figures of the "independent worker," "partner," and "micro business owner," we have shown how the operations present in the platform are based on active practices of surveillance and remote monitoring. Canceling orders or receiving negative evaluations from clients may lead to account suspension, which is a "digital" word for "firing." Uber Eats deploys what we have called "friendly surveillance," that is, they operate in a personalized and automated, gamified, and indirect manner, promoting an apparent experience of equality and participation among delivery staff. However, their effects materialize the precariousness of delivery partners' work conditions.
We argue that the specific normativities inscribed in the platform, which are based on the idea of a flexible, autonomous, and self-sufficient subject, activate this form of biopower: a form of "driving conduct" from a distance. This form of distance "persuasion" and regulation that the Uber platform activates presents an ontological performance by seeking to design a certain type of "delivery subject." Under this remote regulatory power, certain forms of being and doing are favored, and they mold and encourage to produce "good" behavior. In contrast to surveillance techniques developed by the Uber platform, we have sought to emphasize and empirically describe the frictions, forms of resistance and practices that delivery staff develop to coexist with the platform's controls. Delivery staff do not endure the instrumentalization of surveillance passively. Rather, they deploy a range of practical knowledge and come "from below" to dispute and undermine the restrictions mobilized by Uber Eats such as account rental, multi-apping or generate new offline arrangements to prevent the facial recognition system from discovering them. Through the observation of these practices, we have affirmed that the platform's operation cannot be defined homogeneously and that there is a field of frictions, rubs, and tensions that are always open to fissures.
We show how delivery staff use the failures and fissures of the logics under which that workspace seeks to be controlled. We note the fundamentally unpredictable nature of the platform, which continues to change and deviate despite the control that it seeks to algorithmically plan. The tactics described, that is, the ways in which the delivery staff understand and reinterpret certain forms of action under the gaze of the platform, are fundamental for the creation of certain types of resistance. Through tactics of resistance which are often implicit and inscribed in routines -delivery staff strengthen a sort of capacity for action in the face of digital control. This includes practices such as account rentals, which can be considered valuable practices in the work of delivery staff in that they generate networks of support among workers. Finally, tactics and features of the platform will change according to the local needs of the delivery workers and the guidelines set out in the Global North, producing new ecologies in friction.