Working the algorithm: Contextual skills of on-demand gig workers

Abstract

This study explores the algorithmic skills of on-demand gig workers, such as food delivery workers and ride-hailing drivers, who navigate algorithmic management on a daily basis. While algorithms often constrain worker autonomy and reduce labor processes to standardized routines, we argue that gig workers cultivate practical algorithmic skills through their experiences. Drawing on interviews with 20 workers, we identify three dimensions of algorithmic skills: algorithmic awareness, algorithm learning and comprehension, and algorithm utilization and mastery, encompassing nine specific indicators: awareness of algorithm presence, recognition of data dependency, awareness of algorithmic evolution, understanding input–output relationships, reverse engineering, collaborative learning of algorithms, exploiting platform rules, leveraging technical tools, and experience-based decision-making. These skills afford workers meaningful agency, enabling them to navigate and occasionally challenge platform control. Compared with analyses rooted in the control-resistance framework, the algorithmic skill framework offers a more constructive and sustainable pathway for the future of human–algorithm collaboration in workplace contexts. This study also highlights the need to contextualize algorithmic skills within specific sociotechnical and occupational frameworks.

Keywords

On-demand gig work digital work algorithmic skills algorithmic management worker agency algorithmic resistance

Introduction

Algorithms have fundamentally transformed how work is organized, distributed, and managed (Jarrahi et al., 2021). They increasingly substitute for human managers in critical aspects of work, including task allocation and coordination, performance evaluation and rewards, and behavioral alignment with organizational goals (Duggan et al., 2023). As a result, workers must engage in continuous interactions with algorithms as a central part of their jobs (Jarrahi and Sutherland, 2019). This mode of “algorithmic management” or “algorithmic control” enhances operational efficiency, often at the cost of worker agency and skill development (Lata et al., 2023; Wood et al., 2019; Gray and Suri, 2019).

Digital gig work under algorithmic management takes diverse forms. It encompasses crowdwork in the remote gig economy—including freelancing, micro-tasking, and online content creation—as well as on-demand work mediated by apps, such as ride-hailing and food delivery (De Groen and Maselli, 2016). The latter is particularly distinctive in that the work is managed online but performed offline. Scholars often distinguish between high-skill and low-skill forms of labor. Short, repetitive, and routine tasks, or so-called “click work,” fall into the low-skill category, alongside occupations such as delivery riders and ride-hailing drivers (De Groen and Maselli, 2016; Cini, 2023).

In China, barriers to entry in the food delivery and ride-hailing sectors are relatively low: workers’ prior education, skills, and other forms of human capital are rarely prerequisites for entry. In the food delivery industry, for instance, many workers migrate from rural areas and previously worked in low-skill, low-entry-barrier sectors such as manufacturing or commercial services (Zhu et al., 2023; Meituan Research Institute, 2023). Most view this work as temporary, in part because they perceive limited opportunities to accumulate transferable skills (Zhu et al., 2023; Vallas and Schor, 2020; Huws et al., 2016; Qiu, 2025). However, the scale of these industries is rapidly expanding. According to Statista, revenue in the online food delivery market is projected to reach USD 1.42 trillion by 2025 (Statista, 2024). In China, the number of workers in new forms of employment—including ride-hailing drivers, food delivery workers, and couriers—reached 84 million in 2023 (All-China Federation of Trade Unions, 2023). This tension—between a rapidly expanding workforce and the perceived absence of skill development—raises a critical question: can algorithmically managed gig work foster systematic, transferable skills? Without skill accumulation, millions of workers risk long-term precarity in an increasingly digitalized labor market, in which familiarity with algorithmic systems is increasingly expected. Conversely, if gig workers do develop skills for navigating algorithmic systems, such competencies could open pathways to upward mobility, reducing their vulnerability to arbitrary platform control.

Importantly, most work on-demand via apps involves conventional tasks—such as driving or delivering food—that are not themselves transformed by digital technologies. These traditional skills are not our focus. Rather, we focus on the new forms of skill development that emerge when workers’ livelihoods depend on algorithmic management and platform rules, and when they are highly motivated to understand and adapt to them (Klawitter and Hargittai, 2018). Specifically, we ask whether workers cultivate algorithmic and platform-related skills in the course of their work, and how these skills may help them adapt to an algorithmic society and gain a sense of control over their work.

Previous studies on algorithmic literacy or skills have largely examined users’ everyday interactions with algorithms. While many studies highlight the importance of algorithmic awareness, knowledge, literacy, or skill (e.g., Devito, 2021; Dogruel et al., 2022; Oeldorf-Hirsch and Neubaum, 2023; Gruber and Hargittai, 2023; Gagrčin et al., 2026), few connect these ideas to workers’ agentic practices or explore how algorithm-related skills are cultivated and applied in the workplace. Research on delivery riders and ride-hailing drivers has largely gravitated toward the control-resistance paradigm, emphasizing how workers reclaim autonomy through collective action or alternative strategies that are not directly related to technology (e.g., Lata et al., 2023; Maffie, 2023; Heiland, 2021). What remains lacking is a more constructive perspective that highlights how workers develop practical skills in response to algorithmic management.

Against this backdrop, we draw on interviews with 20 food delivery workers and ride-hailing drivers. We ask two guiding research questions: (1) What dimensions of algorithm-related skills are developed by so-called “low-skill” gig workers in their everyday labor? (2) How do these skills enable workers to exercise agency within the constraints of algorithmic management?

Literature review

Algorithmic management and digital gig work

Algorithmic management refers to the use of algorithms by platforms to replace human decision-making in work-related processes such as task allocation, performance evaluation, and rule enforcement (Duggan et al., 2020; Kellogg et al., 2020; Van Doorn, 2017). This system enables platforms to match workers with clients at scale, while simultaneously monitoring their performance (Rosenblat, 2018; Prassl, 2018; Gandini, 2019). Beyond direct control, platforms employ gamification to make rules and compensation structures more visible, nudging workers toward compliance and mutual competition (Cameron, 2022; Healy et al., 2017).

Algorithmic management reduces platforms’ operational costs while presenting workers with a system that appears neutral, objective, and—to some—more trustworthy than human managers (Lee et al., 2015; Lin and de Kloet, 2019). Yet scholars caution that under digital capitalism, labor exploitation has grown more subtle, mediated by the very platforms that promise efficiency and fairness (Scholz, 2017). Algorithmic management erodes autonomy, produces significant asymmetries of information and power, and can arbitrarily penalize or terminate workers (Duggan et al., 2020; Kadolkar et al., 2024; Purcell and Brook, 2022; Shapiro, 2018). Platforms’ constant data collection creates a panoptic system of surveillance (Newlands, 2021), in which gig workers not only perform physical labor but also function as invisible data producers who sustain the platform's algorithmic infrastructure (Chen, 2022). Scholars argue this dynamic produces new class divisions: what Burrell and Fourcade (2021) term a “coding elite” that controls data and software, and a “cybertariat” whose labor feeds and trains the very algorithms that may eventually replace them. This pervasive control undermines workers’ sense of autonomy (Glavin et al., 2021) and harms their well-being (Wood et al., 2019).

Against this backdrop, a key question concerns how workers reclaim autonomy and agency in the gig economy. Past work has largely focused on resistance to algorithmic control. Studies highlight how workers leverage supportive communities and political traditions to mobilize (Cini, 2023), bypass platforms by directly contacting clients (Cameron and Rahman, 2022), or establish informal “off-platform” practices such as unregistered ride-hailing services to reduce platform dependence (Maffie, 2023). Collective action has also been emphasized as a vital strategy of resistance (Heiland, 2021). However, these studies primarily frame agency as rooted in social and organizational practices, such as interpersonal networks or collective resistance, rather than in workers’ everyday engagements with platform technologies. Moreover, as Howcroft and Bergvall-Kåreborn (2019) note, platforms’ classification of workers as independent contractors often isolates them, making traditional forms of collective resistance, such as strikes, difficult and ineffective.

Rather than focusing solely on resistance, this study foregrounds a skills-based perspective, asking how gig workers may benefit from work by developing algorithmic skills. By shifting attention from collective resistance to the micro-level cultivation of competencies in navigating algorithms, this study highlights an underexplored yet constructive dimension of agency. Such a perspective allows us to examine how workers build practical skills through continuous interaction with algorithmic systems, thereby carving out space for autonomy, improving their earning capacity, and fostering occupational identity within the constraints of platform labor.

Algorithmic skills and worker agency

Several studies have highlighted the role of digital skills in shaping workers’ agency in the gig economy. Fiers (2024) shows that freelancers with strong digital skills can creatively exploit platform rules to secure more opportunities. Similarly, food delivery workers have been found to employ tactics such as creating fake orders, transferring or selectively accepting tasks, and operating on multiple platforms to increase their earnings and gain greater control over task selection (Sun, 2019). Chen (2018) demonstrates that ride-hailing drivers in China enhance their working conditions not only through strikes but also through “proactive algorithm usage,” including plug-ins and multi-platform strategies. Ferrari and Graham (2021) further categorize such practices into manipulation, subversion, and disruption, highlighting the diverse ways workers carve out autonomy under algorithmic control. Yet these studies either address digital skills broadly without focusing on algorithms specifically, or they stop short of connecting workers’ digital practices to a systematic framework of algorithmic skills.

As Bucher (2017: 42) notes, “while algorithms certainly do things to people, people also do things to algorithms.” However, algorithmic skills remain inconsistently defined. Scholarship has generated a cluster of related concepts—algorithmic awareness, knowledge, literacy, and competencies—though mostly in the context of ordinary users’ interactions with digital platforms. For instance, algorithmic awareness captures users’ recognition of algorithms’ role in shaping information and recommendations (Eslami et al., 2015; Zarouali et al., 2021). Algorithmic knowledge builds on this by encompassing an understanding of how algorithms function (Cotter and Reisdorf, 2020). Dogruel et al. (2022: 118) define algorithmic literacy as “being aware of the use of algorithms in online applications, platforms, and services, knowing how algorithms work, being able to critically evaluate algorithmic decision-making as well as having the skills to cope with or even influence algorithmic operations.” Klawitter and Hargittai (2018) define algorithmic skills as the ability to discern how specific algorithms operate and to leverage that knowledge in content production and decision-making. Jarrahi and Sutherland (2019) extend this notion into gig work by conceptualizing algorithmic competencies as central to platform labor. Taken together, these studies point to four dimensions of algorithm-related capacities: awareness, knowledge, skills/competencies, and critical thinking.

Drawing on this body of work, we define algorithmic skills as the awareness, knowledge, and practices that on-demand gig workers develop in their everyday labor, with particular emphasis on how they interact with platform algorithms, engage in diverse forms of experimentation and adaptation, and critically reflect on algorithmic systems. This definition, we argue, offers three contributions to the existing literature.

Distinction from general digital skills. Definitions and scale designs of digital skills center on operational proficiency in various aspects of digital technologies. For instance, Van Deursen et al. (2014) developed a framework encompassing five dimensions of digital skills: operational, information navigation, social, creative, and mobile skills. The measurements focus on concrete, standardized operations—such as the statement “I know how to open downloaded files”—which can be easily judged against clear criteria.

Algorithms are frequently described as “black boxes” (Pasquale, 2015). Their opacity stems from multiple sources: continual evolution (Gillespie, 2014), dependence on user-generated input (Shin, 2021), and platforms’ deliberate non-disclosure (Hargittai et al., 2020). Take, for example, the causal relationship from algorithmic inputs to outputs: different individuals form divergent judgments on this, and even the programmers who build the algorithms can rarely grasp the full picture. As such, the conceptualization and operationalization of algorithmic skills can never be universal; instead, they must be understood as situated and interpretive. They incorporate not only workers’ understanding of the technology itself, but also their perceptions of platforms, industry rules, and the practical demands of their work. It is therefore necessary to explore algorithmic skills in relation to the specific characteristics of occupations that involve sustained interaction with algorithms. This is also why existing measures of algorithmic literacy or skills, developed primarily with ordinary users in mind, cannot be directly applied to on-demand gig workers.

From knowledge to practice. Most previous studies emphasize users’ awareness or knowledge of algorithms, while paying limited attention to practice-oriented dimensions of algorithmic skills, especially within work settings. Even when such practices are studied, the focus is disproportionately on “highly skilled” workers such as content creators and freelancers, rather than so-called “low-skill” gig workers like delivery riders and drivers.

Processual nature of skills. Research on gig literacies suggests that skills are not fixed outcomes but ongoing processes of problem-solving and adaptation (Sutherland et al., 2020; Lloyd, 2014). In the context of ever-evolving algorithms, examining gig workers’ algorithmic skills requires attention to the pathways of understanding and learning—not just end results. Prior research on algorithmic skills has paid insufficient attention to this developmental, process-oriented dimension.

Algorithmic skills play a crucial role in shaping gig workers’ agency. Agency—the capacity to make independent choices and enact change (Schwartz and Mahnke, 2021)—is central to how workers navigate algorithmic management. Bonini and Treré (2024) introduce the concept of algorithmic agency, referring to workers’ ability to maneuver within and manipulate platform constraints to gain more control than formal rules allow. A deeper understanding of algorithms and decision-making structures enables workers to exercise autonomy and self-determination (Cotter and Reisdorf, 2020; Dogruel et al., 2022; Cotter, 2019; Klawitter and Hargittai, 2018). This resonates with Schuppan's (2014) argument that technological change does not simply de-skill or upskill workers, but often drives reskilling—a process our concept of algorithmic skills aims to capture.

We argue that algorithmic skills are central to how workers exercise agency under platform control. At the same time, we acknowledge the coexistence of platform control and worker autonomy. This duality arises from the decentralized nature of platform management and from workers’ capacity to interpret, exploit, and even reverse algorithmic rules. If algorithms act as instruments of labor discipline, workers’ strategic engagements with them represent an important site of negotiation, revealing the limits of platform control.

Research context

We situate this study in the Chinese context for several reasons. First, most previous work on algorithmic management and gig work has centered on Western settings, with few studies focusing on China or East Asia. Second, on-demand gig work has boomed in China. The official framing of this sector as “new forms of employment” signals the country's investment in the growth of China's digital economy (Chen and Soriano, 2022). Yet the country’s platform economy remains dominated by informal workers lacking collective bargaining power, formal contracts, or employment benefits (Chen, 2018; Sun, 2019). This raises the question of whether these workers can develop skills that expand their opportunities in an increasingly algorithmic labor market.

China's platform economy has distinct characteristics that make it a particularly suitable setting in which to develop a skills-based alternative to the prevalent resistance framework. First, China's gig sector features low worker organization, with platforms wielding significant power through capital and technological advantages. Despite strong government regulation, competitive pressures and the normalization of long working hours drive Chinese gig workers to adopt survival-oriented strategies. Instead of contesting algorithmic rules directly, they develop practical workarounds to earn more and get penalized less, making algorithmic skill a practical survival tool. Second, China's collectivist culture and “guanxi” norms foster informal knowledge-sharing networks—such as WeChat groups and mentor–apprentice relationships—where algorithmic know-how is developed through both individual experimentation and peer exchange. While these features should not be treated as absolute—Chinese workers do also engage in collective action and organized resistance—the overall institutional context makes a skills-based framework especially pertinent.

China's largest food delivery platforms are Meituan and Ele.me, while Didi dominates the ride-hailing sector. The labor processes of food delivery and ride-hailing share many similarities. In both cases, workers wait for the platform to assign orders, occasionally “grab” orders themselves, and retain limited authority to cancel or transfer tasks. They are classified into hierarchical “levels” based on performance indicators such as time spent online and customer ratings. Across both sectors, workers operate under close algorithmic surveillance and management.

Methodology

Study design

We conducted semi-structured, in-depth interviews. Given the opacity of algorithmic systems (Hargittai et al., 2020), algorithmic skills constitute tacit, experiential knowledge that is difficult to capture through standardized surveys. Semi-structured interviews offer the flexibility to track emerging themes while probing individual sense-making processes, making them well suited to our inductive, framework-building approach.

Since most on-demand gig workers are unfamiliar with the specialized term “algorithm,” we avoided using it at the outset of our interviews. Instead, we prompted discussion using more familiar terms such as “system,” “platform,” and “rules” to encourage participants to reflect on their experiences (Hargittai et al., 2020). The interview guide covered participants’ work backgrounds and platform experience; their understanding of dispatch rules, scoring mechanisms, and income calculation; their attitudes and emotional responses toward platform governance; and their proactive strategies for navigating algorithmic constraints. The interviews, conducted via voice calls or face-to-face, lasted 1–2 hours each. We assigned each participant an alphanumeric code (e.g., Z3) to ensure anonymity.

Data collection

We recruited 20 on-demand gig workers—10 food delivery couriers and 10 ride-hailing drivers—through snowball sampling. Initial participants were recruited through the researchers’ personal networks and subsequently helped facilitate further recruitment by introducing us to local online chat groups and referring colleagues. Informed consent was obtained prior to all interviews. Participants’ work experience ranged from 2 months to 5 years; most had worked on more than one platform. Most were full-time workers, though some worked part-time.

Although we sought demographic diversity, our final sample included no female participants. This constitutes a limitation. However, given that the vast majority of food delivery couriers and ride-hailing drivers in China are male (Zhu et al., 2023), and considering that our analysis does not address gender differences in algorithmic skills, we believe the findings still capture the predominant patterns of how workers in these sectors understand and engage with algorithms. Table 1 provides detailed participant information.

Table 1.

Participant information.

Measure	Item	Frequency	Percentage
Primary platform	Meituan	6	30%
	Didi	5	25%
	Ele.me	4	20%
	Other platforms	2	10%
	Multi-platform	3	15%
Employment status	Full time	17	85%
Employment status	Part time	3	15%
Duration in role	Less than 1 year	2	10%
	1–2 years	6	30%
	2–3 years	9	45%
	More than 3 years	3	15%

Data analysis

Following the grounded theory procedures outlined by Strauss and Corbin (1998), we analyzed the interview data using NVivo 12, conducting open and axial coding to develop core concepts. This process yielded 90 first-level codes and 10 major categories.

In the first stage—open coding—we focused on algorithm-related terms used by participants, such as “data,” “calculation,” “rankings,” and “machines.” From these, we derived initial concepts like “data centrality,” “performance-based rewards,” “sense of fairness,” and “gamified bonuses.” Previous research suggests that algorithmic knowledge is not only conceptual but also embedded in practice (Bishop, 2018; Cotter, 2024). This informed our approach to the second stage—axial coding—in which we organized participants’ algorithmic skills into two initial dimensions: “knowledge” and “action.” The “knowledge” dimension refers to their understanding of algorithmic principles, while “action” encompasses the practices they adopt during work based on this understanding. Under these core dimensions, we identified nine subcategories.

As Cotter (2024) argues, practical knowledge is inseparable from action. Consistent with this view, our analysis revealed that the boundaries between the “knowledge” and “action” dimensions were not always distinct, prompting us to restructure our framework. We finally identified three key dimensions of algorithmic skills: Algorithmic Awareness, Algorithm Learning and Comprehension, and Algorithm Utilization and Mastery. These three dimensions comprise nine indicators, detailed in Figure 1. The following sections elaborate on each dimension and the relationships among them.

Figure 1.

Dimensions and indicators of algorithmic skills in the on-demand gig work context.

Results

Algorithmic awareness

The first level of algorithmic skills is algorithmic awareness, as being conscious of a technology is the foundational step toward understanding and mastering it. Within this level, we identify three key indicators of algorithmic skills: Awareness of Algorithm Presence, Recognition of Data Dependency, and Awareness of Algorithm Evolution.

Awareness of algorithm presence

The foundational step in developing algorithmic skills among on-demand gig workers is recognizing the presence and operation of algorithms in their work processes. Many on-demand gig workers may not be familiar with the term “algorithm,” yet they recognize its presence through alternative terms such as “machine,” “computer program,” “system,” or “backend,” which refer to the same underlying system in their everyday parlance. As Interviewee Z3 stated: “We don’t really understand these rules, and the platform won’t tell you. We just know that the orders are dispatched by the system, not by a person.”

In the Chinese context, the term 算法 (algorithm) literally translates to “the rules of calculation.” This literal meaning allows workers, even those without specialized knowledge of computer science, to form associations based on their lived experiences. These associations take two broad forms. The first aligns with mathematical rules, particularly those governing income distribution and order volume. The second anthropomorphizes the system, imagining the algorithm as a calculating agent that actively strategizes around its labor. As Interviewee Z2 stated, “The system evaluates what kind of person you are and calculates everything meticulously. While we take care of customers, we also have to take care of the system.”

Recognition of data dependency

This indicator refers to whether on-demand gig workers recognize the connection between data input and the output. Our interviews suggest that most workers have a clear awareness of how their data feeds into algorithmic operations. For example, Interviewee S1 mentioned, “In this line of work, data is key. The system uses your data to assign orders.” Many on-demand gig workers are well aware that platforms and algorithms are constantly collecting their data—such as their performance rankings (Interviewee W1), on-time delivery rates (Interviewee T1), and complaint records (Interviewee L3). Workers frequently emphasized that data is central to the system's functioning. In China, public discourse around “big data” has been more prominent than discussions of “algorithms,” which may partly explain workers’ heightened sensitivity to data collection.

Workers’ awareness that their data drives algorithmic operations suggests a more active relationship with the system than the passive “data subject” framing implies. As one interviewee, C2, explained:

The data provided by riders constantly feeds and refines the algorithm. For example, top-tier riders might not rely heavily on system-dispatched orders. However, if everyone opted out of system dispatching and relied solely on manual order grabbing, then the lack of fresh data input would cause the algorithm to stagnate.

C2's account reflects an awareness of algorithmic interdependence—a recognition that workers’ actions continuously supply the system with data that shapes future decision-making. Recognizing the role of data in platform operations is a further demonstration of algorithmic skills.

Awareness of algorithmic evolution

Awareness of algorithmic evolution refers to workers’ recognition that the algorithms governing their work are not static but undergo continual change. Algorithms, particularly those based on machine learning, are continuously updated and refined, drawing on new data and instructions to improve their functionality. Workers who have developed stronger algorithmic awareness tend to recognize these shifts and adjust their strategies accordingly. This indicator captures a general awareness that algorithms are dynamic, rather than knowledge of any specific code-level modification.

However, the specific perceptions of algorithmic evolution among participants vary significantly. Some workers view these changes as positive and perceive the algorithms as becoming more considerate and user-friendly. For instance, Interviewee L3 noted, “The platform's algorithms have improved a lot—they’re more considerate of us now.”

Conversely, other workers express frustration, believing that each update introduces new inefficiencies. As Interviewee Z6 explained, “The algorithm is constantly being optimized, but honestly, each optimization might solve a small problem while introducing a bigger bug.” Interviewee T1 echoed this sentiment, noting: “In the past, most orders were conveniently routed, but now random dispatching is far more common—we often get one order on the east side and another on the west.”

These divergent interpretations suggest more than just technical perceptions. Instead, they reveal how on-demand gig workers view the broader social forces—such as the platform itself and its managers—shaping these algorithms.

Algorithm learning and comprehension

The second level of algorithmic skills we observed is algorithm learning and comprehension, which refers to workers’ capacity to acquire and make sense of information about how algorithms operate. Given the opacity of algorithmic systems, no worker—nor indeed any single actor—can claim a complete or objectively “correct” understanding of how they function. Nevertheless, we identify three indicators that mark meaningful progress toward such understanding: grasping basic input–output relationships within algorithmic processes, experimenting to infer how algorithms function, and drawing on diverse resources to learn about algorithmic operations. The latter two indicators emphasize the process of knowledge acquisition rather than its outcomes, and in practice they often contribute to the first. We suggest that the effort to engage with algorithms is itself significant: because algorithms are adaptive and continually evolving, the skills needed to navigate them are necessarily ongoing and iterative rather than fixed. Together, these indicators lay the groundwork for a deeper, if always partial, understanding of algorithmic systems.

Understanding input–output relationships

Understanding input–output relationships refers to on-demand gig workers’ perceptions of the factors that influence algorithmic outputs, such as whether the order acceptance rate impacts the system's order assignment. During interviews, workers frequently mentioned factors such as duration in role (Interviewee S1), level (Interviewee L2), and order acceptance rate (Interviewee Y1) as significant “causes,” while “effects” included income (Interviewee Z3), the volume of assigned orders (Interviewee L3), and order quality (Interviewee L1). This capacity to reason about input–output relationships represents a deeper engagement with algorithmic logic than the awareness discussed above. Interviewee S1 identified three factors that he believed shape order dispatch logic:

It's essentially three things: First, time — you definitely need to spend more time than others to take more orders. Second, data — our system has levels like bronze, gold, diamond, up to the king level. The higher your level, the higher your priority in order assignment. Third, the platform tends to assign orders based on your average efficiency (how long it takes per order). If your reputation is good, with no negative reviews or complaints, you’ll get priority in order distribution. (Interviewee S1)

From the views expressed by on-demand gig workers during interviews, their perceptions of how the platform's task allocation algorithm operates can be broadly divided into two categories. The first group emphasizes factors such as level and past performance, viewing these as the most significant influences. They believe that differences in workers’ abilities create stratification among them. The second group subscribes to a “work more, earn more” logic, asserting that online time is the most critical factor influencing final income. Workers in this group often describe the platform's algorithm as “fair” or “balanced,” believing that the system evens out order distribution over time. However, regardless of their perspective, both are rooted in a fundamental understanding of the causal relationships within the task allocation system.

Algorithms, as sociotechnical systems, are shaped through negotiation among various stakeholders (Van der Nagel, 2018). This characteristic creates space for on-demand gig workers to exercise agency, as they recognize that their own actions and behaviors also play a role in shaping algorithmic operations. From the perspective of boundary resource theory, the algorithmic rules and incentive mechanisms that govern gig workers can be defined as core boundary resources of the platform—interfaces that connect platform designers and workers and are subject to dynamic adjustments through multi-party interactions (Eaton et al., 2015). Despite platforms’ persistent efforts to mold gig workers into obedient “sheep” through training and rule enforcement, workers frequently question whether platform guidance is technically justified or merely designed to serve the platform's interests. As Interviewee C2 remarked: “They [Platforms] say rejecting orders affects the algorithm and reduces the likelihood of getting future orders. But I doubt that—I think Meituan is just trying to scare riders into rejecting fewer orders. I don’t think it actually makes any difference.” This skepticism illustrates how gig workers often interpret platform directives differently, challenging official narratives and instead relying on their own experiences to navigate algorithmic systems. This ongoing negotiation constitutes what Eaton et al. (2015) term a “distributed tuning process.” Workers assert agency not by defying the system outright, but by strategically accommodating algorithmic control while maintaining their own interpretive independence.

Reverse engineering

Reverse engineering is a widely utilized analytical technique in fields such as software engineering and industrial design. Its core principle lies in deducing the original design process or logic from an existing product or system (Canfora and Di Penta, 2007). Sun (2019) described the process of understanding algorithmic management through labor practices as the “reverse engineering” of algorithms.

In this study, we define reverse engineering as the process of inferring the operational logic of algorithms by analyzing various inputs and corresponding outputs. For instance, some workers manipulate specific variables and compare differing outcomes to deduce causal relationships. This practice enables workers to gain a more comprehensive understanding of the technology, thereby allowing them to adjust their working patterns more purposefully. For example, Interviewee L2 described experimenting with a colleague to test the logic of task assignment:

If you are in an area with few tasks but many drivers, task allocation is based on your service score. We confirmed this through an experiment: my colleague and I placed our phones side by side; he had a higher service score and was assigned larger tasks. However, when there are many tasks but few drivers, the allocation seems random.

Past work has highlighted that algorithms are technologies learned through use (Cotter and Reisdorf, 2020; Swart, 2021). On-demand gig workers employ reverse engineering to learn algorithmic logic from the bottom up through experiential practices. They then adapt their work methods based on these inferred patterns, which can strengthen their sense of control over the labor process. Moreover, this approach may partially offset the information asymmetry between workers and platforms. Through such practices, workers move beyond passive compliance toward active interpretation of algorithmic logic, gaining a degree of informational agency within the system.

Collaborative learning of algorithms

In addition to employing reverse engineering to understand algorithms, on-demand gig workers also draw on indirect channels to learn about algorithmic operations. Based on our interviews, we identified three primary methods: (1) Interpersonal Channels: Workers share tips and observations about platform algorithms through conversations in chat groups or online forums. (2) Online Channels: Workers access algorithmic insights by reading industry-related articles and watching videos on social media platforms. (3) Official Channels: Workers learn and understand algorithmic logic through official guidelines and training materials provided by the platforms. Workers’ experience with algorithms outside of job-related platforms also helps them accumulate algorithm-related knowledge and insights. For example, Interviewee C2 said he was no stranger to algorithms. When using shopping sites like Taobao and Pinduoduo, he noticed that “the items recommended are all related to what you’ve browsed before; you can easily pick up on that after using them for a while.” This awareness may lack technical precision, but it indicates a broader algorithmic sensibility that extends beyond the workplace. “I may not know exactly how the underlying algorithm works,” he explained, “but I do know it's exerting some kind of control—like something is guessing what you like.”

Many interviewees mentioned sharing personal work experience through interpersonal communication and pooling collective wisdom to boost their work efficiency and earnings. They noted joining online groups for gig workers or chatting on platform forums, where they exchanged a wide range of tips—including where to find high-value orders (Interviewee W1), the best times to secure orders (Interviewee Z6), and how to get the system to assign better-paying tasks (Interviewee L3).

In recent years, the rapid expansion of China's food delivery and ride-hailing industries has sparked widespread online discussion, with numerous media reports and academic studies highlighting the conditions of digital work. As the term “algorithm” gains prominence in such accounts, many workers have developed a greater familiarity with the concept. However, workers do not passively accept all information they encounter from various sources. Instead, they critically assess its credibility based on their own work experiences. As interviewee O1 remarked: “I’ve heard about algorithms before—there's been a lot of talk about it last year or this year. Honestly, it's not as bad as these articles described, like being ‘trapped in the system’.”

Algorithm utilization and mastery

The highest level of algorithmic skills for on-demand gig workers is algorithm utilization and mastery, which entails not only understanding how algorithms work but also translating that understanding into strategic action within the constraints of platform governance. The indicators discussed below are more specific to the on-demand gig work context than those outlined above.

Exploiting platform rules

Exploiting platform rules refers to leveraging a thorough understanding of system rules to streamline work processes and enhance convenience. As previously discussed, while on-demand gig workers may not be familiar with the technical term “algorithm,” they are highly knowledgeable about platform rules. Because algorithms operationalize and enforce platform rules, a worker's ability to exploit those rules reflects an applied understanding of the algorithmic system. Therefore, the ability of on-demand gig workers to exploit platform rules for personal gain is a manifestation of their algorithmic skills. Interviewee Y1 described a commonly used strategy:

Many people hang around airports refreshing orders, accepting the good ones and ignoring the bad ones. But the platform's backend monitors this. There's a trick: instead of canceling an unwanted order, you refresh the screen, and after about 20 seconds, the order disappears without affecting your acceptance rate. Some people like using this method to secure large orders.

Other strategies, such as “setting preferred routes” (Interviewee Z2), “multi-platform order-taking” (Interviewees W1 and L1), or “batching orders” (Interviewee C2) are also widely used among on-demand gig workers. However, since platforms continuously update their algorithms and policies, many of these loophole-exploiting strategies are subject to monitoring and potential penalties. Thus, these examples do not represent long-term, sustainable practices.

Despite constant rule changes, workers consistently find new ways to adapt—though always within narrow margins. These adaptive practices are rooted in workers’ perceptions and understanding of platform algorithms, transforming abstract knowledge into actionable skills.

Leveraging technical tools

Leveraging technical tools refers to the practice of on-demand gig workers utilizing additional technological tools to “interfere with” or “manipulate” algorithmic processes, maximizing their personal benefits. We include this dimension in the concept of “algorithmic skills” because, while understanding and adapting to algorithmic logic is an essential aspect of skills, the ability to resist the algorithm's influence and even exploit it for one's advantage is a strong demonstration of the agency of on-demand gig workers.

Specifically, on-demand gig workers engage with technical tools in two main ways to influence algorithms: (1) Enhancing Hardware: Workers invest in more powerful devices, such as smartphones with better functionality and faster internet connections, to increase their chances of securing desirable orders (Interviewee W1, C1). (2) Using Third Party/External Tools: Some employ plugins or external software, often prohibited by platforms, to disrupt the algorithm's original task-assignment logic and selectively target high-value or large orders (Interviewee Y1, L2). Interviewee Z5 used a vivid analogy to describe this dynamic, likening orders to fish in a pond: “Using hacks is like this: the system assigns fish to different slots—it's all calculated. But people using hacks can see in advance which fish are better and pull them into their slot.”

Experience-based decision-making

Experience-based decision-making is the most integrative indicator in our framework, as it draws on and synthesises the skills described in previous sections. It refers to how workers draw on accumulated experience to work more effectively within—or independently of—algorithmic systems. In some cases, workers bypass algorithmic guidance entirely, relying on personal judgment to achieve greater efficiency. At every level, this indicator reflects workers’ insistence on the value of experiential judgment alongside—or in place of—algorithmic direction. Some workers go further, questioning whether algorithmic decision-making is actually superior to their own judgment and challenging the very premise on which these systems are built. As Interviewee C1 states:

As long as a rider gains enough experience, or achieves a certain understanding of the algorithmic rules, they can completely replace the system. At this point, human initiative and agency become far stronger than the mechanical algorithm. A truly skilled rider does not need to rely on system-assigned orders; they can manually select orders, making decisions based on their own expertise. As the best judge of their own abilities, they understand how to optimize their workflow to maximize efficiency far better than the algorithm does.

Chinese food delivery workers operate under two distinct models: “ZhuanSong” and “ZhongBao.” ZhuanSong workers function essentially as full-time platform employees, assigned to specific stations with defined delivery zones and strict service hours. Their orders come primarily through algorithmic dispatch rather than manual selection. ZhongBao workers, by contrast, include many part-timers who maintain flexible schedules and exercise greater discretion in choosing orders. Within China's delivery industry, legends abound about “order kings” and “masters”—workers who have achieved top-tier rankings and earnings. Interviewee C2 observed that genuine masters tend to emerge from the ZhongBao ranks. During peak hours especially, skilled ZhongBao workers can view orders across the entire city, spanning 3–10 km. They can instantly identify which orders lie along compatible routes and which deliveries will be straightforward—judgments made entirely through experience. C2 learned one particular strategy:

Grab a distant order first, then assess which additional orders can be completed along the way. The route materializes immediately in my mind. Masters specifically target orders that ZhuanSong workers leave behind during rush periods, and this alone generates substantial income. It all comes down to experience.

C2's account illustrates an advanced form of algorithmic skill, in which experienced workers can match or outperform algorithmic dispatch through accumulated local knowledge. However, experience-based decision-making does not require reaching this elite level. Many delivery workers and ride-hailing drivers develop proficiency in platform and algorithm mechanics, enabling them to regulate their work rhythm for optimal algorithmic coordination and maximum efficiency. Interviewee L1 recalled frequent delays when first starting out due to inexperience. Over time, he gained understanding of the platform's algorithmic logic and learned to manage its rules, which improved his delivery planning: “Initially, I blamed problems on poor algorithm design. Gradually, though, I accumulated experience and grasped the essentials—knowing how many orders to accept at different times. Eventually, I realized this job offered considerable room for personal initiative, which made the work much more satisfying.”

Together, C2's and L1's accounts show how experienced workers integrate algorithmic knowledge with personal judgment to optimize their work. This capability demonstrates a profound level of agency. These workers demonstrate not only practical mastery of algorithmic systems but also a critical awareness that human judgment retains value that automation cannot fully capture.

Discussion

Our interviews reveal a hierarchical framework of algorithmic skills comprising three dimensions—algorithmic awareness, algorithm learning and comprehension, and algorithm utilization and mastery—each operationalized through specific indicators (as shown in Figure 1). These range from basic recognition of algorithmic presence to experience-based decision-making that can rival or bypass algorithmic direction. Unlike the fragmented tactical accounts found in prior studies, these indicators form a coherent hierarchical structure—one that extends existing work on algorithmic literacy/skills into the specific conditions of platform labor. This framework offers a basis for future research to further develop and test measures of algorithmic skills in gig work contexts.

Beyond the resistance paradigm

Past work on algorithmic labor relations predominantly adopts a resistance lens, emphasizing how workers in asymmetric power structures reassert agency through critical or subversive actions. For instance, Huang (2026) conceptualizes workers’ tactics as everyday resistance emphasizing conflict and subversion, while Heiland (2021) highlights collective actions against algorithmic systems. Other studies call for stronger regulatory oversight to address platform–worker power asymmetries (Di Porto and Zuppetta, 2021). However, such actions tend to be episodic and difficult to sustain, rarely consolidating into organized, lasting solidarity.

While acknowledging adversarial dimensions of platform–worker relations, we argue that framing agency solely through resistance is insufficiently constructive. A more critical question emerges: what comes after resistance and critique? How can we foster better human–algorithm relations? Since algorithmic management is likely to intensify rather than recede, a pressing question is how workers can develop productive competencies within these systems.

Our skill-building framework offers a systematic alternative. Spanning algorithmic awareness, causal inference, rule exploitation, collaborative learning, and experience-based decision-making, these indicators offer a structured way of assessing workers’ algorithmic skills across multiple levels. This framework may also have practical implications, offering workers and policymakers a vocabulary for recognizing and cultivating algorithmic competencies in platform labor.

Algorithmic skills as foundation for solidarity

Our framework also has implications for understanding barriers to worker solidarity. Platforms atomize workers by classifying them as “independent contractors,” dismantling collective foundations and forcing them into competitive rather than cooperative relations (Barratt et al., 2020). We find that enhancing algorithmic skills offers a viable pathway to break this impasse: it not only reduces “involutionary” competition among workers but also creates the preconditions for more informed collective engagement with algorithmic systems.

Our empirical evidence illustrates this dynamic. Several interviewees reported that some workers accelerate deliveries to maximize earnings, feeding algorithms data that subsequently tightens delivery timeframes for all workers. This individual profit-seeking intensifies workloads collectively, reinforcing algorithmic strictness. However, as interviewee C1 explained, such workers are not algorithmic “accomplices” but act unconsciously—they lack the algorithmic awareness to recognize that their behavior reshapes algorithmic logic in ways that will eventually worsen their own working conditions.

If workers possessed higher algorithmic skills—understanding operational logic and causal relationships rather than mechanically following platform directives—such collectively harmful behaviors probably would diminish. This would weaken destructive competition and create conditions for solidarity. Greater algorithmic awareness could, over time, provide a shared knowledge base from which more informed collective responses—whether cooperative or adversarial—might emerge.

Skills as dynamic, context-embedded practice

While other studies also move beyond resistance to emphasize worker skills (Weber et al., 2025; Fiers, 2024; Jarrahi and Sutherland, 2019), our innovation lies in foregrounding how algorithmic skills are progressively accumulated through algorithmic interaction and work practice. These are not pre-existing generalized digital skills workers bring to gig work, nor static categorical labels, but dynamic, processual knowledge developed in situ. While related to digital skills, algorithmic skills can only be understood in relation to the broader sociotechnical context in which they develop. Drawing on Leu et al.'s (2004) framework of new literacies, we recognize that the relationship between workers’ skills and algorithmic technology is transactional—workers not only adapt to algorithms but also transform how these technologies function through their practices. Moreover, in line with Leu et al.'s emphasis on critical literacies, more proficient workers maintain an evaluative stance toward algorithmic directives, testing them against their own experience.

Limitations and future research

This study develops a conceptual framework of algorithmic skills grounded in the experiences of on-demand gig workers. However, these indicators are closely tied to the specific conditions of on-demand gig labor, which may limit their transferability to other occupational settings. Future research could investigate which indicators might extend to other algorithm-mediated occupations such as digital creative work, logistics management, or platform-based professional services. Understanding which skills transfer across platform contexts would contribute to a broader theory of worker competency in algorithmically managed work. Additionally, while our interview data illuminate the relationship between algorithmic skills and worker agency, future research could complement these findings through large-scale survey-based studies. The indicators developed here could be adapted into survey instruments to examine what factors shape the development of algorithmic skills, and how these skills relate to outcomes such as job satisfaction and earnings.

Footnotes

Acknowledgments

The authors would like to extend their gratitude to Dr. Kelley Cotter for her helpful comments on earlier versions of this article. They also thank colleagues from the LIKED Lab at Pennsylvania State University for insightful discussions related to this work.

ORCID iDs

Xinyi Hong

Xinyi Cheng

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

All-China Federation of Trade Unions (2023) The Ninth Survey on the Status of Chinese Workers (Report Volume). Beijing: Workers’ Publishing House. in Chinese.

Barratt

Goods

Veen

(2020) I’m my own boss…’: Active intermediation and ‘entrepreneurial’ worker agency in the Australian gig economy. Environment and Planning A: Economy and Space 52(8): 1643–1661.

Bishop

(2018) Anxiety, panic and self-optimization: Inequalities and the YouTube algorithm. Convergence: The International Journal of Research into New Media Technologies 24(1): 69–84.

Bonini

Treré

(2024) How people develop their algorithmic agency. Algorithms 13(6): 16–23.

Bucher

(2017) The algorithmic imaginary: Exploring the ordinary affects of Facebook algorithms. Information, Communication & Society 20(1): 30–44.

Burrell

Fourcade

(2021) The society of algorithms. Annual Review of Sociology 47(1): 213–237.

Cameron

(2022) Making out’ while driving: Relational and efficiency games in the gig economy. Organization Science 33(1): 231–252.

Cameron

Rahman

(2022) Expanding the locus of resistance: Understanding the co-constitution of control and resistance in the gig economy. Organization Science 33(1): 38–58.

Canfora

Di Penta

(2007) New frontiers of reverse engineering. In: Future of Software Engineering (FOSE’07), pp.326–341. IEEE.

10.

Chen

Soriano

(2022) How Do Workers Survive and Thrive in the Platform Economy? Evidence from China and the Philippines. In: Digital Work in the Planetary Market. Cambridge: The MIT Press, 41–58.

11.

Chen

(2018) Thrown under the bus and outrunning it! the logic of didi and taxi drivers’ labour and activism in the on-demand economy. New Media & Society 20(8): 2691–2711.

12.

Chen

(2022) Two worlds and dual identities: The labor process and labor relations in the platform economy era [in Chinese]. Sociological Research 6: 81–100, 228.

13.

Cini

(2023) Resisting algorithmic control: Understanding the rise and variety of platform worker mobilisations. New Technology, Work and Employment 38(1): 125–144.

14.

Cotter

(2019) Playing the visibility game: How digital influencers and algorithms negotiate influence on Instagram. New Media & Society 21(4): 895–913.

15.

Cotter

(2024) Practical knowledge of algorithms: The case of BreadTube. New Media & Society 26(4): 2131–2150.

16.

Cotter

Reisdorf

(2020) Algorithmic knowledge gaps: A new horizon of (digital) inequality. International Journal of Communication 14: 21.

17.

De Groen

Maselli

(2016, June 3) The impact of the collaborative economy on the labour market (CEPS Special Report No. 138). CEPS. https://ssrn.com/abstract=2790788

18.

DeVito

(2021) Adaptive folk theorization as a path to algorithmic literacy on changing platforms. Proceedings of the ACM on Human-Computer Interaction 5(CSCW2): 1–38.

19.

Di Porto

Zuppetta

(2021) Co-regulating algorithmic disclosure for digital platforms. Policy and Society 40(2): 272–293.

20.

Dogruel

Masur

Joeckel

(2022) Development and validation of an algorithm literacy scale for internet users. Communication Methods and Measures 16(2): 115–133.

21.

Duggan

Carbery

McDonnell

, et al. (2023) Algorithmic HRM control in the gig economy: The app-worker perspective. Human Resource Management 62(6): 883–899.

22.

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(1): 114–132.

23.

Eaton

Elaluf-Calderwood

Sørensen

, et al. (2015) Distributed tuning of boundary resources. MIS quarterly 39(1): 217–244.

24.

Eslami

Rickman

Vaccaro

, et al. (2015) I always assumed that I wasn’t really that close to [her]’: Reasoning about invisible algorithms in news feeds. In: Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems, pp.153–162.

25.

Ferrari

Graham

(2021) Fissures in algorithmic power: Platforms, code, and contestation. Cultural Studies 35(4–5): 814–832.

26.

Fiers

(2024) Resilience in the gig economy: Digital skills in online freelancing. Journal of Computer-Mediated Communication 29(5): zmae014.

27.

Gagrčin

Naab

Grub

(2026) Algorithmic media use and algorithm literacy: An integrative literature review. New Media & Society 28(1): 423–447.

28.

Gandini

(2019) Labour process theory and the gig economy. Human Relations 72(6): 1039–1056.

29.

Gillespie

(2014) The relevance of algorithms. In: Gillespie

Boczkowski

Foot

(eds) Media Technologies. Cambridge, MA: The MIT Press, 167–194.

30.

Glavin

Bierman

Schieman

(2021) Über-alienated: Powerless and alone in the gig economy. Work and Occupations 48(4): 399–431.

31.

Gray

Suri

(2019) Ghost work: How to stop Silicon Valley from building a new global underclass. New York: Eamon Dolan Books.

32.

Gruber

Hargittai

(2023) The importance of algorithm skills for informed internet use. Big Data & Society 10(1): 20539517231168100.

33.

Hargittai

Gruber

Djukaric

, et al. (2020) Black box measures? How to study people’s algorithm skills. Information, Communication & Society 23(5): 764–775.

34.

Healy

Nicholson

Pekarek

(2017) Should we take the gig economy seriously? Labour and Industry 27(3): 232–248.

35.

Heiland

(2021) Controlling space, controlling labour? Contested space in food delivery gig work. New Technology, Work and Employment 36(1): 1–16.

36.

Howcroft

Bergvall-Kåreborn

(2019) A typology of crowdwork platforms. Work, Employment and Society 33(1): 21–38.

37.

Huang

(2026) Weapons of the riders: Everyday algorithmic resistance of migrant food-delivery workers in China’s gig economy. Journal of Contemporary China 35(158): 1450–1465.

38.

Huws

Spencer

Joyce

(2016) Crowd work in Europe: Preliminary results from a survey in the UK, Sweden, Germany, Austria and the Netherlands.

39.

Jarrahi

Newlands

Lee

, et al. (2021) Algorithmic management in a work context. Big Data & Society 8(2): 14.

40.

Jarrahi

Sutherland

(2019) Algorithmic management and algorithmic competencies: Understanding and appropriating algorithms in gig work. In Information in Contemporary Society: 14th International Conference, iConference 2019, Washington, DC, USA, March 31–April 3, 2019, Proceedings 14 , p. 578–589. Springer International Publishing.

41.

Kadolkar

Kepes

Subramony

(2024) Algorithmic management in the gig economy: A systematic review and research integration. Journal of Organizational Behavior 46(7): 1057–1080.

42.

Kellogg

Valentine

Christin

(2020) Algorithms at work: The new contested terrain of control. Academy of Management Annals 14(1): 366–410.

43.

Klawitter

Hargittai

(2018) It’s like learning a whole other language’: The role of algorithmic skills in the curation of creative goods. International Journal of Communication 12: 3490–3510.

44.

Lata

Burdon

Reddel

(2023) New tech, old exploitation: Gig economy, algorithmic control and migrant labour. Sociology Compass 17(1): e13028.

45.

Lee

Kusbit

Metsky

, et al. (2015) Working with machines: The impact of algorithmic and data-driven management on human workers. In: Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems, pp.1603–1612.

46.

Leu

Kinzer

Coiro

, et al. (2004) Toward a theory of new literacies emerging from the internet and other information and communication technologies. In: Rudell

Unrau

(eds) Theoretical Models and Processes of Reading, 5th ed. Newark, DE: International Reading Association, 1568–1611.

47.

Lin

de Kloet

(2019) Platformization of the unlikely creative class: Kuaishou and Chinese digital cultural production. Social Media + Society 5(4): 2056305119883430.

48.

Lloyd

(2014) Following the red thread of information in information literacy research: Recovering local knowledge through interview to the double. Library & Information Science Research 36(2): 99–105.

49.

Maffie

(2023) Becoming a pirate: Independence as an alternative to exit in the gig economy. British Journal of Industrial Relations 61(1): 46–67.

50.

Meituan Research Institute (2023, August 29) Research on the occupational skills and skill formation of food delivery couriers. https://mri.meituan.com/research/report?typeCodeOne=5 [in Chinese].

51.

Newlands

(2021) Algorithmic surveillance in the gig economy: The organization of work through lefebvrian conceived space. Organization Studies 42: 719–737.

52.

Oeldorf-Hirsch

Neubaum

(2023) What do we know about algorithmic literacy? The status quo and a research agenda for a growing field. New Media & Society 27(2): 681–701.

53.

Pasquale

(2015) The black box society: The secret algorithms that control money and information. Boston: Harvard University Press.

54.

Prassl

(2018) Humans as a service: The promise and perils of work in the gig economy. Oxford: Oxford University Press.

55.

Purcell

Brook

(2022) At least i’m my own boss! explaining consent, coercion and resistance in platform work. Work, Employment and Society 36: 391–406.

56.

Qiu

(2025) Beyond Datafied Labor: The Algorithmic Dilemmas and Resistance of Delivery Riders. The National High School Journal of Science: 1–9.

57.

Rosenblat

(2018) When your boss is an algorithm. International New York Times, 15 October. Available at: https://link.gale.com/apps/doc/A558283529/AONE?u=anon∼e4224565&sid=googleScholar&xid=5e8eaf1f (accessed 10 February 2025).

58.

Scholz

(2017) Uberworked and underpaid: How workers are disrupting the digital economy. Cambridge: John Wiley & Sons.

59.

Schuppan

(2014) E-Government at work level: Skilling or de-skilling? In: 2014 47th Hawaii International Conference on System Sciences, pp.1927–1934: IEEE.

60.

Schwartz

Mahnke

(2021) Facebook use as a communicative relation: Exploring the relation between Facebook users and the algorithmic news feed. Information, Communication & Society 24(7): 1041–1056.

61.

Shapiro

(2018) Between autonomy and control: Strategies of arbitrage in the ‘on-demand’ economy. New Media & Society 20: 2954–2971.

62.

Shin

(2021) The effects of explainability and causability on perception, trust, and acceptance: Implications for explainable AI. International Journal of Human-Computer Studies 146: 102551.

63.

Statista (2024) Online food delivery - worldwide. Available at: https://www.statista.com/outlook/emo/online-food-delivery/worldwide (Accessed 9 February 2025).

64.

Strauss

Corbin

(1998) Basics of Qualitative Research: Techniques and Procedures for Developing Grounded Theory. Thousand Oaks, CA: Sage Publications.

65.

Sun

(2019) Your order, their labor: An exploration of algorithms and laboring on food delivery platforms in China. Chinese Journal of Communication 12(3): 308–323.

66.

Sutherland

Jarrahi

Dunn

, et al. (2020) Work precarity and gig literacies in online freelancing. Work, Employment and Society 34(3): 457–475.

67.

Swart

(2021) Experiencing algorithms: How young people understand, feel about, and engage with algorithmic news selection on social media. Social Media + Society 7(2): 20563051211008828.

68.

Vallas

Schor

(2020) What do platforms do? Understanding the gig economy. Annual Review of Sociology 46(1): 273–294.

69.

Van der Nagel

(2018) Networks that work too well’: Intervening in algorithmic connections. Media International Australia 168(1): 81–92.

70.

Van Deursen

AJAM

Helsper

Eynon

(2014) Measuring Digital Skills. From Digital Skills to Tangible Outcomes project report. Available at: www.oii.ox.ac.uk/research/projects/?id=112 (Accessed 18 January 2026)

71.

Van Doorn

(2017) Platform labor: On the gendered and racialized exploitation of low-income service work in the ‘on-demand’ economy. Information, Communication & Society 20(6): 898–914.

72.

Weber

Remus

Geiger

, et al. (2025) Between proactive and reactive coping: How food delivery workers Cope with algorithmic management threats. European Journal of Information Systems: 1–29.

73.

Wood

Graham

Lehdonvirta

, et al. (2019) Good gig, bad gig: Autonomy and algorithmic control in the global gig economy. Work, Employment and Society 33(1): 56–75.

74.

Zarouali

Boerman

de Vreese

(2021) Is this recommended by an algorithm? The development and validation of the algorithmic media content awareness scale (AMCA-scale). Telematics and Informatics 62: 101607.

75.

Zhu

Cui

Zheng

, et al. (2023) The world of riders: A social survey of new occupational groups. Social Sciences Academic Press. (in Chinese).