Between automation and gamification: forms of labour control on crowdwork platforms

View author details View Less
  • 1 WZB Berlin Social Science Center Reichpietschufer, , Germany
  • | 2 Weizenbaum Institute for the Networked Society, , Germany
  • | 3 Helmut Schmidt University Hamburg, , Germany
  • | 4 WZB Berlin Social Science Center, , Germany
Full Access
Get eTOC alerts
Rights and permissions Cite this article

Algorithmic management is a core concept to analyse labour control on online labour platforms. It runs the risk, however, of oversimplifying the existing variety and complexity of control forms. In order to provide a basis for further research, this article develops a typology of labour control forms within crowdwork and discusses how they influence perceptions of working conditions. It identifies the two most relevant forms of labour control in crowdwork: direct control mainly takes the form of automated output control, while indirect control aiming at creating motivation and commitment is mainly exerted through ranking and reputation systems (gamification). The article shows that these forms of control and their combination are linked with very different ways of how platform workers perceive working conditions on platforms. In addition, the analysis shows significant differences regarding the perception of working conditions between those who work on platforms in addition to a regular employment as opposed to those who are self-employed and rely more strongly, if not fully, on their income from platform work. The analysis is based on qualitative and quantitative research on crowdwork platforms. In particular, it builds on an online survey conducted with 1,131 crowdworkers active on different types of platforms.

Abstract

Algorithmic management is a core concept to analyse labour control on online labour platforms. It runs the risk, however, of oversimplifying the existing variety and complexity of control forms. In order to provide a basis for further research, this article develops a typology of labour control forms within crowdwork and discusses how they influence perceptions of working conditions. It identifies the two most relevant forms of labour control in crowdwork: direct control mainly takes the form of automated output control, while indirect control aiming at creating motivation and commitment is mainly exerted through ranking and reputation systems (gamification). The article shows that these forms of control and their combination are linked with very different ways of how platform workers perceive working conditions on platforms. In addition, the analysis shows significant differences regarding the perception of working conditions between those who work on platforms in addition to a regular employment as opposed to those who are self-employed and rely more strongly, if not fully, on their income from platform work. The analysis is based on qualitative and quantitative research on crowdwork platforms. In particular, it builds on an online survey conducted with 1,131 crowdworkers active on different types of platforms.

Introduction

Work in the platform economy has become a core topic in labour sociology research. An important question is how platform management ensures the transformation of labour and mobilises the performance of platform workers (Thompson, 1983; Smith, 2006). Early literature has focused on the new possibilities for surveillance and rigid control due to the technological mediation of work. In an analogy to Braverman’s (1974) analysis of Taylorism and the relationship between technology and the degradation of work, these first studies have characterised the control forms of labour platforms as a digital version of Taylorism (Kittur et al, 2013; Bergvall‐Kåreborn and Howcroft, 2014). In the last couple of years, however, empirical studies have increasingly emphasised indirect control via the use of ranking systems and other gamification mechanisms that promote the self-responsibility and self-discipline of workers (Gandini, 2016; Gerber and Krzywdzinski, 2019; Kellogg et al, 2019).

One part of the literature thus describes platform work as characterised by direct technical control, while other concepts position labour control of platforms as variations of indirect control and responsible autonomy approaches. This schematic view, however, runs the risk of overlooking the diversity of tasks and complexity of control forms. Our article tries to move the discussion forward by developing a typology of labour control forms on crowdwork platforms and discussing how they influence perceptions of working conditions.

At the same time, we argue that for the analysis of work-related outcomes, not only labour control but also the extent of dependence of workers on platforms must be taken into consideration. One can both find platform workers who entirely depend on their platform income, while others use it as an additional source of income (Schor, 2015; Rosenblat, 2018; Robinson and Vallas, 2020). This heterogeneity of socio-economic backgrounds (which characterises many low-hours or marginal work arrangements) impacts the power relations between platforms and workers, and as a result also shapes perceptions and behaviour at work.

Our main questions are:

  1. Which forms of labour control can we distinguish within crowdwork?

  2. How are these forms of labour control related to the perception of working conditions?

  3. How does the degree of dependence on platform income influence the perception of working conditions?

The analysis is based on qualitative and quantitative research on a specific form of platform work: platform-based online labour or so-called ‘crowdwork’. In particular, we build on data from an online survey conducted with 1,131 crowdworkers active on different types of platforms. This allows us to compare different constellations of control forms by platform and task types (for example micro- and macrowork platforms).

The article proceeds as follows. First, we present the state of current research, including central debates on the question of labour control from LPT as well as existing research on platform work. Thereafter, we present the data and methodology and then, in the third part, discuss the empirical findings. First, different labour control forms and their effects on the perceptions of work are discussed. Second, these are compared with crowdworkers’ different degrees of economic dependence. Finally, we consider our conclusions and their implications with regards to collective action.

State of research

Forms of labour control in platform work

Platform work is a broad term that has not been used entirely consistently in the literature. What is common to all contributions is the ‘on-demand’ mediation of jobs to formally independent workers via internet-based platforms. A central distinction lies, however, in the relevance of space. One the one hand, platforms such as Uber, Deliveroo, Instacart or Helpling act as intermediaries for service activities that are delivered locally and are therefore location-dependent. Platforms such as Amazon Mechanical Turk (AMT) or Upwork, on the other hand, focus on location-independent jobs that are performed online and by a largely anonymous mass of people, the so-called ‘crowd’. We refer to location-independent online labour also as crowdwork.

Only estimates are available for employment in the platform sector, as statistical sources are lacking. In the case of the US, recent studies put the figure at between 600,000 (Katz and Krueger, 2019) and 1.6 million people (Bureau of Labor Statistics, 2018); in the case of Germany, conservative estimates range from 300,000 to 400,000 people (Bonin and Rinne, 2017; Pongratz and Bormann, 2017). These estimates include both location-dependent and location-independent platform work.

Location-dependent and location-independent forms of platform work differ in terms of how work is organised and how far collective worker organisation is possible (Elmer et al, 2019). The individualisation and flexibility of working hours is greater in crowdwork than in location-dependent work. In addition, the latter is dominated by low-skill tasks, while crowdwork includes a huge range of low skill, but also high skill tasks. There are, however, also similarities between the different forms of platform work with regards to labour control on the one hand and the motives of workers on the other. While our empirical analysis focuses on crowdwork, we connect our literature review to platform work more generally.

For platform work, similarly to ‘classical’ dependent wage labour, the central question in the labour process is to what extent management (or the platforms) can ensure the transformation of potential labour power into actually spent and value-generating labour. In the tradition of the labour process analysis, this has been understood as the problem of labour control. We refer to labour control as the way om which companies obtain the desired behaviour from workers (Edwards, 1979: 17) or more specifically as forms with which the companies ‘direct work tasks’, ‘supervise and evaluate performance’ and organise ‘discipline and rewards’ (Thompson, 1983: 122). The early days of labour process theory were dominated by a controversy over the role and prevalence of Taylorist labour control based on technical and bureaucratic means in comparison to approaches based on responsible autonomy, job enrichment and involvement (Braverman 1974; Friedman 1977; Burawoy, 1979). In this controversy, Thompson (1983), Edwards (1979) and other researchers early on referred to the classical Marxian theory of the labour process as a ‘contested terrain’ (Edwards, 1979) characterised by workplace conflicts and bargaining (see also Lawrence and Robinson, 2007, for more recent research). As a result, we can observe that forms of direct, Taylorist control as well as ‘responsible autonomy’ may coexist in the same company.

In the context of platform work, the problem of labour control arises in a particular way. Since there is no workplace and no employment relationship, the forms of labour control available to the platforms are limited. In particular, there is no direct personal control, no hierarchical bureaucratic control. A core role is rather played by the technical infrastructure of the platforms. Reflecting this particularity, research early on coined the term ‘algorithmic management’. This term goes back to the work of Aneesh (2009) who describes it as a form of labour control that functions ‘by shaping an environment […] in which there are only programmed alternatives for the execution’ of the work tasks (p 71).

As research discussions during the recent years have shown, however, the term ‘algorithmic management’ is often used in a variety of different ways. One group of researchers has highlighted the automation of task assignment and surveillance. Möhlmann and Zalmanson (2017) provided one of the first definitions of algorithmic management on platforms:

We argue that algorithmic management is characterized by continuously tracking and evaluating worker behavior and performance, as well as automatic implementation of algorithmic decisions. In algorithmic management practices, workers interact with a ‘system’ rather than with humans. In many cases, the system has less transparency, and workers have no knowledge of the set of rules governing the algorithms. (p 4)

Irani (2015) emphasised automated filtering and task allocation as examples of algorithmic management: workers who do not meet certain criteria (for example language, past performance scores) never see certain tasks. Elmer et al (2019) and Kittur et al (2013) associated algorithmic management with automated surveillance, referring to the case of Upwork. For specific jobs, Upwork takes desktop screenshots and counts keystrokes to ensure that the remote crowdworker is actually active.

Against the background of such automated work-process management, researchers focusing on the microwork platform Amazon Mechanical Turk (AMT or MTurk), in particular, characterised platform work as ‘digital Taylorism’ (Kittur et al, 2013; Altenried, 2019). This connects to broader research on the extent to which information technologies can be used to monitor and measure ‘white-collar work’ and thereby create an ‘assembly line in the head’ (Bain and Taylor, 2000), for example with regards to office work or call centres (Woodcock, 2016). According to (Bergvall‐Kåreborn and Howcroft 2014), ‘when workers register with MTurk, they are allocated an alphanumeric ID identifier, which renders the workers invisible, thereby anonymising the underlying social relations of labour’ (p 218).

In contrast to the characterisation of platform work through automated control and Taylorist approaches, other studies associated ‘algorithmic management’ with the use of control mechanisms aiming at creating commitment and self-direction (Rosenblat and Stark, 2016; Woodcock and Johnson, 2018; Shapiro, 2018; Wood et al, 2019; Fieseler et al, 2019). Wood et al (2019) argued that even on microwork platforms ‘control is operated at the end of the labour process rather than during it’ (p 64), which differs fundamentally from the Taylorist approach to control. A major form of this kind of indirect control is the so-called ranking and reputation systems (RRS) (Kellogg et al, 2019). At the core, they regulate access to job offerings based on ratings for past jobs. But they are more than this. Gandini (2016) suggested that reputation becomes the social capital of freelancers on platforms. Schörpf et al (2017) and Gandini (2016) showed that RRS are an important source of pride and motivation of platform workers. They analysed how platforms induce freelancers to spend effort and unpaid extra time because personal reputation is a key to being visible to clients and thus jobs and earning opportunities. Platforms often include a number of metrics to regulate and incentivise performance quality, to promote swift response to clients. Uber or Lyft, for instance, require drivers to meet a certain acceptance rate (Shapiro, 2018).

The RRS are linked to a greater debate about the gamification of work. According to Woodcock and Johnson (2018), gamification ‘is understood as the application of game systems – competition, rewards, quantifying player/user behaviour – into non-game domains, such as work’ (p 542). Several studies on platform work showed that awards or playfully presented personal bests and daily goals are used to mobilise productivity or evoke specific behaviours (Gandini, 2016; Lehdonvirta, 2018). Schörpf et al (2017) for instance found that creative freelance platforms such as Upwork use ‘badges for acquiring and maintaining a high completion rate of projects which can be interpreted as an incentive not to cancel a project once accepted’ (p 50). In this study we understand gamification as a broader concept that must not necessarily be linked to symbolic awards, and whose core is the use of visible rankings and reputation scores to induce work performance – which is often enough to create competition between workers.

We argue that the usage of one uniform concept – such as algorithmic management – to characterise labour control on platforms is of limited analytical benefit. It runs the danger of effacing a wide variety of approaches, of which some resemble Digital Taylorism while others correspond to the gamification approach. The main aim of our contribution is to push forward the development of a differentiated typology of labour control forms on crowdwork platforms, taking up a suggestion by Griesbach et al (2019). Based on interviews with food delivery platform workers, they identified a considerable diversity of workers’ experiences with labour control. They characterised the Instacart platform as an ‘algorithmic despotism regime’ based on a tight control of working hours and activities. At the same time, however, they emphasised that other food delivery platforms make much less use of such direct control. They argued: ‘Specifying the varieties of algorithmic control and their impact on workers’ earnings and experiences is also critical for informing legal and political debates around the future of work’ (Griesbach et al, 2019: 13).

External factors: dependence on platforms

In the early debates about the labour process, it was pointed out that the development of labour control is influenced by external factors. Burawoy (1990) saw the development of state intervention and the regulation of the labour markets as a factor that promoted the development of the hegemonic factory regimes, which gave greater weight to forms of indirect control. Similarly, Edwards (1979) highlighted the existence of different labour control regimes, which he attributed to the different conditions of production, for example the differing market powers of companies and fractions of the workforce. He developed the notion of ‘fractional politics’ to describe how management benefited from creating different strata of the workforce with distinct interests.

Linking to these previous debates, Robinson and Vallas (2020) as well as Krzywdzinski and Gerber (2020) emphasised the role of external factors in the development of labour control in the platform economy. In particular, they pointed to the heterogeneity of the workforce with respect to motives for working and economic dependence on platforms. Robinson and Vallas (2020) stressed that ‘hegemonic and despotic labour regimes can and do co-exist’ on platforms, not only because platforms organise different tasks or follow different strategies, but because of the fractured nature of the platform workforce itself (see also Huws et al, 2016; Popiel, 2017; Berg et al, 2018; Pesole et al, 2018; Krzywdzinski and Gerber, 2020). Platform crowds differ regarding dimensions like employment status (that is, whether platform work constitutes the primary income source or is combined with another form of employment), time worked on the platforms and education. In terms of this heterogeneity, platform work shows similarities to other forms of temporary or marginal work.

Based on empirical research with Uber drivers in Boston, Robinson and Vallas (2020) emphasised in particular the differences between full-time and part-time (or discretionary) workers. These main groups experience risks related to platform work very differently, depending on their level of economic dependence. Robinson and Vallas (2020) suspected that the platform economy has generated new inequalities among workers by drawing together workers with distinct market power and embeddedness in the socio-economic system:

platforms attract workers with such fractured orientations toward their work that they can side step the forms of industrial conflict that have long bedevilled Fordist organisations. Because hobbyist and part time workers ‘have fewer economic incentives to advocate better pay’ and are ‘tolerant of working conditions that are anathema to occupational drivers trying to support their families,’ ([Rosenblat] 2018: 53–54) Uber is able to secure a floating source of workers whose very presence undermines the pressures the firm would otherwise face. (Robinson and Vallas, 2020: 4)

This shows that for the analysis of labour control regimes on platforms, it is important to consider the heterogeneity of the workforce with regards to the dependence on income from platform work.

Varieties of work experiences

One of the central arguments in our analysis is that work behaviour and the perception of the working conditions vary not only by the nature of the labour control but also by the living conditions of the workers and, in particular, their dependence on platform work. Such an approach can help to order the variety of existing findings on the perception of platform work.

Early studies on platform work have emphasised the aspect of control and atomisation to characterise platform workers’ experiences. The study by Lee et al (2015) on Uber argued that the forms of labour control are perceived as unfair and arbitrary due to the intransparency and power asymmetries enshrined in the algorithmic systems. Möhlmann and Zalmanson (2017) provided similar findings but at the same time placed greater emphasis on the drivers’ perception of autonomy. Thereafter, drivers have developed ‘gaming’ strategies, which undermine the control of the platform (see also Rosenblat and Stark, 2016). A number of studies have dealt with work on the AMT platform. Lehdonvirta (2016), for example, argued that AMT perfects the fragmentation, anonymisation and technical control of work, and thus tends to produce alienation in work. Fieseler, Bucher, and Hoffmann (2019) also emphasised the ‘unfairness by design’ in the design of the AMT platform. Their empirical data, however, showed a surprisingly positive perception of work by the crowdworkers. The authors attributed this to the formal autonomy of workers and the associated perception of being able to decide on the work themselves.

A number of studies suggested that platform workers may also use these digital infrastructures to build networks of solidarity and resistance. Wells, Attoh and Cullen (2020) showed how Uber drivers in Washington DC organised a strike by daily meetings in the parking lot of an airport. Similarly, Heiland and Schaupp (2020) highlighted the key role of informal communication channels among food delivery riders to build solidarity networks. Such experiences of resistance, however, are absent with regards to geographically dispersed crowdwork. Wood, Lehdonvirta, and Graham (2018) showed that crowdworkers organise as internet communities based on a common freelance identity. Yet, their collective action continues to be confronted with a ‘fragmentation […] along national, occupation and platform line’ (p 109).

The argument of the fragmentation of platform labour is reminiscent of Edwards’ (1979) concept of ‘fractional politics’. However, it is important to stress that Edwards sees fractional politics as the result of a conscious management strategy to control and weaken the labour side. In contrast, the fragmentation in the case of platform labour is more the result of decisions on the workers’ side: platforms offers opportunities for both people interested in full-time work and those looking for a supplementary income.

Summary

To sum up, the existing research leads us to three major expectations which will guide our analysis:

  • First, we expect to find a variety of forms of labour control in crowdwork. There are two major poles: on the one hand, there is automated output control which can be interpreted in the tradition of the technical and bureaucratic forms of control discussed since the 1970s. On the other hand, we have control based on gamification by means of RRS.

  • Second, we expect to find different degrees of dependence of workers on the crowdwork platforms. This is based on the high variety of motives of workers joining the platforms.

  • Third, we expect that the constellations of labour control and dependence on the platforms influence workers’ perceptions of working conditions on platforms.

Our aim is to combine the answers to these questions in order to develop a typology of control forms on crowdwork platforms and the impact they have on workers.

Data and methods

In the analysis, we used data from the project ‘Between digital bohemia and precarity. Work and performance in the crowd’ (2016–2019), which was funded by the Thyssen Foundation. The analysis focused on crowdwork as a specific form of platform work. The project combined qualitative and quantitative methods. In the qualitative part, 15 case studies of crowdwork platforms in Germany and the United States were conducted. The central question was how the companies design their labour control approaches. In particular, the study examined approaches to the recruitment of crowdworkers, the control and regulation of performance and communication with crowdworkers. The case studies were based on an analysis of platform websites and official documents, their work interfaces for crowdworkers and their communication forums. The study also involved conducting 32 one- to two-hour semi-structured interviews with representatives of the platforms and crowdworkers (Gerber and Krzywdzinski, 2019).

The platform-selection process took place in several stages. Our central aim was to include platforms that focused on different types of work (in terms of task content and skill requirements). We distinguished between two task types: micro- and macrotasks. Microtasks can be defined as routine tasks (for example, image categorisation, lead data verification, short audio transcriptions) or as tasks that do not require specific knowledge (for example, short product descriptions, testing of apps). The nature of these tasks allows them to be broken down into short and standardised components that can be completed within seconds or minutes. Several crowdworkers can work simultaneously on the same task without having to interact. Macrotasks, by contrast, are complex and require a high degree of creativity and specific, sometimes professional knowledge (for example, design activities, software programming, development of product concepts). These tasks cannot be broken down and are therefore organised as projects lasting several days or weeks. The focus is on quality. Platforms often organise competitions to generate proposals, from which the client, a jury, or the crowd community can select the winners.

To select the platforms, we compiled a list of platforms in Germany and selected metropolitan regions in the United States (Bay Area, Boston, Chicago, New York) on the basis of internet research. We identified 60 platforms, which we contacted in several rounds by phone and email. We continued to contact them until we reached a sample of 15 case studies that equally represented micro- and macroplatforms.

We combined this qualitative approach with a quantitative survey of crowdworkers which is at the core of the following analysis. The survey was sent to crowdworkers who registered as German or US residents on the platforms. We selected Germany and the United States in order to capture different institutional settings and their impact on platform work (see Krzywdzinski and Gerber, 2020). In the USA in particular, there are often hardly any alternatives to platform work because of a rudimentary welfare state. Platform work often plays the role of a last lifeline when other sources of income dry up. In Germany, platform work is used more as a secondary source of income, not least because the relatively strong social security system offers alternatives.

The total sample consists of 1,131 participants. On four platforms, the survey was placed on the platform as a paid task. On the other platforms, the survey invitation was sent to the crowdworkers via the platform itself (email lists) or, if the platforms refused to participate in the survey, via social media (Facebook, LinkedIn, Reddit, Xing) and the platforms’ communication forums. A table in the annex gives an overview of the platforms.

Our data therefore represent a ‘convenience sample’, which means we cannot generalise beyond the platforms we surveyed. We are, however, not aware of any sources of potential bias other than willingness to agree to take the survey. Our sample is also broadly similar in composition to those of the available studies on Germany (Leimeister et al, 2016; Bonin and Rinne, 2017) as well as the United States (Popiel, 2017; Difallah et al, 2018). Most importantly, however, our main aim is not to describe general characteristics of the total population of crowdworkers, but rather to analyse differences between subgroups. In this regard, an advantage of our sample is that we covered platforms with different types of tasks in one survey, thus avoiding a focus on individual platforms, which is the dominant approach in current research. Table 1 provides demographic information on our sample.

Table 1:

Socio-demographic composition of the sample

GenderWomen48.6%
Men49.9%
Other/third gender1.5%
AgeUp to 19 years3.8%
20–2940.2%
30–3933.5%
40–4914.0%
50–6410.7%
Above 651.5%
EducationHigh-school diploma and below40.7%
Tertiary degree59.3%
Platform typeMicrotasks76.0%
Macrotasks24.0%
CountryUSA45.2%
Germany50.9%
Other*3.9%
N1,131

Source: Krzywdzinski and Gerber, 2019. *Some platform workers were registered on the platforms as German or US residents, but actually reported living/working in other countries.

The questionnaire used in the survey consists of 28 questions and covered the following topics: (a) basic information about the working situation (working hours, types of tasks and so on); (b) the form of performance monitoring on the platform; (c) reputation and ranking systems; (d) interaction with other crowdworkers; (e) perception of working conditions and stress; and (f) socio-demographic information (age, gender and so on). We introduce the specific variables used in the analysis in the following section.

Forms of labour control and the perception of working conditions

A typology of labour control on platforms

We have argued that a particular feature of the platforms in contrast to physical workplaces is the importance of technical infrastructure for control of work performance and output. There is no supervisor and no place of work where someone could supervise the execution of the work. Based on the existing research, we can distinguish two major forms of technical control used by the crowdwork platforms:

  • The first is direct automatic output control and evaluation of performance. This approach corresponds to the technical or bureaucratic control as described in analyses of tayloristic systems (for example Edwards, 1979). These automatic controls are sometimes relatively simple. In the case of microtasks like the tagging of images, platforms intersperse test tasks, or random samples of three crowdworkers perform the same task and the majority answer is automatically taken as the correct one. In other cases, the automatic controls use human labour. For example, if a test task is answered incorrectly, or simply at random intervals, the software initiates a check by a further crowdworker. Some macrotask platforms work on prediction algorithms, combining data on the performance in previous tasks, the platform workers’ profiles and the results of crowd community voting to distinguish suitable solutions for a task from unsuitable ones. For example, one of the platforms we investigated – concerned with the production of medical diagnoses and recommendations – worked on an algorithm based on artificial intelligence, which was intended to identify those solutions that were most likely to achieve the desired quality criteria.

  • The second form of control is gamification, and in particular the use of reputation and ranking systems (RRS) to generate incentives and sanctions for performance. It represents an indirect form of control, that serves not only to stratify the crowd by performance and regulate access to work, but is also a core mechanism which platforms use to motivate workers and generate commitment. However, the systems vary in complexity between micro- and macrotasks (Gerber and Krzywdzinski, 2019). The former primarily use ‘objective’ performance criteria, such as past job evaluations or activity. On macro platforms, more complex algorithms can be found. Aside from ‘objective’ performance criteria, they often also contain ‘subjective’ variables such as customer communication, interaction with or likes by other crowdworkers. They also take care of linking the RRS to symbolic rewards like stars, badges, or awards for crowdworkers, and in this way making them a highly visible part of the workers’ profiles on the platforms. For one thing, the RRS serve a much finer regulation of task access, a process known as ‘matching’ of customers and workers. In addition, the RRS are supposed to provide incentives: on the one hand, to deliver high-quality work despite the uncertain earning opportunities; on the other hand, to mobilise unpaid extra work into the collaborative finalisation of products. We assume that the ranking of crowdworkers and the presentation of their scores on the platforms has an influence on the workers’ work behaviour, even if it is not symbolically exaggerated by badges, titles and other means. As the rankings have an impact on access to tasks and remuneration, they likewise create competition without being accompanied by symbolic awards and titles.

Of course, the labour control via the technical infrastructure has limits. Direct output control by the customers remains a third, important form of control on platforms. Customer control, however, is used by all platforms and does not constitute a useful basis for developing a typology.

To measure the use of the aforementioned forms of labour control, we chose a specific approach, namely the experiences or perceptions of the crowdworkers themselves. We asked the crowdworkers to what extent automatic controls are used and RRS play a role in the tasks they perform on the platforms. Such an approach has advantages and disadvantages. One obvious disadvantage is that it may well be that the use of automatic controls remains invisible to the crowdworkers (though this is less likely for RRS, since their effectiveness is based precisely on their visibility to the crowdworkers). This possibility exists, however our qualitative interviews with crowdworkers indicate that crowdworkers do perceive such controls (Gerber and Krzywdzinski, 2019). Evidence with regard to crowdworkers ‘gaming’ the platforms’ systems (Möhlmann and Zalmanson, 2017) also suggests that automatic controls cannot be hidden from crowdworkers. We believe that crowdworkers are aware of the use of such automatic controls, even though their specific mechanisms are certainly often not transparent and are also deliberately kept secret by the platforms. Thus, although the issue of labour control visibility is a concern due to the nature of our data, we do not consider it so fundamental that it would call into question the validity of our findings.

In addition, our own interviews with platforms (Gerber and Krzywdzinski, 2019) illustrate that implementing automated controls for many activities is not straightforward. Accordingly, control often continues to be exercised ‘manually’, with clients or other crowdworkers checking the work. This is true, for example, for many macrotasks, but also for microtasks such as texting. It is also important to emphasise that even on platforms which use automatic checks, the latter are often not used for all types of tasks. This is where the advantage of measuring labour control by surveying crowdworkers becomes apparent: we avoid making blanket assumptions about the role of automatic controls and RRS for activity domains where they are not important.

In order to measure automated control, we use the question: ‘The system recognises immediately if a task has not been performed according to the requirements.’ In our survey, about two-thirds of the microworkers said that the platforms use automated performance controls (Table 2). Given that many, but by no means all, microtasks allow automation of labour control, this share seems plausible to us. In the case of macroworkers, only about a quarter of our respondents report automatic controls, which again seems plausible given the complexity of macrotasks.

Table 2:

Crowdworkers’ perceptions of automation controls of work

All surveyed crowdworkersMicroworkersMacroworkers
Automatic controls are used (automation)53.8% (608)62.5% (537)26.1% (71)
Only workers reporting the use of automatic controls: The feedback I receive on rejected work is usually fair and sufficient.
 – Strongly agree12.0% (79)13.0% (70)12.7% (9)
 – Agree37.3% (227)37.4% (201)36.6% (26)
 – Neither agree nor disagree29.8% (181)30.0% (161)28.2% (20)
 – Disagree9.2% (56)9.1% (49)9.9% (7)
 – Strongly disagree3.8% (23)3.7% (20)4.2% (3)
 – No answer6.9% (42)6.7% (36)8.4% (6)

While we assume that workers are aware of automatic controls of their work, our survey gives some evidence about how far workers perceive the functioning of these controls as transparent. Around half of our respondents agreed (or strongly agreed) that platforms provide a sufficient explanation if performed work is rejected, while the other half was less satisfied.

In order to measure gamification, we included two items: first, the question of whether RRS are used on the platform. Given that they might be more or less relevant for different types of jobs, in addition, we added a second item on whether RRS influence income. The functioning of RRS presupposes their visibility to crowdworkers, insofar as we assume that our measurement does not pose any problems. In our survey, slightly more than half of the microworkers and more than three-quarters of the macroworkers indicated that RRS are used (Table 3).

Table 3:

Crowdworkers’ perceptions of reputation and ranking systems

All surveyed crowdworkersMicroworkersMacroworkers
RRS are used (gamification)62.9% (711)58.1% (499)77.9% (212)
Only workers reporting the use of RRS: I understand how the reputation/ranking is calculated.
 – Strongly agree22.4% (159)18.8% (94)30.7% (65)
 – Agree38.5% (274)40.3% (201)34.4% (73)
 – Neither agree nor disagree26.4% (188)27.9% (139)23.1% (49)
 – Disagree10.1% (72)10.4% (52)9.4% (20)
 – Strongly disagree1.7% (12)1.6% (8)1.9% (4)
 – No answer0.8% (6)1.0% (5)0.5% (1)

Again, it should be noted that the exact operation of RRS may well be more or less transparent. About two-thirds of respondents answered this question positively; for about one-third, the RRS are less or not at all transparent.

Based on these two forms of direct automated output control and indirect gamification through RRS, we can develop a typology of labour control constellations in crowdwork (Table 4). First, an approach that relies only on reputation and ranking systems and does not use automated control can be called gamification. It is based on a logic of labour control through competition and play. Second, we call an approach that relies on automated control and does not include gamification, automated despotism. It is, in return, based on a logic of work control through surveillance and sanctions. Introducing the term ‘despotism’, we link to the recent studies by Wood (2020) and Griesbach et al (2019) that use this term to emphasise the coercive elements of technological infrastructures. We would like, however, to emphasise that the ‘despotic’ coercion embodied in platform infrastructures differs fundamentally from the nineteenth-century despotism analysed by Burawoy (1985) or in recent studies on despotic labour control in China and other countries (for example Zhang, 2008). Third, we suggest the term gamified automation for constellations of labour control that combine automated controls and gamification. And finally, fourth, we have the potential case of platforms that for a range of tasks rely neither on automated control nor on gamification, but rather purely on control by the customers themselves. On microtask platforms this is sometimes referred to as the ‘self-service’ option, in which customers organise the whole control process themselves. It is worth mentioning, that even in this constellation, platforms still influence the labour process by defining the terms of trade and by providing interfaces for workers and customers.

Table 4:

Forms of labour control in platform work

Automated output control
NoYes
RRSNoPure customer controlAutomation
YesGamificationGamified automation

Source: Authors.

Table 5 shows the distribution of crowdworkers in our sample. Our sample is not representative, so we want to emphasise that the table is not intended to give an indication of the size of the respective groups in the total population of crowdworkers. Rather, it is important for us to have all four types represented in our sample to a relevant extent in order to analyse the differences between these groups.

Table 5:

Crowdworkers by constellation of labour control (in %, n= 1,131)

Constellations of labour controlAll crowdworkersMicroworkersMacroworkers
Pure customer control16.2% (183)16.1% (138)16.5% (45)
Automated despotism20.9% (237)25.8% (222)5.5% (15)
Gamification30.1% (340)21.4% (184)57.3% (156)
Gamified automation32.8% (371)36.7% (315)20.6% (56)
1,131859272

The constellations of labour control are related to the platform types. Macrotask platforms are oriented primarily to gamification, but in many cases rely on pure customer control only. Automatic control is mainly used in combination with gamification, while ‘automated despotism’ seems to be rather a niche phenomenon. This highlights that greater task complexity and the professionalism of freelancers requires an incentive-based approach. In the area of microtask platforms, the picture is more mixed and we find all four types to a relevant extent.

Relationship between labour control, dependence and perception of working conditions

We expect that labour control and the dependence of crowdworkers on the respective platform will influence perceptions of the working conditions. Regarding working conditions, we use three different types of indicators. First, we discuss aggregate indicators, such as work satisfaction or the intention to pursue platform work in the long term. Second, in order to get a more in-depth understanding, we examine perceptions of stress (which can be understood as directly influenced by the forms of labour control) related to time pressure, flexibility requirements and other factors. Third, we focus on the perception of general working conditions like pay, work content or career opportunities.

We measure dependence mainly using the primary employment status of the platform workers. People who have a regular full-time job besides working on platforms are less dependent on platform work than people who combine a part-time job with platform work or who are self-employed. A further measure of dependency is the proportion of platform income to total household income. The higher this proportion, the greater the dependence on the platform. Since many platform workers are active on multiple platforms, dependence is related to the overall income from platform work and not from individual platforms.

As we show elsewhere, gender and education play an important role in terms of dependence on platform work (Krzywdzinski and Gerber, 2020). Platform work is often a more important source of income for women than for men, since men more often than women have a regular (full-time) job in addition to platform work. These differences also exist with regard to education. Platform workers with a university degree are more likely to have a regular job in addition to platform work, which reduces their dependence on income from platform work. We take these factors into account in the following by using appropriate control variables.

How do labour control and the dependencies on platforms affect perceptions of the work situation? Let us first look at aggregate indicators, such as work satisfaction or the intention to pursue platform work in the long term (Table 6). We used the items ‘Overall, I am satisfied with working on the platform’ and ‘I see my personal future in platform work’, both with a five-level Likert scale. Table 6 presents two models for both variables: one that focuses on the forms of labour control, dependence on the platforms, and socioeconomic control variables (columns I and III); and a second that includes additional variables on crowdworkers’ perceptions of working conditions (columns II and IV).

Table 6:

Work satisfaction and attractiveness of platform work, ordinal logistic regressions (odds ratios, in brackets robust std. err.)

Work satisfactionPlatform work as future
IIIIIIIV
Constellations of labour control (reference: pure customer control)
• Automated despotism1.64 (0.38)*1.11 (0.30)1.63 (0.39)*1.20 (0.32)
• Gamification1.17 (0.25)0.97 (0.24)1.60 (0.36)*1.34 (0.33)
• Gamified automation2.49 (0.49)**1.27 (0.34)3.77 (0.82)**1.42 (0.36)
Perceptions of platform work
Income uncertainty is a worry0.70 (0.05)**1.22 (0.08)**
Have to work flexibly (late nights…)1.35 (0.09)**1.11 (0.09)
Feel rushed and under time pressure0.75 (0.06)**1.07 (0.09)
Feely performance in constantly monitored0.97 (0.08)1.02 (0.08)
Pay on the platform is reasonable1.71 (0.17)**1.30 (0.12)**
Career opportunities are available1.14 (0.11)3.03 (0.33)**
Work is interesting4.96 (0.66)**1.42 (0.15)**
Feel part of a team1.10 (0.09)1.23 (0.11)*
Dependence
• Share of platform income in total income1.09 (0.07)1.23 (0.10)**1.11 (0.08)1.01 (0.08)
• Typical weekly working hours on platform1.19 (0.08)**1.08 (0.08)1.36 (0.09)**1.22 (0.09)**
• Primary employment status (reference: full-time regular job)
 ◦ Part-time employed0.65 (0.13)*1.15 (0.27)0.78 (0.16)0.84 (0.18)
 ◦ Freelancer0.75 (0.15)0.70 (0.17)1.39 (0.25)1.36 (0.27)
 ◦ Student0.84 (0.21)1.32 (0.41)0.58 (0.17)0.72 (0.23)
 ◦ Other (retired, unemployed…)0.99 (0.28)1.62 (0.48)1.74 (0.44)*1.99 (0.50)**
Socio-demographics
• Women (compared to men)1.09 (0.15)0.73 (0.13)1.19 (0.18)0.93 (0.15)
• Tertiary education0.78 (0.12)1.28 (0.22)0.64 (0.10)**0.69 (0.11)*
• Country of residence (US compared to Germany)1.32 (0.23)1.05 (0.21)1.92 (0.32)**1.00 (0.17)
• Age0.98 (0.01)**0.98 (0.01)**1.00 (0.01)1.03 (0.01)**
Pseudo R20.050.320.100.26
N1,0469431,007927

Source: Krzywdzinski and Gerber, 2019. * p < 0.05, ** p < 0.01. Controls for platform dummies.

Let’s start with columns I and III. The picture is clear: gamified automation is associated with the most positive, and the ‘pure customer control’ constellation with the least positive results, while gamification and automated despotism approaches are in the middle. Taking an example: people working under gamified automation have a 2.49 times higher odds of reporting high work satisfaction than workers in the pure customer control category. The odds of describing platform work as something people want to pursue in the long term is 3.77 times higher under gamified automation than in the ‘pure customer control’ category.

Dependence on the platforms and socio-demographic variables play an important role regarding the long-term attractiveness of platform work. Those who are ‘100%’ freelancers tend to have lower work satisfaction on platforms compared to people doing platform work as a side-job in addition to regular full-time employment – at the same time they have higher odds of considering platform work as their long-term prospect. It is very important to distinguish between these groups in the analysis of labour control regimes on platforms. With regard to socio-demographic variables, people with academic education also show a significantly weaker attachment to platform work than people without. For US-Americans, platform work often represents a more long-term income source than for Germans – which can be explained by the different institutional regulations of the labour markets and the different social security systems (Krzywdzinski and Gerber, 2020). Work satisfaction and the interest in doing platform work on a long-term basis are slightly higher if people work more hours on the platform and if the share of platform income in total income is higher. This might be due to a self-selection effect: people dissatisfied with platform work tend to leave this labour market segment.

How can we explain the particularly strong satisfaction with work in gamified automation? Columns II and IV provide some explanation. When we include variables on the crowdworkers’ perception of working conditions, the forms of labour control lose their explanatory value in the model. Apparently, the forms of labour control are associated with certain patterns of how working conditions are perceived, with interesting work, reasonable pay, flexibility requirements, income uncertainty and time pressure being the most relevant for work statistics. For the intention to pursue platform work in the long term, career opportunities, interesting work, reasonable pay, team relations, and income uncertainty play a role.

How are perceptions of working conditions and forms of labour control related? Table 7 shows the results of ordinal logistic regressions with two groups of dependent variables. The first group of variables represents potential outcomes of labour control regarding stress, whereby we focus on income volatility/uncertainty, flexibility requirements, time pressure and perceptions of surveillance. The second group comprises variables related to general perceptions of working conditions (fairness of pay, career opportunities, work contents, social cohesion).

Table 7:

Perceptions of platform work, ordinal logistic regressions (odds ratios, in brackets robust std. err.)

Dependent variables: outcomes of labour controlDependent variables: general working conditions
Independent variablesIncome uncertainty is a worryHave to work flexibly (late nights, weekends)Feel rushed and under time pressureFeel my performance is constantly monitoredPay on the platform is reasonableCareer opportunities availableWork is interestingFeel part of a team
Constellations of labour control (reference: pure customer control)
• Automated despotism0.97 (0.22)1.42 (0.33)0.83 (0.18)1.06 (0.24)1.94 (0.44)**1.45 (0.33)1.77 (0.44)*1.41 (0.34)
• Gamification1.05 (0.25)1.16 (0.26)1.16 (0.23)2.02 (0.44)**1.57 (0.34)*1.67 (0.37)*1.27 (0.27)1.05 (0.23)
• Gamified automation1.36 (0.30)1.66 (0.35)*1.49 (0.29)*2.14 (0.46)**3.05 (0.61)**4.33 (0.89)**2.80 (0.58)**2.81 (0.59)**
Dependence
• Share of platform income in total income1.16 (0.07)*1.13 (0.07)*1.13 (0.07)*1.00 (0.06)1.20 (0.08)**1.05 (0.6)0.89 (0.06)1.09 (0.07)
• Typical weekly working hours on platform1.26 (0.08)**1.39 (0.09)**1.08 (0.07)1.22 (0.08)**0.94 (0.06)1.21 (0.07)**1.30 (0.08)**1.36 (0.09)**
• Primary employment status (reference: full-time regular job)
 ◦ Part-time employed1.86 (0.37)**0.96 (0.18)1.04 (0.21)1.10 (0.20)0.71 (0.13)0.90 (0.18)0.53 (0.10)**0.78 (0.15)
 ◦ Freelancer1.67 (0.34)**1.48 (0.26)*0.88 (0.17)1.02 (0.18)0.95 (0.18)1.13 (0.22)1.00 (0.19)0.70 (0.13)
 ◦ Student1.28 (0.30)0.88 (0.22)0.94 (0.25)1.16 (0.28)0.73 (0.19)0.68 (0.19)0.52 (0.15)*0.68 (0.17)
 ◦ Other (retired, unemployed…)2.09 (0.51)**1.33 (0.32)0.80 (0.17)1.48 (0.32)0.78 (0.19)1.24 (0.32)0.69 (0.19)0.80 (0.18)
Socio-demographics
• Women (compared to men)1.09 (0.15)1.36 (0.18)*1.15 (0.16)1.03 (0.14)1.09 (0.14)1.29 (0.18)1.62 (0.24)**1.29 (0.17)
• Tertiary education1.31 (0.18)0.87 (0.12)1.53 (0.22)**1.42 (0.21)*0.64 (0.09)**0.84 (0.13)0.60 (0.09)**0.82 (0.12)
• Country of residence (US compared to Germany)1.97 (0.31)**1.61 (0.25)**1.21 (0.19)2.07 (0.34)**1.47 (0.25)*2.10 (0.36)**1.56 (0.26)**1.56 (0.25)**
Age0.98 (0.01)*0.99 (0.01)1.00 (0.00)0.98 (0.01)*0.96 (0.00)**0.97 (0.01)**0.99 (0.01)1.00 (0.01)
Pseudo R20.050.070.050.050.080.090.070.06
N1,0391,0481,0481,0191,0489891,0481,031

Source: Krzywdzinski and Gerber, 2019. * p < 0.05, ** p < 0.01. Controls for platform dummies.

It is important to note that (ordinal logistic) regressions do not necessarily have to imply causal relationships. With regards to the first column we can, however, assume causal relationships based on theoretical reasoning: ranking systems (gamification) can be expected to create competition und uncertainty, flexibility and time pressure; automated controls can be expected to amplify time pressure and create the feeling of being under constant surveillance. We call this group of variables therefore ‘outcomes of labour control’. In the case of the second group of variables, we do not assume a specific causal relationship but interpret the outcomes as correlation only. It might be that constellations of labour control that exert particularly high performance pressures are viable only in arrangements that provide sufficiently good working conditions regarding pay, career opportunities and other factors, in order to retain platform workers.

When we look at the outcomes of labour control, gamified automation stands out. The permanent control of performance is perceived as particularly pronounced, the time pressure and income insecurity are highest, as well as the pressure to work flexibly (in the evening, on weekends and so on). For instance, workers in gamified automation have 1.49 times higher odds of feeling rushed and under time pressure than workers in the ‘pure customer control’ regime; they have 2.14 times higher odds of complaining about permanent performance monitoring. However, the differences with the other forms of labour control are gradual: for example, the intensity of performance monitoring in the case of ‘gamification’ is similar to that of gamified automation.

Given that gamified automation is linked to the highest stress levels among the labour control constellations examined here, the high satisfaction with work under this form of labour control reported in Table 6 (column I) seems all the more puzzling. The second group of variables considered in Table 7 offers the potential explanation. Workers working under gamified automation report the highest level of satisfaction with their pay levels (3.05 times higher than the ‘pure customer control’ category, twice as high as the ‘gamification’ category), the by far best career opportunities (4.33 times higher than in the case of ‘pure customer control’), the most interesting work content (2.80 times higher) and the strongest perception of being part of a team (2.81 times higher).

As we have emphasised earlier, we do not see the forms of labour control as the cause of these general working conditions. Rather, gamified automation creates performance pressure and thus stress. At the same time, however, it is mainly used in jobs with somewhat higher pay, more interesting work content, and more opportunities for cooperation within the team. We argue, therefore, that gamified automation represents a powerful combination of labour control forms, which is however only viable when the general working conditions are attractive enough to bind workers to the platforms and generate acceptance of this form of labour control. Gerber (2021) analyses how, in particular, macrotask platforms specialising in particularly complex and demanding tasks try to create and maintain communities of specialists. The outcomes of these efforts are visible here.

In addition, the living and employment situations of platform workers play a role. Freelancers subject themselves to the flexibility pressure of platform work (that is, the need to work at night and at weekends as well) much more than platform workers who, in addition to the platform, also have a regular full-time or part-time job and perform this work in free timeslots. Freelancers also see significantly more opportunities for advancement in platform work and in addition they feel much less replaceable in this activity than people who perform platform work alongside another regular job.

With regard to the sociodemographic control variables considered, crowdworkers with a university degree are less satisfied with the income from platform work, complain about more time pressure and monitoring and rate the work as less interesting compared to platform workers without a university degree. A comparison between the United States and Germany shows that institutional differences also matter. US-Americans emphasise both the positive and negative aspects of platform work much more than Germans. Since social security in the United States is much weaker than in Germany, platform work plays a much greater role as last resort to secure an income if other forms of employment are not available. Because of this dependency, US-Americans feel the time and performance pressure on the platforms more strongly, but at the same time also evaluate income opportunities in platform work more positively.

Conclusions

The article connects to a lively discussion on the topic of labour control within platform work. To this end, we discuss empirical data from research on a specific type of platform work, namely crowdwork. We argue that the increasingly popular notion of algorithmic management is too broad to serve as an analytical concept, as it entails a variety of different labour control regimes. There is a need for a systematic discussion of the diversity and types of labour control in platform work, as Kellog et al (2019), Griesbach et al (2019) or Wood (2020) suggest.

If we put the role of customers in controlling work outputs aside, there are two most relevant forms of labour control in crowdwork: direct control mainly takes the form of automated output control via technical platform infrastructure, while indirect control aiming at creating motivation and commitment is mainly exerted through ranking and reputation systems that create competition and which we call gamification. Many existing studies focus on a few prominent ride-sharing and delivery platforms that are indeed characterised by pronounced direct control. However, especially in the realm of crowdwork jobs, forms of indirect, ‘normative’ control ‘seeking to win their [the platform workers’] hearts and minds through feelings of “fun” and excitement’ (Kellog et al, 2019: 388) play an important role. We follow Kellog et al’s (2019) recommendation to analyse the role of this indirect control more systematically.

The combination of automation and gamification creates four potential constellations. First, some tasks remain regulated by pure customer control; second, automated despotism, that is constellations that use automated task checks; third, gamification constellations that rely on RRS to influence income opportunities; and fourth, gamified automation that combines automated controls with RRS. These constellations are not exclusively identical with a specific platform type; we find different mixes of labour control on crowdwork platforms depending on which tasks and jobs are offered.

Regarding work-related outcomes, gamified automation stands out. It displays the strongest combination of control forms and generates the highest stress level among the survey participants. At the same time, it also shows the highest work satisfaction levels. This can be attributed to the positively perceived work contents, salaries and opportunities for advancement. Put differently: gamified automation seems to be a kind of ‘golden cage’. It is stressful, but such stress goes along with attractive tasks and working conditions. Compared to this, platforms that use neither automated output control nor gamification – the ‘pure customer control’ category – score the worst with regards to work satisfaction. This constellation is more strongly associated with microwork, partially lower stress levels, but also less interesting work contents, no career opportunities and lower pay.

These findings suggest scepticism regarding a simple ‘control – resistance’ theoretical model, such as that proposed by Kellog et al (2019). Certainly, the forms of labour control in platform work evoke resistance: in some areas as collective action, in others as individualised shirking strategies. However, a critical labour process perspective must also be able to deal with the fact that in relevant parts of the platform workforce there is a positive perception of earnings opportunities and work contents that leads to an acceptance of algorithmic forms of labour control. Labour agency often manifests itself not in resistance but in the construction of individual career paths that sometimes remain in platform work, sometimes lead out of platform work into other labour market areas.

Why is this so? We argue that explanations of the variety of perceptions of working conditions must combine platform-specific factors (for example, disciplining and rewarding mechanisms of labour control, working conditions) and the dependency of platform workers on platforms: our data shows significant differences between those who work on platforms in addition to a regular employment as opposed to those who are self-employed and rely more strongly, if not fully, on their income from platform work. Self-employed respondents describe platform work as particularly stressful, but also as a particular opportunity. Our results imply that working on the same platform does not necessarily lead to shared work experiences, in particular due to different degrees of economic dependence. In terms of organising and collective action, one implication could be that rather than a universal approach, fragments of a platforms’ workforce should be specifically addressed, for instance precarious part-timers or professional freelancers. Labour process theory has historically focused on workplace-internal factors, but needs to include labour market conditions in a more systematic way.

One major limitation of this study is the need to rely on information on labour control that is provided by workers. An important point for future research is to compare this information with a more detailed analysis of the technical infrastructures, including their invisible aspects. This could be done with a comparative research design combining qualitative and quantitative evidence. Another limitation is the focus on crowdwork. We expect that forms of labour control will differ in location-dependent platform work. For future research, the development of a more comprehensive typology for the whole field of platform work can be a promising task.

Conflict of interest

The authors declare that there is no conflict of interest.

References

  • Altenried, M. (2019) The Platform as Factory: Crowdwork and the Hidden Labour behind Artificial Intelligence, SSRN Scholarly Paper ID 3486472, Rochester, NY: Social Science Research Network, https://papers.ssrn.com/abstract=3486472.

    • Search Google Scholar
    • Export Citation
  • Aneesh, A. (2009) Global labor: algocratic modes of organization, Sociological Theory, 27(4): 34770. doi: 10.1111/j.1467-9558.2009.01352.x

    • Search Google Scholar
    • Export Citation
  • Bain, P. and Taylor, P. (2000) Entrapped by the ‘electronic panopticon’? Worker resistance in the call centre, New Technology, Work and Employment, 15(1): 218. doi: 10.1111/1468-005X.00061

    • Search Google Scholar
    • Export Citation
  • Berg, J., Furrer, M., Harmon, E., Rani, U. and Six Silberman, M. (2018) Digital labour platforms and the future of work. Towards Decent Work in the Online World, Geneva: ILO.

    • Search Google Scholar
    • Export Citation
  • Bergvall‐Kåreborn, B. and Howcroft, D. (2014) Amazon mechanical turk and the commodification of labour, New Technology, Work and Employment, 29(3): 21323, https://doi.org/10.1111/ntwe.12038.

    • Search Google Scholar
    • Export Citation
  • Bonin, H. and Rinne, U. (2017) Omnibusbefragung zur verbesserung der datenlage neuer beschäftigungsformen, Kurzexpertise im Auftrag des Bundesministeriums für Arbeit und Soziales (IZA Research Report Nr. 80), Bonn: IZA, https://www.iza.org/publications/r/188/omnibusbefragung-zur-verbesserung-derdatenlage-neuer-beschaftigungsformen.

    • Search Google Scholar
    • Export Citation
  • Braverman, H. (1974) Labor and Monopoly Capital, New York, NY: Monthly Review.

  • Burawoy, M. (1985) Manufacturing Consent: Changes in the Labor Process under Monopoly Capitalism, Chicago, IL: University of Chicago Press.

    • Search Google Scholar
    • Export Citation
  • Burawoy, M. (1990) The Politics of Production: Factory Regimes Under Capitalism and Socialism, London: Verso.

  • Bureau of Labor Statistics (2018) Electronically mediated work: new questions in the contingent worker supplement, https://www.bls.gov/opub/mlr/2018/article/electronically-mediated-work-new-questions-in-the-contingent-worker-supplement.htm.

    • Search Google Scholar
    • Export Citation
  • Difallah, D., Filatova, E. and Ipeirotis, P. (2018) Demographics and dynamics of mechanical turk workers, in Proceedings of the Eleventh ACM International Conference on Web Search and Data Mining, New York, NY: Association for Computing Machinery, pp 13543, WSDM ’18, https://doi.org/10.1145/3159652.3159661.

    • Search Google Scholar
    • Export Citation
  • Edwards, R. (1979) Contested Terrain. The Transformation of the Workplace in the Twentieth Century, New York, NY: Basic Books.

  • Elmer, M., Herr, B., Klaus, D. and Gegenhuber, T. (2019) Platform workers centre stage! Taking stock of current debates and approaches for improving the conditions of platform work in Europe, Working Paper No. 140, Düsseldorf: Hans Böckler Stiftung.

    • Search Google Scholar
    • Export Citation
  • Fieseler, C., Bucher, E. and Hoffmann, C.P. (2019) Unfairness by design? The perceived fairness of digital labor on crowdworking platforms, Journal of Business Ethics, 156(4): 9871005. doi: 10.1007/s10551-017-3607-2

    • Search Google Scholar
    • Export Citation
  • Friedman, A.L. (1977) Industry and Labour: Class Struggle at Work and Monopoly Capitalism, New York, NY: Springer.

  • Gandini, A. (2016) The Reputation Economy: Understanding Knowledge Work in Digital Society, New York, NY: Springer.

  • Gerber, C. (2021) Community building on crowdwork platforms: autonomy and control of online workers?, Competition and Change, 25(2): 190211. doi: 10.1177/1024529420914472

    • Search Google Scholar
    • Export Citation
  • Gerber, C. and Krzywdzinski, M. (2019) Brave new digital work? New forms of performance control in crowdwork, in V. Steven and A. Kovalainen (eds) Work and Labor in the Digital Age, Bingley: Emerald, pp 12143.

    • Search Google Scholar
    • Export Citation
  • Griesbach, K., Reich, A., Elliott-Negri, L. and Milkman, R. (2019) Algorithmic control in platform food delivery work, Socius, 5(January): 2378023119870041. doi: 10.1177/2378023119870041

    • Search Google Scholar
    • Export Citation
  • Heiland, H. and Schaupp, S. (2020) Digitale Atomisierung oder neue Arbeitskämpfe? widerständige Solidaritätskulturen in der plattformvermittelten Kurierarbeit, Momentum Quarterly-Zeitschrift für Sozialen Fortschritt, 9(2): 5067.

    • Search Google Scholar
    • Export Citation
  • Huws, U., Spencer, N. and Joyce, S. (2016) Crowd work in Europe: preliminary results from a survey in the UK, Sweden, Germany, Austria and the Netherlands FEPS Studies December 2016, Brussels: Foundation for European Progressive Studies.

    • Search Google Scholar
    • Export Citation
  • Irani, L. (2015) The cultural work of microwork, New Media & Society, 17(5): 720739, doi: 10.1177/1461444813511926.

  • Katz, L.F. and Krueger, A.B. (2019) The rise and nature of alternative work arrangements in the United States, 1995–2015, ILR Review, 72(2): 382416. doi: 10.1177/0019793918820008

    • Search Google Scholar
    • Export Citation
  • Kellogg, K.C., Valentine, M.A. and Christin, A. (2019) Algorithms at work: the new contested terrain of control, Academy of Management Annals, 14(1): 366410. doi: 10.5465/annals.2018.0174

    • Search Google Scholar
    • Export Citation
  • Kittur, A., Nickerson, J.V., Bernstein, M., Gerber, E., Shaw, A., Zimmerman, J., Lease, M. and Horton, J. (2013) The future of crowd work, in Proceedings of the 2013 Conference on Computer Supported Cooperative Work, New York, NY: ACM, pp 130118.

    • Search Google Scholar
    • Export Citation
  • Krzywdzinski, M. and Gerber, C. (2019) Working conditions of crowdworkers, Survey of Crowdworkers in Germany and the United States, Version 1.00, Berlin: WZB Berlin Social Science Center, unpublished data set.

    • Search Google Scholar
    • Export Citation
  • Krzywdzinski, M. and Gerber, C. (2020) Varieties of Platform Work: Platforms and Social Inequality in Germany and the United States, Weizenbaum Series #7, Berlin: Weizenbaum Institute for the Networked Society, https://doi.org/10.34669/wi.ws/7.

    • Search Google Scholar
    • Export Citation
  • Lawrence, T.B. and Robinson, S.L. (2007) Ain’t misbehavin: workplace deviance as organizational resistance, Journal of Management, 33(3): 37894. doi: 10.1177/0149206307300816

    • Search Google Scholar
    • Export Citation
  • Lee, M.K., Kusbit, D., Metsky, E. and Dabbish, L. (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, New York: Association for Computing Machinery, pp 160312, CHI ’15, https://doi.org/10.1145/2702123.2702548.

    • Search Google Scholar
    • Export Citation
  • Lehdonvirta, V. (2016) Algorithms that divide and unite: delocalisation, identity and collective action in ‘Microwork’, in J. Flecker (ed) Space, Place and Global Digital Work, London: Palgrave Macmillan, pp 5380, https://doi.org/10.1057/978-1-137-48087-3_4.

    • Search Google Scholar
    • Export Citation
  • Lehdonvirta, V. (2018) Flexibility in the gig economy: managing time on three online piecework platforms, New Technology, Work and Employment, 33(1): 1329. doi: 10.1111/ntwe.12102

    • Search Google Scholar
    • Export Citation
  • Leimeister, J.M., Durward, D. and Zogaj, S. (2016) Crowdworker in Deutschland: Eine empirische Studie zum Arbeitsumfeld auf externen Crowdsourcing-Plattformen, HBS Study, No. 323, Düsseldorf: Hans Böckler Stiftung.

    • Search Google Scholar
    • Export Citation
  • Möhlmann, M. and Zalmanson, L. (2017) Hands on the wheel: navigating algorithmic management and uber drivers, 38th ICIS Proceedings.

  • Pesole, A., Brancati, M.C., Fernández-Macías, E., Biagi, F. and Vázquez, I.G. (2018) Platform workers in Europe, Luxembourg: Publications Office of the European Union.

    • Search Google Scholar
    • Export Citation
  • Pongratz, H.J. and Bormann, S. (2017) Online-arbeit auf internet-plattformen: empirische befunde zum ‘crowdworking’ in Deutschland, AIS-Studien, 10(2): 15881.

    • Search Google Scholar
    • Export Citation
  • Popiel, P. (2017) ‘Boundaryless’ in the creative economy: assessing freelancing on upwork, Critical Studies in Media Communication, 34(3): 22033. doi: 10.1080/15295036.2017.1282618

    • Search Google Scholar
    • Export Citation
  • Robinson, H.C. and Vallas, S.P. (2020) Variations in the lived experience of risk among ride-hailing drivers in boston, Paper presented at the 32nd SASE Annual Conference, Cologne: SASE.

    • Search Google Scholar
    • Export Citation
  • Rosenblat, A. (2018) Uberland: How Algorithms Are Rewriting the Rules of Work, Oakland, CA: University of California Press.

  • Rosenblat, A. and Stark, L. (2016) Algorithmic labor and information asymmetries: a case study of uber’s drivers, International Journal of Communication, 10: 375884.

    • Search Google Scholar
    • Export Citation
  • Schor, J. (2015) Homo varians: diverse economic behaviors in new sharing markets, Unpublished Paper, Boston College.

  • Schörpf, P., Flecker, J., Schönauer, A. and Eichmann, H. (2017) Triangular love–hate: management and control in creative crowdworking, New Technology, Work and Employment, 32(1): 4358, https://doi.org/10.1111/ntwe.12080.

    • Search Google Scholar
    • Export Citation
  • Shapiro, A. (2018) Between autonomy and control: strategies of arbitrage in the ‘on-demand’ economy, New Media & Society, 20(8): 295471. doi: 10.1177/1461444817738236

    • Search Google Scholar
    • Export Citation
  • Smith, C. (2006) The double indeterminacy of labour power: labour effort and labour mobility, Work, Employment and Society, 20(2): 389402. doi: 10.1177/0950017006065109

    • Search Google Scholar
    • Export Citation
  • Thompson, P. (1983) The Nature of Work: An Introduction to Debates on the Labour Process, Basingstoke: Macmillan International Higher Education.

    • Search Google Scholar
    • Export Citation
  • Wells, K.J., Attoh, K. and Cullen, D. (2020) ‘Just-in-place’ labor: driver organizing in the uber workplace, Environment and Planning A: Economy and Space, August, 0308518X20949266, https://doi.org/10.1177/0308518X20949266.

    • Search Google Scholar
    • Export Citation
  • Wood, A. (2020) Despotism on Demand. How Power Operates in the Flexible Workplace, Ithaca, NY: ILR Press.

  • Wood, A.J., Graham, M., Lehdonvirta, V. and Hjorth, I. (2019) Good gig, bad gig: autonomy and algorithmic control in the global gig economy, Work, Employment and Society, 33(1): 5675. doi: 10.1177/0950017018785616

    • Search Google Scholar
    • Export Citation
  • Woodcock, J. (2016) Working the Phones: Control and Resistance in Call Centres, London: Pluto Press.

  • Woodcock, J. and Johnson, M.R. (2018) Gamification: what it is, and how to fight it, The Sociological Review, 66(3): 54258. doi: 10.1177/0038026117728620

    • Search Google Scholar
    • Export Citation
  • Zhang, L. (2008) Lean production and labor controls in the Chinese automobile industry in an age of globalization, International Labor and Working-Class History, 73(1): 2444, doi: 10.1017/S0147547908000033

    • Search Google Scholar
    • Export Citation

Appendix

Table:

Survey participants by platform

PlatformPlatform typeCountryNumber of registered crowdworkersMode of survey distributionNumber of (valid) respondents*
P1MicroGermany>500,000Paid job on platform278
P2MicroGermany50,000–100,000Paid job on platform147
P3MicroGermany<50,000Platform email list39
P4MicroGermany>500,000Platform email list193
P5MacroGermany50,000–100,000Platform email list2
P6MacroGermany100,000–500,000Social media and email43
P7MicroUS100,000–500,000Paid job on platform190
P8MicroUS<50,000Social media12
P9MacroUS>500,000Social media and email200
P10MacroUS>500,000Social media11
P11MacroUS>500,000Social media9
P12MacroUS>500,000Social media4
P13MacroUS>500,000Social media1
P14MacroUS>500,000Social media1
P15MacroUS<50,000Social media1

The respondents from each platform may include both German and US residents.

  • Altenried, M. (2019) The Platform as Factory: Crowdwork and the Hidden Labour behind Artificial Intelligence, SSRN Scholarly Paper ID 3486472, Rochester, NY: Social Science Research Network, https://papers.ssrn.com/abstract=3486472.

    • Search Google Scholar
    • Export Citation
  • Aneesh, A. (2009) Global labor: algocratic modes of organization, Sociological Theory, 27(4): 34770. doi: 10.1111/j.1467-9558.2009.01352.x

    • Search Google Scholar
    • Export Citation
  • Bain, P. and Taylor, P. (2000) Entrapped by the ‘electronic panopticon’? Worker resistance in the call centre, New Technology, Work and Employment, 15(1): 218. doi: 10.1111/1468-005X.00061

    • Search Google Scholar
    • Export Citation
  • Berg, J., Furrer, M., Harmon, E., Rani, U. and Six Silberman, M. (2018) Digital labour platforms and the future of work. Towards Decent Work in the Online World, Geneva: ILO.

    • Search Google Scholar
    • Export Citation
  • Bergvall‐Kåreborn, B. and Howcroft, D. (2014) Amazon mechanical turk and the commodification of labour, New Technology, Work and Employment, 29(3): 21323, https://doi.org/10.1111/ntwe.12038.

    • Search Google Scholar
    • Export Citation
  • Bonin, H. and Rinne, U. (2017) Omnibusbefragung zur verbesserung der datenlage neuer beschäftigungsformen, Kurzexpertise im Auftrag des Bundesministeriums für Arbeit und Soziales (IZA Research Report Nr. 80), Bonn: IZA, https://www.iza.org/publications/r/188/omnibusbefragung-zur-verbesserung-derdatenlage-neuer-beschaftigungsformen.

    • Search Google Scholar
    • Export Citation
  • Braverman, H. (1974) Labor and Monopoly Capital, New York, NY: Monthly Review.

  • Burawoy, M. (1985) Manufacturing Consent: Changes in the Labor Process under Monopoly Capitalism, Chicago, IL: University of Chicago Press.

    • Search Google Scholar
    • Export Citation
  • Burawoy, M. (1990) The Politics of Production: Factory Regimes Under Capitalism and Socialism, London: Verso.

  • Bureau of Labor Statistics (2018) Electronically mediated work: new questions in the contingent worker supplement, https://www.bls.gov/opub/mlr/2018/article/electronically-mediated-work-new-questions-in-the-contingent-worker-supplement.htm.

    • Search Google Scholar
    • Export Citation
  • Difallah, D., Filatova, E. and Ipeirotis, P. (2018) Demographics and dynamics of mechanical turk workers, in Proceedings of the Eleventh ACM International Conference on Web Search and Data Mining, New York, NY: Association for Computing Machinery, pp 13543, WSDM ’18, https://doi.org/10.1145/3159652.3159661.

    • Search Google Scholar
    • Export Citation
  • Edwards, R. (1979) Contested Terrain. The Transformation of the Workplace in the Twentieth Century, New York, NY: Basic Books.

  • Elmer, M., Herr, B., Klaus, D. and Gegenhuber, T. (2019) Platform workers centre stage! Taking stock of current debates and approaches for improving the conditions of platform work in Europe, Working Paper No. 140, Düsseldorf: Hans Böckler Stiftung.

    • Search Google Scholar
    • Export Citation
  • Fieseler, C., Bucher, E. and Hoffmann, C.P. (2019) Unfairness by design? The perceived fairness of digital labor on crowdworking platforms, Journal of Business Ethics, 156(4): 9871005. doi: 10.1007/s10551-017-3607-2

    • Search Google Scholar
    • Export Citation
  • Friedman, A.L. (1977) Industry and Labour: Class Struggle at Work and Monopoly Capitalism, New York, NY: Springer.

  • Gandini, A. (2016) The Reputation Economy: Understanding Knowledge Work in Digital Society, New York, NY: Springer.

  • Gerber, C. (2021) Community building on crowdwork platforms: autonomy and control of online workers?, Competition and Change, 25(2): 190211. doi: 10.1177/1024529420914472

    • Search Google Scholar
    • Export Citation
  • Gerber, C. and Krzywdzinski, M. (2019) Brave new digital work? New forms of performance control in crowdwork, in V. Steven and A. Kovalainen (eds) Work and Labor in the Digital Age, Bingley: Emerald, pp 12143.

    • Search Google Scholar
    • Export Citation
  • Griesbach, K., Reich, A., Elliott-Negri, L. and Milkman, R. (2019) Algorithmic control in platform food delivery work, Socius, 5(January): 2378023119870041. doi: 10.1177/2378023119870041

    • Search Google Scholar
    • Export Citation
  • Heiland, H. and Schaupp, S. (2020) Digitale Atomisierung oder neue Arbeitskämpfe? widerständige Solidaritätskulturen in der plattformvermittelten Kurierarbeit, Momentum Quarterly-Zeitschrift für Sozialen Fortschritt, 9(2): 5067.

    • Search Google Scholar
    • Export Citation
  • Huws, U., Spencer, N. and Joyce, S. (2016) Crowd work in Europe: preliminary results from a survey in the UK, Sweden, Germany, Austria and the Netherlands FEPS Studies December 2016, Brussels: Foundation for European Progressive Studies.

    • Search Google Scholar
    • Export Citation
  • Irani, L. (2015) The cultural work of microwork, New Media & Society, 17(5): 720739, doi: 10.1177/1461444813511926.

  • Katz, L.F. and Krueger, A.B. (2019) The rise and nature of alternative work arrangements in the United States, 1995–2015, ILR Review, 72(2): 382416. doi: 10.1177/0019793918820008

    • Search Google Scholar
    • Export Citation
  • Kellogg, K.C., Valentine, M.A. and Christin, A. (2019) Algorithms at work: the new contested terrain of control, Academy of Management Annals, 14(1): 366410. doi: 10.5465/annals.2018.0174

    • Search Google Scholar
    • Export Citation
  • Kittur, A., Nickerson, J.V., Bernstein, M., Gerber, E., Shaw, A., Zimmerman, J., Lease, M. and Horton, J. (2013) The future of crowd work, in Proceedings of the 2013 Conference on Computer Supported Cooperative Work, New York, NY: ACM, pp 130118.

    • Search Google Scholar
    • Export Citation
  • Krzywdzinski, M. and Gerber, C. (2019) Working conditions of crowdworkers, Survey of Crowdworkers in Germany and the United States, Version 1.00, Berlin: WZB Berlin Social Science Center, unpublished data set.

    • Search Google Scholar
    • Export Citation
  • Krzywdzinski, M. and Gerber, C. (2020) Varieties of Platform Work: Platforms and Social Inequality in Germany and the United States, Weizenbaum Series #7, Berlin: Weizenbaum Institute for the Networked Society, https://doi.org/10.34669/wi.ws/7.

    • Search Google Scholar
    • Export Citation
  • Lawrence, T.B. and Robinson, S.L. (2007) Ain’t misbehavin: workplace deviance as organizational resistance, Journal of Management, 33(3): 37894. doi: 10.1177/0149206307300816

    • Search Google Scholar
    • Export Citation
  • Lee, M.K., Kusbit, D., Metsky, E. and Dabbish, L. (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, New York: Association for Computing Machinery, pp 160312, CHI ’15, https://doi.org/10.1145/2702123.2702548.

    • Search Google Scholar
    • Export Citation
  • Lehdonvirta, V. (2016) Algorithms that divide and unite: delocalisation, identity and collective action in ‘Microwork’, in J. Flecker (ed) Space, Place and Global Digital Work, London: Palgrave Macmillan, pp 5380, https://doi.org/10.1057/978-1-137-48087-3_4.

    • Search Google Scholar
    • Export Citation
  • Lehdonvirta, V. (2018) Flexibility in the gig economy: managing time on three online piecework platforms, New Technology, Work and Employment, 33(1): 1329. doi: 10.1111/ntwe.12102

    • Search Google Scholar
    • Export Citation
  • Leimeister, J.M., Durward, D. and Zogaj, S. (2016) Crowdworker in Deutschland: Eine empirische Studie zum Arbeitsumfeld auf externen Crowdsourcing-Plattformen, HBS Study, No. 323, Düsseldorf: Hans Böckler Stiftung.

    • Search Google Scholar
    • Export Citation
  • Möhlmann, M. and Zalmanson, L. (2017) Hands on the wheel: navigating algorithmic management and uber drivers, 38th ICIS Proceedings.

  • Pesole, A., Brancati, M.C., Fernández-Macías, E., Biagi, F. and Vázquez, I.G. (2018) Platform workers in Europe, Luxembourg: Publications Office of the European Union.

    • Search Google Scholar
    • Export Citation
  • Pongratz, H.J. and Bormann, S. (2017) Online-arbeit auf internet-plattformen: empirische befunde zum ‘crowdworking’ in Deutschland, AIS-Studien, 10(2): 15881.

    • Search Google Scholar
    • Export Citation
  • Popiel, P. (2017) ‘Boundaryless’ in the creative economy: assessing freelancing on upwork, Critical Studies in Media Communication, 34(3): 22033. doi: 10.1080/15295036.2017.1282618

    • Search Google Scholar
    • Export Citation
  • Robinson, H.C. and Vallas, S.P. (2020) Variations in the lived experience of risk among ride-hailing drivers in boston, Paper presented at the 32nd SASE Annual Conference, Cologne: SASE.

    • Search Google Scholar
    • Export Citation
  • Rosenblat, A. (2018) Uberland: How Algorithms Are Rewriting the Rules of Work, Oakland, CA: University of California Press.

  • Rosenblat, A. and Stark, L. (2016) Algorithmic labor and information asymmetries: a case study of uber’s drivers, International Journal of Communication, 10: 375884.

    • Search Google Scholar
    • Export Citation
  • Schor, J. (2015) Homo varians: diverse economic behaviors in new sharing markets, Unpublished Paper, Boston College.

  • Schörpf, P., Flecker, J., Schönauer, A. and Eichmann, H. (2017) Triangular love–hate: management and control in creative crowdworking, New Technology, Work and Employment, 32(1): 4358, https://doi.org/10.1111/ntwe.12080.

    • Search Google Scholar
    • Export Citation
  • Shapiro, A. (2018) Between autonomy and control: strategies of arbitrage in the ‘on-demand’ economy, New Media & Society, 20(8): 295471. doi: 10.1177/1461444817738236

    • Search Google Scholar
    • Export Citation
  • Smith, C. (2006) The double indeterminacy of labour power: labour effort and labour mobility, Work, Employment and Society, 20(2): 389402. doi: 10.1177/0950017006065109

    • Search Google Scholar
    • Export Citation
  • Thompson, P. (1983) The Nature of Work: An Introduction to Debates on the Labour Process, Basingstoke: Macmillan International Higher Education.

    • Search Google Scholar
    • Export Citation
  • Wells, K.J., Attoh, K. and Cullen, D. (2020) ‘Just-in-place’ labor: driver organizing in the uber workplace, Environment and Planning A: Economy and Space, August, 0308518X20949266, https://doi.org/10.1177/0308518X20949266.

    • Search Google Scholar
    • Export Citation
  • Wood, A. (2020) Despotism on Demand. How Power Operates in the Flexible Workplace, Ithaca, NY: ILR Press.

  • Wood, A.J., Graham, M., Lehdonvirta, V. and Hjorth, I. (2019) Good gig, bad gig: autonomy and algorithmic control in the global gig economy, Work, Employment and Society, 33(1): 5675. doi: 10.1177/0950017018785616

    • Search Google Scholar
    • Export Citation
  • Woodcock, J. (2016) Working the Phones: Control and Resistance in Call Centres, London: Pluto Press.

  • Woodcock, J. and Johnson, M.R. (2018) Gamification: what it is, and how to fight it, The Sociological Review, 66(3): 54258. doi: 10.1177/0038026117728620

    • Search Google Scholar
    • Export Citation
  • Zhang, L. (2008) Lean production and labor controls in the Chinese automobile industry in an age of globalization, International Labor and Working-Class History, 73(1): 2444, doi: 10.1017/S0147547908000033

    • Search Google Scholar
    • Export Citation
  • 1 WZB Berlin Social Science Center Reichpietschufer, , Germany
  • | 2 Weizenbaum Institute for the Networked Society, , Germany
  • | 3 Helmut Schmidt University Hamburg, , Germany
  • | 4 WZB Berlin Social Science Center, , Germany

Content Metrics

May 2022 onwards Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 19 19 16
PDF Downloads 18 18 14

Altmetrics

Dimensions