Republican freedom and on-demand work

(1)

Republican freedom and on-demand

work

Thesis European Politics and External Relations (Political Science)

Author: Patrick Hoop Student number: 10438114 Project: Alternatives to capitalism Supervisor: dr. Paul Raekstad Second reader: dr. Gordon Arlen Hand-in date: 29-06-2018

Graduate School of Social Sciences, 2018

(2)

Republican freedom and on-demand work

Introduction

Modern-day advocates of the platform economy laud the way that businesses such as Uber, Deliveroo or Helpling use efficient algorithmic matchmaking to ‘free the worker’. Freed from office, nearly anyone can simply activate the app, or login to the online platform of choice to accept a job. This utopian picture is one where anyone can work autonomously, wherever and whenever, without being burdened by the hierarchy of a manager, going to an office, or inflexible working hours. Work is efficiently distributed through platforms, connecting consumer and workers in need of income. The relatively well-paid on-demand workers, often self-employed, can enjoy an alternative arrangement to standard contractual work, enabling them to combine work-hours with other responsibilities more freely (Friedman, 2014). Above all, on-demand work promises freedom. No doubt, some of those working in the platform economy do enjoy this arrangement.

Others however, have a different understanding of the platform economy. Opponents see the specific way that the platform economy operates as a neoliberal device, which uses the veil of entrepreneurial opportunities to mask the dismantling social policies in employment (Wright, 2018). By hiring workers as self-employed contractors for ad-hoc ‘gigs’ rather than making them part of the company as employees, employers can adjust employment and even wages in response to demand conditions. Making employment and wages more flexible, gig employments shift the risk of economic fluctuations onto the workers, leaving them with uncertain employment. Worrying, some authors see that evading traditional regulations may even be a key rationale for establishing digital businesses in the first place (Steward & Stanford, 2017, p. 421). What follows from this critique is the claim that these firms in the platform economy use online platforms and matching algorithms to offer sham employment: disguising what is in substance an employment arrangement into forms of self-employment. This debate shares many features also present in the wider labour market and fits within the broader development of casualisation of labour. One of the key questions in the political arenas on both the European and national level thus became: ‘’should on-demand workers be classified as self-employed contractors, or

workers deserving of a contract? ’’. As is the case with any fundamental question, this dilemma gave

birth to many more questions, mainly concerning whether traditional models of employment are in fact sufficient for this ‘new economy’. The most important question for this thesis is the question to what degree the shape of this new economy might in fact be a controlling force, instead of a liberating force. At the time of writing, many of these practical policy questions still remain unresolved.

(4)

Thesis objective and main arguments

As seen, proponents of the platform economy stress freedom and autonomy. This thesis aims to examine these claims of liberation from the perspective of republican freedom as put forward by Philip 1 Pettit (1997; 2002; 2014), with some additions by Thompson (2013). Re-visiting one of the aforementioned questions, this thesis applies the theoretical lens of republican freedom and workplace republicanism to understand in what way the shape of this new economy is in fact a source of freedom for workers. In other words, we are interested in how ‘control’ is exerted on those working in the on-demand sector. The question which we thus set out to answer is ‘in what way do workers in the

on-demand sector experience control?’ Logically, definite conclusions about the general state of freedom

for workers in the on-demand sector is beyond the scope of this thesis . That said, we acknowledge that 2

using republican freedom carries normative implications for the shape of this thesis and the outcome.

In short, the main argument of this thesis is that on-demand workers actually experience significant forms of managerial control while doing their job, regardless of whether they are currently classified as either ‘self-employed’ or ‘contracted’. Using the typology by De Stefano (2018) which puts forward three distinct forms of control, we argue that on-demand workers are subjected to these forms, albeit very different from the way found in more traditional authority relation. In fact, on-demand workers are subjected to all the three distinct forms of control distinguished by this aforementioned typology, sometimes direct, sometimes ‘outsourced’, either via reputation mechanisms, internal ranking systems, or algorithmic management all together. Logically, being controlled has profound effects on the degree to which on demand workers experience republican freedom. Although not exhaustive, one form of subjection can be found in the fact that rating-systems may partially transfer the power of monitoring and compliance to peers, granting them a deciding ‘voice’ in whether an on-demand worker for instance keeps his or her job. Another form of subjection is due to algorithmic management: the framework of platforms or applications themselves are hardcoded to give unilateral orders and directives, monitor, sanction, subjecting workers to an arbitrary dominationing power. The fact that on-demand workers are subjected in this way erodes the freedom of workers. This is because the theory of republican freedom

1

The word republican here signifies a specific way of thinking about freedom, grounded in political theory and political philosophy. Explicitly, this does not refer to the conservative tradition found in the United States, with which the term is also associated.

2_{While certainly interesting, this would require a comparative analysis of the ways in which these platforms interfere with} workers, which most likely would be problematic in terms of feasibility. This thesis is mainly interested in various forms that control can take.

(5)

sees severe managerial interference with little or no justification in terms of the worker’s interests, as a situation which is unfree for workers.

Consequently, this thesis also explores two preliminary pathways for increasing the republican freedom for on-demand workers. The first pathway seeks to limit the degree to which workers are subject to interference through for instance regulation, collective bargaining and the adequate classification of workers. When it comes to regulation, we see that present-day legislation limits the interfering power of ‘traditional’ managers far more than it does the systemic interfering powers of reputation mechanisms and algorithmic management. Although we provide some suggestions, exactly how to limit domination resulting from algorithmic management and reputation systems via regulation requires additional research. Hereafter, the capability of labour unions to limit domination is reviewed. In short, limiting interference through collective bargaining will be difficult because the nature of on-demand work inhibits collective organizing efforts, making workers hard to find, hard to reach, and difficult to engage (Johnston & Land-Kazlauskas, 2018). Furthermore, classifying on-demand workers as self-employed must be avoided when the goal is to advance republican freedom because it surpasses general employment regulation which protects fundamental principles and rights at work (Drahokoupil, 2016; Aloisi, 2015).

The second pathway examines ways to limit arbitrariness for on-demand workers. Upon examining workplace constitutionalism and workplace democracy, we argue that workplace democracy is preferable. While workplace constitutionalism can raise a ‘minimum floor’, workplace democracy enables workers to protect themselves against arbitrary interference by putting them in a position to contest managerial decisions. Workplace democracy can provide solace for on-demand workers especially, because it encourages transparency in algorithmic management as to see whether decisions are in fact in the interest of workers in a firm. Additionally, on-demand work is at risk of being more arbitrary than regular work since traditional managers may provide (informal) windows for workers to have a say in decision making; a trait which is non-existent in algorithmic management.

The final chapter before the concluding remarks explores ways of reducing the dependence on managerial authority. The idea of a basic income is briefly visited since it reduces the material dependence of workers, increasing their ability to bargain collectively. Lastly, the republican ideal of self-employment is visited. While this ideal does not provide a pathway, we can contrast this ideal with the realities of on-demand work, underlining that the subjection found in on-demand work as an self-employed worker that these situations are far apart.

(6)

On relevance

In what way, than, is a republican examination on the situation of on-demand workers relevant?

First of all, we aim to make a case for two additions to republican thinking coined by Thompson (2013) via examples found in on-demand work. Practically, we accept Pettit’s idea that a person is unfree when in a position where another can interfere with a him or her on the basis of an interest or opinion that need not be shared by this person. However, we diverge from Pettit’s view firstly when we add that this interference can also be the result of a structure or logics of institutions, and secondly when we add that cases of interference are not always known by the actors whom it may concern. By addressing the situations of on-demand workers, this thesis attempts to show that firstly a structure such as a specific framework of an online platform can indeed act as an interfering device, and secondly, that arbitrary interference is not always known to all actors involved.

Secondly, although there are numerous authors concerned with changing labour relations and the shift towards self-employment as found in the platform economy, very few articles consider the situation of on-demand workers explicitly from the scope of republican tradition. This thesis aims to contribute to the tradition of workplace republicanism by expanding the theoretical debate regarding on-demand work. Thirdly, this thesis aims to add to the literature on algorithmic management by demonstrating that workplace republicanism is a broad enough concept to function as a guideline to ensuring worker freedom in workplaces where management takes similar forms.

The fourth and perhaps most important from the perspective of policymakers is found in the argument that on-demand workers are in fact subjected by intransparent managerial power structures in the form of algorithms and reputation mechanisms. We argue that this is erosive of freedom and contrasts with the republican ideal of a free self-employed worker. This in term provides a novel line of reasoning in the ongoing debate about whether on-demand workers should be classified as independent contractors. Importantly, our aim is not to provide a definitive answer to the complex debate whether on-demand workers can, or should be classified as regular employees. That said, assessing the phenomenon of on-demand work through a lens of republican thought does provide us with an image wherein on-demand workers are controlled to a degree that includes the standard criteria for employment status.

(7)

Thesis outline

Before we go into the actual argumentation we briefly outline the chapters of this thesis.

The first chapter aims to create some conceptual clarity between the wide variety of often confusing concepts present in the literature regarding the platform economy, by defining what a platform is, what sets apart on-demand work from ‘crowd-work’ and what constitutes the ‘sharing economy’.

The second chapter will first examine the ideas of Isaiah Berlin about positive and negative freedom. We then use Berlin’s work about negative freedom to make a comparison between liberal ideas about freedom to the republican conception of freedom, to understand where the two diverge and to come to the core of what republican freedom entails. In doing so, this comparison shows that different conceptions of freedom can lead to very different logical results about whether one can be considered free in a specific situation.

The third chapter examines the baseline of republican thinking by Pettit, defining freedom as a situation wherein people are not arbitrarily dominated. Furthermore, two principles coined by Thompson (2013) are examined to help understand how the disciplinary and opaque institutional context experienced by on-demand workers can be dominating. What follows is an elaboration on when a situation can be called ‘interfering’ and ‘arbitrary’ or ‘dominating’, followed by the acknowledgement that these concepts are not easily quantifiable and often dependent on context.

The fourth chapter examines the specifics of workplace republicanism derived from the broader republican ideas examined in chapter three. Here, it is argued that the workplace is one of the areas particularly at-risk of arbitrary domination, because of the pervasiveness of managerial power to interfere in workers’ lives. Furthermore, the conditions of workplace domination by managers or employers are outlined by Hsieh (2005), whereafter we visit the typology by De Stefano (2018) in order to have a systematic way of understanding the forms of control which managerial power can take.

In chapter five, the ideas about workplace republicanism and the typology by De Stefano are used to assess some of the realities of on-demand work. Specifically, we argue that in reputation mechanisms and algorithmic management in practice are dominating devices which render on-demand workers unfree in the republican sense. Chapters six and seven explore two different pathways on how to

(8)

increasing the republican freedom for on-demand workers. Chapter six examines limiting the degree to which workers are subject to interference trough for instance regulation, collective bargaining and the classification of workers. Chapter seven goes into limiting arbitrariness via workplace constitutionalism and workplace democracy. Finally, chapter eight explores reducing the dependence on managerial authority. While this is not a pathway specific to on-demand work, the chapter explores the generally liberating traits of a basic income. Furthermore, it visits the republican ideal of self-employment in order to contrast it to the realities of on-demand workers.

(9)

1: Definitions

Before this thesis moves on to examine some of the central ideas regarding freedom, it is important to create conceptual clarity when it comes to the object of research: on-demand work. Currently, on-demand work remains an active topic of discussion in several political arena's because of its novelty, complexity and political consequences. However, subtle yet important distinctions are often overlooked in the political and public debate (Maselli, 2016). In an attempt to avoid ambiguity, this paragraph will first examine the distinct features of platforms and they differ from each other. Hereafter, the difference between ‘crowd-work’ and ‘on-demand’ work is assessed, before settling on a definition of on-demand work.

First, on-demand work is facilitated by platforms. The term “platform” points to a set of digital arrangements whose algorithms serve to organize and structure economic and social activity (Kenney & Zysman, 2015). The term is used for the digitalization of retail including infrastructures such as Ebay, as well as for the digitalization of service providing infrastructures such as Uber, Lyft, Helpling and Deliveroo. Importantly, service providing platforms are different from the sharing economy; although the lines between the two are not always clear cut, participating in the sharing economy revolves around maximising the use of an underutilized asset (Goudin, 2016) and not so much about buying or selling services through digital means.

When it comes to service providing infrastructures, it is possible to distinguish two different forms of digital service (De Stefano, 2015). One form of digital service is called ‘crowd-work’. Crowdwork is work that is executed globally through online platforms that very often involves so-called microtasks. These microtasks are extremely parcelled activities, often menial and monotonous, which still require some sort of judgement beyond the understanding of artificial intelligence (De Stefano, 2015, p.3). Examples of these microtasks can be looking up specific kinds of information, valuing the appropriateness of a text or completing surveys via online environments such as Amazon Mechanical Turk.

The second and the main object for this thesis is called ‘on-demand work’, a form of digital service distinct from crowdwork. On-demand work is more locally oriented than crowd-work and often seen as the digitized equivalent of more traditional forms of work such as transport, cleaning and running errands. Firms such as Uber, Helpling and Deliveroo provide the platform whereon supply and demand

(10)

for these jobs can be matched, providing a structural framework for those who supply this labour: the on-demand workers. Importantly, Steward & Stanford (2017) posit that many firms utilising platforms for on-demand work retain control over important aspects of the work such as setting prices and standards, and selecting and managing the workforce. 3

In other words, on-demand workers work via online platforms where the shape of their work is organized by clusters of algorithms, traditional when it comes to the type of work. Most importantly, when it comes to on-demand work, some platforms retain control over important aspects of the work such as setting prices and standards, and selecting and managing the workforce. As expressed in the introduction of this thesis, the specific shape that this control may take has consequences for the amount of freedom a worker can enjoy. However, before we can assess their situation by ways of republican freedom, we must first go into republican thinking and workplace republicanism. In order to get a thorough understanding of these two ideas, it is useful to first visit the dichotomy by Berlin (1969), because it provides a baseline in thinking about freedom.

3

(11)

2: Various ideas of freedom

This chapter will first examine Berlin's dichotomy in order to lay out a baseline for thinking about freedom. Examining the ideas of Berlin provides us with the concept of ‘negative freedom’, which thereafter can be compared with ‘non-domination’, a core concept of Republican freedom. In order to provide further clarity, hereafter this chapter will briefly visit how republican freedom relates to Berlin's dichotomy.

2.1: Negative freedom and positive freedom

Berlin's ideas about negative freedom refer to the ideal-type situation wherein an individual does not experience constraints on his or her doing by other people (Berlin, 1969). Being free in this sense means

freedom from interference, or, enjoying ‘non-interference’. When thinking about negative freedom, it is

important to note that constraining factors always include a social element such as is the case when being confronted by acts of coercion, or for instance slavery. This said, negative freedom specifically does not include incapacities or limits derived from factors outside of human control: ‘ ’If I say that I am

unable to jump more than ten feet in the air, or cannot read because I am blind, or cannot understand

the darker pages of Hegel, it would be eccentric to say that I am to that degree enslaved or coerced. ’’

(Berlin, 1969, p. 160). In other words; for an individual to enjoy freedom in Berlin’s idea he or she must not be interfered with by others. Positive freedom on the other hand goes beyond this. Contrasted with the former, positive freedom is to be understood as freedom to,derived from a person's’ wish to be his or her own master (Berlin, 1969). Positive freedom, in turn, requires the individual to take an active part in gaining control or mastery of themselves (Pettit, 1997, p. 17).

2.2: freedom as non-domination and freedom as non-interference

In recent times, there has been an increased interest in the republican idea of ‘non domination’. This interest may stem from the fact that this republican idea quite convincingly provides a response to the idea of freedom as ‘non-interference’, which is quite limited in its scope. As said, comparing the two is helpful in understanding the republican position. The difference between ‘freedom as non-domination’ and ‘freedom as non-interference’ can be best illustrated through the analogy of a horse-rider, also used by Pettit (2014). Wandering through the fields, a horse rider can let the reins hang loose so that a horse

(12)

can head in any direction it chooses. When doing so, the horse rider does not have actual control over the path they go together. Strictly speaking, the horse enjoys freedom from interference - it is under nobodies operative control. While this may be so, that does not mean the horse is free from domination, or the situation of being in a position where another can interfere with the life of an individual at will. In other words, despite the fact that actual interference is not present, the mere possibility of that interference itself is an obstacle to freedom (Thompson, 2013, p.281). This difference can have serious consequences. For example, in the early 19 thcentury the state went from being viewed as a provider of basic liberties ensured via laws and regulations to being viewed as an ‘interfering’ player in peer-to-peer relationships (Pettit, 2014). In terms of working relationships, classical liberal thinkers during this period reasoned that a contractual agreement between two agreeing parties, whether powerful or marginalized, should happen as unconstrained as possible. In doing so classical liberals settled on defining freedom merely as the absence of interference. In turn, some thinkers advocated to give industrialists a carte blanche when it came to setting the terms on hiring workers, effectively dragging basic liberties into the bargaining room (Pettit, 2014). As a result, republican thinkers saw men, woman and even children accepting working conditions in complete absence of freedom from domination on the workfloor. (Pettit, 2014).

As we have seen, the republican tradition idea of non-domination is different from Berlin’s positive conception (Pettit, 2002), and although it shares some traits (Larmore, 2003), also different from negative freedom, making it an odd fit in terms of Berlin's dichotomy. More importantly, to understand freedom as ‘non-domination’ or ‘non-interference’ leads to different logical results about whether one can be considered free in a specific situation. The republican perspective is broader and more demanding; as will be illustrated in chapter 4, this is especially true when it comes to the workplace. Here, defining freedom as ‘non-interference’ would fail to capture the nature of managerial relations between actors in the workplace. Before we visit this chapter, however, first a more detailed conceptual framework of republicanism is needed.

(13)

3: Republican Freedom: a definition

This chapter provides a general definition of Republican freedom by Pettit (1997; 2002; 2014), taking into the account some of the critiques by Thompson (2013). From there on, it continues into an examination of about the limits of the concepts ‘interference’, ‘domination’, and ‘arbitrariness’.

Firstly, whether Republican freedom indeed can be equated to non-domination is not undisputed. For instance, an alternative reading by Skinner argues that republican freedom contains both non-domination and Berlin’s idea of noninterference (Pettit, 2002). That said, this thesis will use the concept of Republican freedom as expressed by Pettit (1997; 2002; 2014), while incorporating some of the objections by Thompson (2013). In this definition, we aim to stay as close as possible of the thought by traditional republican writers. Pettit defines republican freedom as a situation wherein an individual is not arbitrarily dominated. To suffer from arbitrary domination then, is to be in a position where another can interfere with the life of an individual on the basis of an interest or opinion that need not be shared by the person affected (Pettit, 1997, p. 22). Non-arbitrary interference, than, is interference that promotes collectively actionable interests of the individuals which are the object of this power. (Larmore, 2003). What is important however, is that we diverge from Pettit’s view on two points regarding the nature of domination: The first addition is that while Pettit is explicit about the fact that domination has to be the act of one agent upon another, some forms of subjection have a structural or systemic origin (Thompson, 2013). Thompson sees thissystemicdimension as prevalent in the writings of Labour republicans but not quite captured by Pettit’s definition: “Consider the office worker who

obeys the dictates of a bureaucratic mandate which in turn has negative effects on others; or consider a

woman who is forced to sell herself into prostitution by economic reasons to place herself at the will of

others’’. (Thompson, 2013, p. 283). Thompson's argument about systemic domination is twofold:

domination works itself (1) into certain structures and logics of institutions; and (2) into the consciousness of agents themselves.

Accepting the second premise in this thesis would require a broad, social-constructivist examination of how norms in the on-demand sector are dominating. While this would indeed be highly interesting on its own merits, it presumes a different way of looking at core concepts such as power and authority. The first premise - certain structures and logics of institutions can be dominating - however corresponds with

(14)

the realist idea of power and authority and therefore fits better with Pettit’s overall understanding of republican freedom. As we will see later on, the idea that certain structures and logics of institutions can be dominating corresponds largely with present-day situation of on-demand workers guided by applications and algorithms. Based on Thompson, we diverge from Pettit’s view that instances of domination are necessarily ‘known’ to the actors. As Thompson states, modern forms of authority and domination cannot be always characterized this way. As elaborated on in chapter 5 , algorithms and reputation mechanisms exercise a form of authority which often goes unnoticed.

Concluding this chapter, we accept the view that republican freedom can be defined as a situation wherein people are not arbitrarily dominated. Additionally, adding to this idea, we accept both Thompsons’ premise that certain structures and logics of institutions can be interfering and therefore dominating and the idea that instances of domination are always known by the actors involved. Arguably, including these two principles helps to understand how the disciplinary and opaque institutional context experienced by on-demand workers can be dominating.

3.1: Into interference, domination, and arbitrariness

Yet, knowing what is meant by republican freedom as a whole, does not provide enough information to look at the situation of on-demand workers. As the the horse-rider analogy shows, pulling the reins on a horse to make it go a certain way is a rather extreme and sudden form of interference. How can we define what counts as instances of interference at all? Pettit defines that interference in republican freedom can contain multiple things: Coercion of the body via for instance physical obstruction, coercion of the will by for instance punishing certain behaviour, and manipulation by for instance deception or rigging the consequence of other people’s actions (Pettit, 1997, p. 53). Naturally, the mere fact that somebody is interfered with does not mean that this person is dominated. Under some definable circumstances republican freedom is permissive of interference, and these circumstances be explained using a typology made by List & Valentini (2016). List & Valentini created a typology examining various concepts of freedom and in what way they are permissive of interference, using “moralization” and “robustness” as qualifications.

The first qualification by List and Valentini is that of robustness. Robustness has to do with whether instances of interference are actualized or not. If an conception of freedom is to be called ‘robust’, the mere power to interfere, even if not actualized, would be one wherein freedom is lacking. (List &

(15)

Valentini, 2016). Republican freedom is such a robust conception of freedom; from a republican standpoint, the capacity of a ‘master’ to arbitrarily interfere a ‘subject’ is enough to call the subject unfree, regardless of whether such capacity is eventually exercised.

The second qualification,_moralization, deals with the question whether specific constraints on freedom (such as imprisonment) are (morally) permissible and therefore do not count as freedom restricting. Most concepts of freedom contain such an exemption clause. Without it, some ideas about freedom would not be permissive of any interference on the individual, making the imprisonment of citizens or laws on taxation unthinkable. Republican freedom includes such a moral exemption clause; namely that of arbitrariness. This exemption argues that constraints on freedom can be ‘permitted’ as long as they are not arbitrarily imposed. In other words, the idea of freedom as non-domination is permissive of interference, as long as this follows from a fair system of law – a non-arbitrary regime, which promotes the interests of its subjects (Pettit, 1997). On a state level, imagine an agency that is allowed to interfere with people, but only on the condition that the interference promises to further the peoples’ collective interest (Pettit, 1997, p. 23). What follows is that interference via the rule of law, as we might find in liberal democracies, is not a straightforward offence against freedom even though it is certainly restrictive. Any system of law, even those very effective in countering domination, interferes with the individual in one way or the other. In the republican view this kind of interference does not compromise freedom, but rather conditions it:

‘’..The presence in the polity of [...] empowering and protective arrangements are like antibodies in the blood; a method of making the individual more or less immune against arbitrary domination. The presence of antibodies represents a way of realizing immunity; it is not something that causally leads to it.’’ (Pettit, 2014, p. 108)

Simply put, securing people against certain forms of intrusion via laws - that promote collectively actionable interests of the individuals which are the object of this power - can lead to net-gains in republican freedom. Thus, although being locked up in accordance with non-arbitrary law would leave the prisoner restrained in many ways, they do not fundamentally compromise freedom as non-domination as a whole. This said, imprisonment is one of the most extreme forms of interference. What is important to note here is that context also plays a large part in determining to what extent certain acts can be called interference, and whether or not this is problematic (Pettit, 1997). Both arbitrariness and the degree of interference come in all shapes and sizes. Even though it is a scale which

(16)

is not easily quantifiable, the level of arbitrariness may be more or less intense, interference can be heavy or intermediate, and the ability of a ‘dominating actor’ to interfere can be greater or lesser (Pettit, 1997, p.58). Similarly, personal areas in which a person can be dominated vary, and as such some forms of domination can be less damaging - such as the dependence of a beginner in sailing and their instructor - than other more central ones - such as the employer dominating the employee in the workplace.

(17)

4: Workplace republicanism

Naturally, the idea of republican freedom can be applied to many aspects of life where relations of authority and discipline plays a role. One of the very central areas where republican freedom may be severely challenged however, is the workplace. This chapter first explains why the workplace is a particularly hazardous realm for republican freedom. Hereafter, it aims to translate republican theory to the specific situation of the workplace. The chapter progresses by visiting the key three features of managerial power as put forward by De Stefano (2018) to look at the different forms in which managerial interference can be interfering, permitting a detailed analysis of freedom in on-demand work in the next chapter. As stated, the workplace is one of the areas particularly at-risk of arbitrary domination, which is due to the fact that labour as a commodity is different from ordinary commodities. Labour republicans around the 19th century acknowledged that ‘labour’ as a commodity is intrinsically linked to the individual workers who are the providing agent of this good. Upon selling this good the worker essentially enters a relationship in which the worker necessarily must resort to the ‘’sale of oneself’’, because the worker cannot be separated from the good itself (Gourevitch, 2014, p.113). Here lies one of the central claims of workplace republicanism: Because the commodity of labour is inseparable from the individual itself, it demands that the seller’s gives up his or her will for the duration of the working day, subjecting the individual to domination (Gourevitch, 2014, p. 113).

Although the degree to which employers can be dominating has changed for the better since the 19th century, republican thinkers argue that the current-day situation is still challenging the freedom of workers. Workplace republicans such as for example Anderson (2017), Breen (2015) and Hsieh (2005; 2008) stress the current-day pervasiveness of managers extensive power to interfere in workers’ lives. Generally, employers generally have the possibility to fire or demote workers, prescribe dress-codes, forbid hairstyles, cut their pay, assign them inconvenient hours, assign them other tasks; set them up to fail or to humiliate and harass them (Anderson, 2017, p.55). In nearly every case, managers enjoy some kind of open-ended authority; economic organizations use open-ended managerial decision-making in steering day-to-day business and unanticipated contingencies (Breen, 2015). The economic rationale is that open-ended authority eliminates the costs of constant negotiation and contracting (Anderson, 2017). What follows is that there is a limit in desirability and efficacy in eliminating such discretion in order to protect against arbitrary interference (Hsieh, 2008, p.36). Hsieh (2008) therefore sees the republican challenge as to limit arbitrary interference up to the point that some discretion can still be

(18)

uphold.

How then, can we know whether a situation is arbitrarily dominating for workers? In other words, where do ‘interference’ and ‘arbitrariness’ begin and end? Derived from the broad idea of freedom as Pettit formules it, Hsieh (2005) defines arbitrariness in the workplace as managerial interference with little or no justification in terms of the worker’s interests upon which the interference is visited (Hsieh, 2005, p. 123). Instances with interference that severely impacts the worker can be called dominating, and the lack of justification in terms of the worker’s interests makes it arbitrary (Hsieh, 2005, p. 123). Surely, employers and managers can behave in a way that is responsive to their employees. Furthermore, Hsieh sees that_{interference can be found in for example restricting the actions of workers, or decisions such} as for example closing business in a city, with the only their place to work. Adding to this definitional work by Hsieh, we include Thompson's general objections that structures and logics of institutions in the workplace can be interfering, and not always known by the actors involved (Thompson, 2013).

4.1: Three categories of control

Before we assess some features of on-demand work in order to see how they are arbitrarily dominating, a more systematic way of understanding the forms which managerial power can take is needed to understand exactly how on-demand workers are disciplined. The typology used in this thesis is one by De Stefano (2018), which contains three key features of power – or control – of employers over employees:

‘‘ ..(i) the power to assign tasks and to give unilateral orders and directives to employees; (ii) the

power to monitor both the performance of such tasks and the compliance with these orders and directives; (iii) and the power to sanction both the improper or negligent performance of the assigned tasks and any disobedience to lawfully-given orders and directives _.’’(De Stefano, 2018, p.16).

Importantly, these three functions categorize the forms of managerial authority in which arbitrary domination might show up. Revisiting Pettit, exercising these forms can result in experiencing interference by workers for instance via coercion of the body, coercion of the will or instances of manipulation. The typology by De Stefano (2018) is useful because, as examined in the next chapter, some of the managerial functions in on-demand work are partially transferred to other actors and even

(19)

structures, which makes them difficult to capture.

5: Republican freedom for on-demand workers

As said, claims of freedom and autonomy unfortunately seem to be frames more than corresponding to reality. In practice, customer approval ratings, internal ranking systems and algorithms set the conditions and context of on-demand work (De Stefano; 2015; Aloisi, 2015). While the controlling features of management may be a process embedded (partially) in technology (Schildt, 2017), it is control nonetheless.

5.1: Reputation mechanisms

A common feature in the on-demand sector is the use of feedback and reputation mechanisms . One of the benefits of a reputation system is that it reduces the need of internal performance-review personnel and mechanisms and leave a good deal of customer care to individual workers (De Stefano, 2015). Additionally, reputation and feedback systems promote trust and trustworthiness by reducing frictions caused by asymmetric information, leading to more efficient matching of supply and demand (Tadelis, 2016, p. 338). On the other hand, numerous risks arise when using peer-to-peer evaluation systems. Some of those risks are well assessed, such as the increased chance of racial discrimination (Leong, 2015), and a general lack of transparency and control in the way this data is weighed and processed (De Stefano, 2018).

The most important risk in terms of republican freedom is the fact that reputation systems can subject on-demand workers to the managerial and disciplining power of peers. In the traditional hierarchical structures of the modern firm, generally, either managers or employers retain the three forms of managerial power as put forward by De Stefano (2018) and thereby the capability to interfere with workers. Interestingly, in on-demand work it is not uncommon that managerial power is (partially) outsourced. For example, Uber drivers and riders can rate each-other according to their personal evaluations of the ride (Langley & Leyshon, 2017). Uber customers are therefore not only clients that use a platform to buy services but also ‘’..managers and HR directors, providing pervasive

micro-feedback in the form of driver ratings that far exceeds the level of supervision experienced by most

(20)

reaching for on-demand workers. In the case of Uber, customers can evaluate the trip ranging from 1 to 5 stars. If the rate falls below a certain threshold, the driver could lose access to Uber application (Aloisi, 2015, p.18).

Although the degree naturally depends on the platform, a rating-system (partially) transfers the power of monitoring performance and compliance (ii) to peers. Even though reviewing is optional (Lee et al., 2015), peers now constitute a disaggregate deciding factor in whether an on-demand worker keeps his or her job. Importantly, while the power to sanction (iii) is partially outsourced to peers it also remains vested in managers: when drivers and on-demand workers in general work take on jobs via platforms, they accept to abide by the policies and instructions unilaterally set by the company (De Stefano, 2015, p. 17). Consider the Uber example: whilede factopeers have the capability to interfere via evaluation systems and hard-coded performance thresholds, a manager can resort to deactivation of workers’ accounts, removing them from the platform even when they have formidable ratings in terms of reputation (De Stefano, 2015, p.17). For example, managers at Uber can autonomously decide to block access to the application when an Uber driver refuses to cooperate with an investigation about misconduct (Uber, 2017). Platform-workers may thus find themselves subjected to not one, but two

different masters, both in a position to severely interfere with the life of workers. Even though Uber

differs from other forms of on-demand work because the company exerts a higher degree of managerial control over the hiring and firing, direction, supervision and payment of workers than most digital platforms (Steward & Stanford, 2017, p. 423), nonetheless ‘terms and conditions’ are widespread practice in for firms in the platform economy. Arguably, to enforce such terms and conditions, some degree of managerial control (for example the power to sanction) is retained by companies in the on-demand sector, making the problem of answering of two masters not a feature specific to the Uber case only.

What does this imply for the republican freedom of workers?

Both the power to monitor (ii) and the power to sanction (iii) are sensitive areas when it comes to interference, since they touch directly upon the livelihoods and economic security of on-demand workers. As said, when firms employ reputation systems, platform-workers can be subjected to _two

masters with the capacity to interfere at will. Revisiting republican thought, domination may be arbitrary

when it is without justification in terms of the worker’s interests (Hsieh, 2005). In the case of reputation systems it is evident that such justification is lacking and that peers might even use this power more arbitrarily than managers. Remaining with the Uber example, the possibility of riders venting their

(21)

frustration against the app by giving the individual driver a poor rate are far from remote (De Stefano, 2015, p.5). For example, a study by Lee et al. (2015) shows that many Uber drivers report that the various physical and psychological states of passengers (such as being in a hurry) influenced the way they rate drivers, with the general feeling that the average of the passengers’ ratings of the drivers was not reflective of their driving performance and services. Taken altogether, the degree of republican freedom that a platform worker may enjoy while faced with a reputation system of some sort is questionable, both in terms of the severe capacity to interfere at will by two parties, and in terms of the level of arbitrariness.

5.2: Algorithmic management

Next to reputation systems, platforms may monitor performance of on-demand workers, optimizing decisions concerning their tasks and future employment. Even though on-demand workers generally face less straightforward interference than traditional workers, platform-workers face managerial4

scrutiny of algorithmic structures within the platform. This ‘algorithmic management’ sets the institutional limits wherein workers can do their job: Algorithms can organize work, for example to assign the fastest employees to work in peak times. (Schildt, 2018, p. 25). While Schildt (2018) argues that algorithmic management shifts power from a hierarchy of managers to those who for instance ‘set the parameters’ of the algorithms, this thesis argues that the algorithmic framework of a platforms itself becomes a structure which de facto replaces the manager . Although not always completely, human5

managerial scrutiny itself is replaced by hard-coded decisions about distribution of work, the processes of monitoring, and in extreme cases even what workers should be fired.

As mentioned in the previous paragraph, the managerial functions (ii) and (iii) are at least partially outsourced. What might be added is the fact that the power to assign tasks and to give unilateral orders and directives (i) is also transferred to the framework of the platform. Revisiting the claims of the advocates of the platform economy, workers should be able to autonomously decide for themselves where and when to work. However, actually being able to work , or ‘getting the gig’ depends in part on the institutional design of the application or online environment. For instance, platforms and apps may use a form of internal ranking allowing some workers to get more ‘gigs’ than others. In other words, although workers are formally under no obligation to show up for work (De Stefano, 2015), internal

4

Traditional workers experience more straightforward steering in the sense that it is clear that the manager or employer has the authority to steer or direct workers.

5

(22)

ranking algorithms may award gigs based on whether the worker shows productive behaviour or even how much time is spent on the platform (Aloisi, 2015). Consider Deliveroo, a firm which designed a platform whereon workers can accept workdelivering meals to the household or office. While productive behaviour of workers using this platform may incur monetary bonuses for them, unproductive behaviour result in lower scores on performance reports, which may result in the individual being assessed differently in terms of internal ranking and consequently having less chance of being awarded shifts on desired times (Lieman, 2018, p. 105). A similar but more extreme example of this can be found in the way the platform designed by Deliveroo exerts its capability to sanction. Deliveroo employs an algorithmic structures which make underperforming employees or contractors redundant automatically - without human involvement (Schildt, 2017, p.25).

Similarly, algorithmic management has a profound effect on how much agency is left for the on-demand worker in the framework of the platform. In a study by Lee et al. (2015) the impact of algorithmic decision on Uber and Lyft drivers is researched. One driver, who worked as both a taxi and rideshare driver, did not like the ridesharing assignment systems because algorithms made decisions that he used to make himself, making him feel he lost agency regarding strategies that he had developed to maximize his income (Lee et al., 2015, p.1610).

Effectively, the managerial power to monitor performance and compliance (ii) and the power to sanction (iii) are partially transferred to the realm of algorithmic management. Consequently, transferring managerial features to algorithmic management completely creates organizational environments where workers are known and treated solely based on quantified data instead of as cultural beings, which gives rise to all kinds of ethical complexities (Gal, Jensen & Stein, 2017, p. 7)

What does this imply for the republican freedom of workers?

Again, the degree to which the loss of human agency on the job is erosive of freedom is very much a matter of context. In the case of Uber, having only 15 seconds to accept a trip request is negligible interference compared with the fact that ‘fares’ fluctuate according to a dynamic pricing algorithm which completely stops them from setting their own fare (Lee et al., 2015, p.1610). Revisiting Hsieh, instances of interference that severely impacts workers can be called dominating, and the lack of justification in terms of the worker’s interests makes it arbitrary. That said, the mere fact that platforms or applications ‘give’ unilateral orders and directives (i), monitor (ii) and sanction (iii) workers in a way that - as seen above - can be called both severe and intransparent in nature subjects on-demand

(23)

workers to the arbitrary domination of algorithmic management. As visited, one of most celebrated successes of the platform economy is the the ability to work ‘wherever and whenever’. However, of course depending on the platform, where and when on-demand workers can actually do their job in term may depend on internal ranking based on ‘productive behaviour’ or the ‘time spent online’ is essential in securing a living as a on-demand worker (Aloisi, 2015). As said, _{definite conclusions about} the general state of freedom in the on-demand sector is beyond the scope of this paper. However, even though the degree to which workers are unfree is highly contextual and dependent on the platform, the existence of extremities wherein underperforming workers lose access to a platform in an automated way strongly emphasises the need for further research and a distinct republican answer to algorithmic management in general.

(24)

6: Limiting interfering capabilities

‘’Given the potential growth in this sector, the promotion of an enabling environment for worker

organizing and collective bargaining can help ensure that the use of digital platforms is not at

the expense of good jobs and decent working conditions’’ (Johnston & Land-Kazlauskas, 2018,

p.4)

Now that we have established that on-demand workers are subjected to far reaching forms of control, the next the next two chapters will explore what a republican answer should be to for instance algorithmic management. These next two chapters will explore some of the possible pathways in which republican freedom for on-demand workers can be increased. Although the amount of individuals who see themselves working under features such as algorithmic management and reputation systems is limited as of yet, the historic growth of people employed in the on-demand sector (Huws et al., 2017) makes exploring republican pathways to freedom for on-demand workers particularly urgent. Generally, ways of increasing freedom (for on-demand workers) can be categorized by three distinct approaches. First of all, as is the subject of this chapter, the degree to which interference is possible can be limited. Secondly, as will be addressed in chapter 7, freedom can be increased through limiting arbitrariness. Limiting arbitrariness can be done via regulation reducing the capabilities of managers, or, more structurally by for instance implementing workplace democracy. Thirdly, as will be seen in chapter 8, limiting the dependence on managerial authority is another possible course of action.

6.1: Reputation systems

When it comes to reputation in the on-demand sector, marginal gains in republican freedom can be made by enforcing data portability. Data portability, or allowing the portability of workers’ existing ratings from one platform to another would reduce the dependency of workers upon one single platform (De Stefano, 2014, p. 23). However, as previously addressed, even if such portability is achieved, on-demand workers will still be quite problematically subject to two masters in the republican sense. For on-demand workers experiencing managerial scrutiny, outsourced or direct, it seems that bringing back the amount of actors with interfering capability from two to one actor is highly preferential. This is not to say that reputation systems in itself necessarily a feature to be avoided; surely it is possible to work as a self-employed individual in the platform economy where reputational scoring

(25)

do not result in being punished, reprimanded or interfered with in any way. Arguably, the fact that reputational scores are made a ‘weighting factor’ in decisions regarding the worker transfers some interfering capabilities to peers. Without this connection, reputation systems would only serve as monitoring device, which is less intrusive. While it is beyond the scope of this paper, a broader examination of reputation systems can clarify the advantages and disadvantages for workers in terms of freedom.

6.2: Regulating internal ranking systems & algorithmic decision making

Arguably, present-day legislation limits the interfering power of ‘traditional’ managers far more than it does the systemic interfering powers of online environments or applications. Similar to the way that Pettit argues that empowering and protective arrangements are a method of making the individual more or less immune against arbitrary domination on the state-level, laws about corporate governance and workplace safety can limit the authoritative power on the on-demand worker, and in doing so set the stage on which negotiation takes place (Anderson, 2017, p. 53). One measure specific for on-demand workers is making sure that any managerial decision suggested by artificial intelligence must be subject to review by human beings who remain legally accountable, together with their organisation, for the decision and its outcomes (De Stefano, 2018, p.28). Examples of such decisions are for instance deactivation of profiles, changes of terms and conditions of use or the payment of workers. Some form of transparency in exactly how firms in the platform economy employ algorithmic management may lead to a slight improvement. Additionally, understanding the contours of algorithmic discipline from the outset helps to workers to negotiate economic relations . That said, transparency might however not be 6

the most feasible pathway to increased freedom, since the platform economy is heavily dependent on competitive algorithms and will likely be unwilling to share features of their core business (Lee et al. 2015). Similar to the challenge in judging the merits of reputation systems in terms of freedom, the exact merits of algorithmic management are unclear and in need of further examination.

6

(26)

6.3: Collective bargaining & regulation

Another instrument to alleviate the levels of control experienced by on-demand workers is unionization. Labour unions can bargain collectively and can use membership dues to pay their members to withhold their labour in an attempt to negotiate less dominating terms , which is often impossible on the7 individual level (Gourevitch, 2014, p. 107). In these negotiations limiting the degree of interference is often a subject of interest. Realizing unionized workers in the on-demand sector however is harder than traditional forms of work because of three reasons. Firstly, it is structurally hard to create solidarity among “scattered” workers (Aloisi, 2015); social relations between workers in the on-demand sector are arguably more fragmented than traditional work. For example, according to the account of Lieman (2017) Deliveroo-riders interact only sporadically at predetermined ‘waiting locations’ where workers gather. The fact that large parts of such work is being carried out individually leads to fragmented social relations. Secondly, although the great accessibility of platforms and apps can create large pools of people available (McKinsey, 2015), on-demand workers are more inclined to seek non-permanent forms of employment than worker-pools in traditional work (Johnston & Land-Kazlauskas, 2018) which makes organizing on-demand workers difficult_{. Thirdly, in some jurisdictions the status of ‘independent} contractor’ prevents workers from forming unions and engaging in collective bargaining altogether (Johnston & Land-Kazlauskas, 2018).

In comparison with traditional work, _{the potential base for labour unions thus consists of more} scattered workers who are more inclined to seek temporal, non-permanent forms of employment. Logically, this inhibits collective organizing efforts, as workers can be hard to find, hard to reach, and difficult to engage (Johnston & Land-Kazlauskas, 2018). This in turn makes collective bargaining even more difficult than in traditional forms of work.

7

Although collective bargaining does not fall neatly in the category of instruments which can be employed to limit interfering, and are in fact a broader instrument, whether bargaining is used to limit employer interference or ensuring that they act with the interest of their workers at heart is a gray area.

(27)

6.4: On-demand workers: Classification matters

‘’The growth of the gig economy, typified by online platforms and isolated independent workers,

poses fundamental challenges to traditional models for regulating work and setting minimum

standards. It is not clear that existing regulations apply to gig workers, let alone that they can be

effectively enforced in the digital economy.’’ (Steward & Stanford, 2017, p. 421)

As De Stefano (2015) puts forward, the terms and conditions set by platforms and apps often explicitly specify that the relationship between worker and the business running the platform or the app will be one of self-employment (De Stefano, 2015, p.12). On-demand work thus shares many features also present in the wider labour market - a trend of casualisation (flexibilization), income instability and (in some instances) and renouncing what is in substance an employment arrangement (Steward & Stanford, 2017, p. 422). Logically, whether a on-demand worker is granted the status of employee or not makes a difference: some benefits might not be available to those working as self-employed providers including paid holiday, paid sick leave, employer pension contributions, maternity and paternity leave etc. (Goudin, 2016)

Most importantly, classifying on-demand workers as self-employed also has two clear consequences in terms of workers’ freedom. Firstly, by classifying their workforce as self-employed individuals, firms circumvent general employment regulation which protects fundamental principles and rights at work (Drahokoupil, 2016; Aloisi, 2015), the existence of which provide the ‘’antibodies’’ (Pettit, 2014) to render the worker more or less immune from arbitrary domination. Secondly, as mentioned, the exclusion of independent contractors from collective bargaining rights in some jurisdictions is an obstacle for on-demand workers to play an active role in defining the scope and conditions of labor performed within this emerging new economy. Depending on the country, these two consequences are true for non-standard forms of work in many sectors other sectors too. Providing adequate classifications to workers in the platform economy should however also be a priority for policymakers: As long as the unique structure of platforms and apps facilitates circumventing employment regulation, this is likely to set a precedent for non-digital firms to follow. This is not a mere hypothetical scenario. As said, authors like Steward & Stanford (2017) see that evading traditional regulations appears to be a key rationale for establishing digital businesses in the first place (Steward & Stanford, 2017, p. 421).

(28)

7: Limiting arbitrariness

Another pathway to increase freedom includes limiting arbitrariness. One of the ways to limit the degree to which employers and managers can make arbitrary decisions at all is limiting the legal space to do so. Workplace constitutionalism is such a limiting instrument: it aims at raising the minimum floor through regulation to ensure employees act with the interest of their workers at heart (Anderson, 2017; González-Ricoy 2014). Workplace constitutionalism however has its limits. Raising the minimum floor can advance the degree of freedom for workers, yet cannot be a catch-all solution because of how the economic process is organized (González-Ricoy 2014). Revisiting the earlier objections by Hsieh about domination, formalizing a harsh regulatory regime to combat arbitrariness can only do so much. Arguably, in a practical sense it is impossible to draft all-encompassing regulation so that managerial discretion can never be arbitrary. What is more effective, according to Hsieh (2005), is enabling workers to protect themselves against arbitrary interference by putting them in a position to contest managerial decisions. One way to do that is workplace democratization. Using the definition by González-Ricoy (2014), workplace democracy can be described as a form of managerial organization in which workers have control rights over the management of the firm. Based on Dahl (1985), Breen argues that people should have the right to participate in collective governance of their firm similar to the right for people to participate in the political system which they are subject of.

‘’Just like states, economic enterprises are cooperative associations governed by general rules

applying to all members and endorsed in terms of their mutual advantage. And just like citizens,

because workers are members of an economic enterprise, have the ability to judge the appropriateness of its general rules and, no less decisively, are subject to the coercive

enforcement of these binding rules, they, too, ought to have the right to participate in its

collective governance.’’ (Breen, 2015, p. 472).

While ambitious, workplace democracy is preferable to workplace constitutionalism when it comes to republican freedom. This is because workplace democracy reduces the arbitrary exercise of such discretion by directly incorporating the interests of workers (the object of this discretion) into their decision-making process (González-Ricoy 2014, p. 247). By doing so, it avoids earlier objections made

(29)

about catch-all regulatory regimes. González-Ricoy acknowledges however that even workplace democracy would be that such a system (as is the case with review systems) transfers managerial capabilities to peers. These peers, in turn, may use their voting rights to dismiss a minority . 8

Nevertheless, workplace democratization is a pathway for traditional and on-demand workers alike. Moreover, there are two reasons why the need for democratization is even bigger in the on-demand sector. Firstly, it is hard to say whether a specific form of algorithmic management is heeding to the actual needs of workers in a firm because of its intranspancy. Arguably, workplace democracy has the potential to partially solve this problem because the underlying features of decision-making must be understandable in order for workers to decide upon its merits. Secondly, traditional managers may provide (informal) windows for workers to have a say in decision making, whereas on-demand workers do not enjoy this feature. As seen in the example of driver ‘fares’, the structure of an application or online environment narrowly dictates what price will be asked at the consumer-side. While a worker may disagree, there is no way to actually object this. Not only does this constitutes high levels of interference in a very important domain, it is hard-coded within the framework of the platform and thus immune from questioning.

8

González-Ricoy states hereon that a solution can be found in multilevel workplace democracy and enforcing workplace constitutionalism. In our view, combining workplace democracy with workplace constitutionalism, regulation and unionism would suffice in making workers more or less immune from arbitrary domination, similar to Pettits’ view on protective arrangements on a state level. Furthermore, in our view multilevel workplace democracy would be unnecessarily complicating and consequently lead to severe forms of voter fatigue, eroding the instrument of workplace democracy itself.

(30)

8: Republican self-employment and worker dependence

Limiting worker dependence can be as simple as limiting the duration of the working week. In fact, one of the immediate aims of European workers’ movements in the mid-nineteenth century was to simply shorten the length of the working day (Anderson, 2017, p. 60). As seen in the objections by Hsieh (2005) and González-Ricoy (2014), limiting overall dependence might be a more wholesome solution to ensure republican freedom for workers in general. Using the words of González-Ricoy:

‘’Given that democracy (in the workplace and elsewhere) cannot fully eliminate domination, the

best means to ensure freedom from arbitrary interference would consist of uprooting the very

source of domination in the workplace—namely, the existence of managerial authority.’’

(González-Ricoy, 2014, p. 251).

Although the list in this chapter is not exhaustive, there are many ways to limit worker dependence at the root . In this chapter, we first examine the idea of self-employment as an instrument to escape the9 realm of authority, which is a typical classic republican ideal. Secondly, this ideal-type is compared with some of the realities of those self-employed in the on-demand sector, to show that self-employment in the platform economy is very different from the republican ideal. Hereafter, we briefly examine a general interplay between republican freedom and workers and the way a ’basic income’ can be a liberating institution.

9

This chapter does not include exit rights as a serious method of limiting dependence based on the arguments put forward by González-Ricoy (2014). Firstly, in imperfect labour markets it is a fact that workers are not in any way guaranteed to be hired back, making quitting costly for workers as well. Secondly, there are all types of non-financial exit costs, ranging from

firm-specific capital to social investedness. Finally, a feature also addressed by Anderson (2017), alternative employers may be evenly despotic, unregulated, and arbitrary as their current job.

Republican freedom and on-demand work