Post-Diversity,Precarious Work for All: Unmaking borders to govern labour in the Amazon warehouse

Abstract

This paper investigates the (un)making of borders as a form of labour governmentality in one of Amazon’s warehouses in Poland. Guided by a critical theory of borders as a form of labour governmentality under global capitalism, we identify organizational practices through which socio-demographic categories traditionally deployed as principles of organizing work (e.g., gender, age, ability) are unmade: the management of deskilled labour through an algorithmic system, the non-selective hiring of workers, the enforcement of social norms of interpersonal respect and a universal system of casualized employment. Together, these practices constitute workers as undifferentiated, interchangeable and equal labour, let them compete with each other under harshly exploitative conditions, and continuously dispose of the least productive among them, keeping all in structural uncertainty. The study contributes to the critical diversity literature by showing a ‘post-diversity’ governmentality that rests on equality, competition and precarization of labour as a whole, rather than segregation and marginalization through an ‘ideal worker’ norm. This labour governmentality operates by eliciting consent from historically subordinated workers and eliminating the advantage of historically relatively privileged ones. Unmaking borders within labour inside the organization, this governmentality at the same time crucially rests on borders outside it.

This paper investigates the possibility that, in advanced capitalism, socio-demographic categories – e.g. gender, ‘race’, ethnicity, age, sexual orientation, disability, religion and language – might no longer be deployed as principles of organizing work and the workforce unequally. This possibility has to date been rarely considered, as critical diversity studies have mainly focused on theorizing and documenting the essential and persistent role of these categories in the reproduction of power inequality (Acker, 1990, 2006; Ashcraft, 2013; Nkomo, 1992; Zanoni, Janssens, Benschop, & Nkomo, 2010). To debunk the myth of organizations as neutral, meritocratic spaces providing equal opportunities to all, they show how some groups are structurally excluded from the labour market or included only in the least recognized and rewarded, most precarious and exploitative jobs (e.g. Reskin & Hartmann, 1986; Tilly, 1998; Tomaskovic-Devey & Avent-Holt, 2019; Zanoni, 2019).

Yet this scholarship overlooks that organizing work along socio-demographic categories is not devoid of risk for employers, as it does not only facilitate the exploitation of labour, it might at the same time hamper it. While segregating workers makes some more vulnerable and exploitable, it also ensures that other ones belonging to historically dominant groups retain higher rewards and protection. Segregation shelters these latter workers by eliminating competition from workers belonging to historically subordinated groups. Sometimes this occurs through legislation excluding certain categories of workers from certain jobs (Hasday, 2009; Hill, 1996), but more often it results from organizational practices that fill jobs with specific socio-demographic categories of workers, assumed to have the right profile to adequately carry them out (Acker, 1990, 2006; Ashcraft, 2013; Cockburn, 1983). Whatever the form, to the extent that socio-demographic categories structurally and durably organize work and the workforce, direct competition between historically subordinated and relatively more privileged groups of workers is excluded, removing an essential mechanism for employers to extract maximal economic value from the workforce in its totality (Heinrich, 2012; Marx, 1976[1867]). Critical diversity studies neglect the contradictory effects of using socio-demographic categories to organize work due to their focus on the vulnerability and experiences of historically subordinated workers, such as women, migrants and racialized individuals, as opposed to (diverse) labour as a more comprehensive category from which capital extracts value to accumulate (Romani, Zanoni, & Holck, 2021).

To unravel this contradiction, we draw on a critical theory of ‘bordering’ (Lazzarato, 2009; Mezzadra & Neilson, 2013; Neilson, 2012). This theoretical approach conceptualizes practices of (un)making borders as a form of labour governmentality under global capitalism. It posits that, while employers fragment labour along socio-demographic lines, skills and employment status, they do so in ways that are dynamic, fragile and uncertain. Through this ‘multiplication of labour’, they constantly remind historically relatively advantaged workers that their own ‘privilege’ should not be taken for granted. ‘Different’, more easily exploitable workers might, in the near future, take their work. Accordingly, we ask: How does global capital (un)make socio-demographic borders to govern labour? To address this question, we draw on a qualitative in-depth study of POZ1, an Amazon warehouse located in Poland, near the border with Germany and mainly serving the German market. This organization is a particularly relevant case as Amazon is known to enforce maximal competition between highly diverse workers and is today considered as epitomizing the future of work under global capitalism (Cattero & D’Onofrio, 2018; Delfanti, 2021).

Informed by a critical theory of borders, this study makes multiple contributions to the critical diversity literature. First, our analysis identifies and unravels organizational practices constituting a ‘post-diversity’ governmentality of labour, that is, a governmentality that unmakes socio-demographic borders that commonly fragment the workforce. We show how the management of deskilled labour through an algorithmic system, the non-selective hiring of workers, the enforcement of social norms of interpersonal respect, and a universal system of casualized employment jointly constitute workers as undifferentiated, interchangeable and thus disposable. By demising socio-demographic categories as principles of organizing, these practices enforce competition, uncertainty and precarity onto the whole workforce, structurally turning all into a reserve army of labour. These practices highlight the key, yet largely neglected, role of competition in processes of precarization and exploitation of diverse labour.

Second, our analysis shows how ‘post-diversity’ labour governmentality fundamentally differs from labour governmentality based on the segregation and exclusion of certain socio-demographic categories of workers based on their difference from an ‘ideal worker’ norm. It namely mobilizes the principle of equality to integrate all workers into the process of capital accumulation as ‘same’ labour, i.e. under the same conditions. Capital deploys this equality to govern and exploit workers in two distinct ways. Equality governs historically subordinated subjects by eliciting their consent, as they experience it as countering their structural disadvantage in the labour market and society (cf. Fraser, 2013; Peticca-Harris, deGama, & Ravishankar, 2020). Equality governs historically advantaged workers rather by disavowing their advantage – their skills, competences and higher social status – to let them compete with all workers, precarizing them to contain their demands.

Third, our analysis points to the need to conceptualize the ‘post-diversity’ governmentality of diverse labour beyond the organization. The unbordering of labour inside POZ1 is predicated on the simultaneous drawing and reaffirming of other borders multiplying labour at various scales. The bordering of labour along national borders within the European Union (EU), between Amazon labour and labour employed by other companies in the region, and the algorithmic bordering between data points capturing workers as more or less performing were essential to the unbordering of POZ1 labour to enforce maximal competition, uncertainty and precarization. Critical border theory thus offers critical diversity studies a powerful conceptual vocabulary to reconstruct how borders proliferate at multiple scales, dynamically organizing diverse labour for capital accumulation (Lazzarato, 2009; Mezzadra & Neilson, 2013; see also Neilson, 2012; Zanoni & Pitts, 2022).

Socio-Demographic Categories, Segregation and Inequality in Organizations

Since the 1990s, critical diversity studies have investigated how socio-demographic identities operate as principles of organizing work and workers in ways that structurally reproduce segregation and inequality. Central to this literature is the idea that work and socio-demographic identities are mutually constituted (Acker, 1990, 2006; Ashcraft, 2013; Cockburn, 1983; Romani et al., 2021; Zanoni et al., 2010). On the one hand, jobs, organizations and whole sectors are defined in ways that reflect the socio-demographic profile of the workers who historically filled them, who are considered to possess the required competences and characteristics (MacKenzie & Forde, 2009; Zanoni, 2011). On the other hand, socio-demographic categories such as man/woman, white/black, young/old do not refer to given, universal ‘essences’ of certain groups. Rather, their meaning is continuously socially produced by organizations to reflect the work associated with them in the specific organizational context (Zanoni & Janssens, 2004). In other words, jobs come to be defined in ways that are imbued with characteristics of the ‘ideal worker’ for which they were originally designed, and those workers that embody that ideal gain access to those jobs most easily precisely because they are seen suitable to carrying them out (Benschop & Dorewaard, 1998). Workers with other socio-demographic profiles conversely tends to be excluded or marginalized, reproducing segregation in inconspicuous ways (e.g. Acker, 1990, 2006; Ashcraft, 2013).

The mutual constitution of socio-demographic categories and work naturalizes segregation along the lines of gender, ‘race’, ethnicity, migration status, language, etc., legitimating and institutionalizing it (Tilly, 1998). Socio-demographic categories thus come to function as key ‘principles of organizing’ not only jobs, occupations and specializations, but also organizations and entire sectors of the economy (Acker, 1990; Jammaers & Zanoni, 2021; Nkomo, 1992). For instance, organizational norms of long hours and high flexibility tend to exclude workers who, next to their paid work, also carry out care tasks; requirements of religious ‘neutrality’ exclude workers who wear religious signs; and specific attire requirements might exclude workers on a number of characteristics including complexion, weight and physical ability, to name only a few.

Crucially for the reproduction of inequality, the mutually constitutive relation between work and socio-demographic categories of workers not only segregates but also hierarchically orders both (Acker, 2006; Holck, 2018). Categories of difference are never neutral, they are relationally established and evaluative (Tomaskovic-Devey & Avent-Holt, 2019). They reproduce the hegemony and dominance of certain groups of workers over others by legitimizing the unequal distribution of symbolic and material rewards, including status and prestige, career opportunities, wage and benefits, social protection, employment security and conditions of work (Acker, 1990; Tilly, 1998). Workers from historically subordinated groups, such as women and migrants, come to be overrepresented in more precarious jobs (Alberti, Bessa, Hardy, Trappmann, & Umney, 2018; Kalleberg & Vallas, 2018; Vallas, 2015; Vosko, 2010), or jobs that are uncertain, unstable and insecure in terms of the statutory conditions (e.g. casual contracts, temporary work, part-time work, (bogus) self-employment), entailing harsh and even hazardous working conditions (e.g. degree of autonomy, pay and benefits, health-related risks, career opportunities) and less institutionally protected (e.g. social rights and statutory protection) (Kalleberg, 2018).

There is ample consensus among critical diversity scholars that socio-demographic categories do not operate as principles of organizing in ways that are fixed once and for all (Ashcraft, 2013; Cockburn, 1983; MacKenzie & Forde, 2009). Employers might reorganize work and employment conditions and tap into new socio-demographic groups. This might be triggered by shortages of labour, the introduction of technology and/or competitive pressures to reduce costs (e.g. MacKenzie & Forde, 2009; Theunissen, Zanoni, & Van Laer, 2022). Historically subordinated groups of workers might also contest their own exclusion from valued and protected work and obtain access. However, typically, when a sizeable share of a job becomes filled with workers belonging to these groups, it tends to become more precarious, less rewarded and protected and to lose status (England, Allison, & Wu, 2007; Murphy & Oesch, 2016). Conversely, when jobs are upskilled and become better rewarded, incumbents from historically subordinated groups tend to be replaced with workers belonging to historically dominant ones (e.g. Cockburn, 1983).

Despite changes over time, socio-demographic categories are understood as playing a lasting role in the organization and valuing of work, reproducing inequality within the workforce in renewed ways. The possibility that employers might not organize through socio-demographic categories remains elusive, as it contradicts the core ambition of critical diversity studies to unveil how inequalities between socio-demographic groups of workers unfold. At best, this possibility features in discussions of colour-blind organizational policies of non-discrimination (Konrad & Linnehan, 1995; Ortlieb & Sieben, 2013), which however are generally critiqued for being ineffective in fostering equality. To envision this possibility, socio-demographic categories need to be conceptualized as means to accumulate capital and therefore contingent on their ability to fulfil this function, rather than long-standing principles of organizing unequally. This requires a theoretical perspective that accounts more for their economic function in capitalist processes.

Studying Diversity in Global Capitalism Through a Critical Theory of Borders

To conceptualize the possibility that socio-demographic categories might be not only made but also unmade as principles of organizing work and workers, we rely on a critical theory of borders. Variously declined as ‘borders’, ‘borderscapes’, ‘border work’ and ‘bordering’, the notion of border is today present across the social sciences, relationally defining ‘us’ and ‘them’, dividing and connecting across space and time through contested practices of symbolic, cultural, social, political and economic bordering, ordering and othering (van Houtum & van Naerssen, 2002).

The works of Mezzadra and Neilson (2013) and Lazzarato (2009) are particularly productive for our purpose. They conceptualize bordering as a form of governmentality involving the ‘multiplication of labour’ under contemporary global capitalism, or the continuous dynamic and contested fragmentation of labour on multiple scales. Importantly, in this perspective, the making and unmaking of borders does not solely refer to national borders and migration, but points more broadly to the dynamic production of difference – be it gender, ‘race’, ethnicity, disability, sexual orientation or their various intersections – and its complex relation to inequality and precarity (see Yuval-Davis, Wemyss, & Cassidy, 2019). The idea of a multiplication of labour has its roots in class composition, a notion advanced by Italian operaismo in the 1970s to explain the changes in the structure and political consciousness of the working class caused by the decline of Fordism. Countering prior understandings of the working class as homogeneous, class composition points to the fragmentation of labour in its relation to capital. Such fragmentation includes the hierarchical division of labour in capitalist workplaces, segregation along socio-demographic lines and different degrees of precarity, and labour’s heterogeneous political consciousness rooted in workers’ partially distinct material conditions (Mezzadra & Neilson, 2013).

Theorizing labour as multiplied provides a novel, fruitful episteme for conceptualizing the relation between diversity – the differences demarcating labour as heterogeneous – and (in)equality. In the contemporary phase of capitalism, characterized by shifting regimes of exploitation, dispossession and domination through capital’s continuous geographical expansion and rescaling (Harvey, 2006), the contradiction between the necessity of abstracting labour to enforce social cooperation and its continuous internal diversification becomes particularly apparent (Mezzadra & Neilson, 2013).

On the one hand, capitalism requires workers to cooperate with each other as labour in ever-expanding production and circulation processes. It therefore socializes and de-individualizes them to transcend the ‘limitations’ of the specific, local context in which they are embedded, a process that is today magnified by the global expansion of the frontier of capital (Adler, 2007; Marx, 1973[1861], 1974[1844]; Harvey, 2006). In this way, concrete workers become the working class and their concrete labour becomes just part of a growing totality of abstract labour, commensurable with any other part:

the apparently simple category of ‘labor’ — just work, not weaving, tailoring, welding, teaching, cooking, driving, typing — is the product of a society that itself is able to see these indifferently, as versions of the same kind of activity, reducible to a multiplier — so many dollars and cents per hour. For not only does labor become a new kind of abstraction, the product of the new economic relations of capitalism, but so does the category of the worker or laborer to name the masses of wage earners in these new relations. (Denning, 2007, p. 128).

On the other hand, categories are deployed by capital to continuously constitute and reconstitute concrete workers and their labour not only as ‘similar’ and ‘commensurable’, but also as ‘different’, ‘heterogeneous’ and ‘incommensurable’. The multiplication of labour points to how borders are continuously instituted, maintained and dismantled through social practices. Explaining capitalist firms’ use of borders to extract value, Lazzarato writes:

Employers [. . .] think and act in accordance with a [. . .] logic [. . .] of locating, constructing and consolidating a multiplicity of ‘normalities’. The goal of the management of these ‘normalities’ does not seek conformity to one model, but to maintain them in a state of ‘equal inequality’, of competition, to encourage differentials and perpetuate a ‘mobilizing’ uncertainty. Inclusion and exclusion, the normal and the abnormal, do not determine a ‘great division’; they are instead variables of governmental action that tends, anyway, to multiply cases, situations or statuses. Government acts less through a divide than through the modulation of divisions and of differences. (Lazzarato, 2009, p. 119)

Clearly, borders can thus not be reduced to walls that exclude some workers. While they sometimes do, in many cases they include them by articulating, multiplying, hierarchizing and stratifying: ‘borders are equally devices of inclusion that select and filter people and different forms of circulation’ (Mezzadra & Neilson, 2013, p. 7). Bordering has historically performed workers’ ‘differential inclusion’ (Wallerstein, 1991), or their unequal incorporation into global capitalism (see Chakrabarty, 1992; Hall (1986) on race and ethnicity and Mies (1998) on gender).

Importantly for our analysis, borders are in this perspective not ‘given’ and unavoidable. They are made and unmade with the purpose of keeping workers in uncertainty, to enforce competition among them. Informed by this body critical theory, we now move to empirically examine the organizational practices of (un)making socio-demographic borders through which one of Amazon’s warehouses in Poland governs labour.

Methodology

Our case is POZ1, Amazon’s warehouse in Poznań. Amazon is one of the biggest on-line retailers, with a network of 320 warehouses operating across the globe, employing about 250,000 grey-collar workers and 25,000 auxiliary personnel (Amazon, 2019), and considered to epitomize contemporary global capital (Cattero & D’Onofrio, 2018; Delfanti, 2021). The main data source are 78 in-depth interviews conducted by the second author in 2018 with workers of POZ1. As Amazon does not allow researchers to access the company, respondents were recruited individually through social media. Most came from the Poznań urban area (about 40%), the rest were based in the surrounding regions. In the warehouse, they were involved in a variety of tasks such as stowing, picking and packing, and included employees recently hired on short-term contracts as well as more experienced workforce on an open-term contract. Their tenure at Amazon ranged between 3 years and 11 months and 3 weeks, averaging 7 months. The majority of interviewees (45) were female, about 30% were under 25 years, 50% were aged 25–45, and the remaining 20% were above 45 years.

The interviews took place in the respondents’ homes or in public spaces (e.g. coffee shops, parks), and lasted between 45 and 180 minutes. They were divided into two parts. The first part was unstructured. Interviewees were asked to introduce themselves and talk about their work, narrating their personal and professional trajectory in order to contextualize their narratives. In the second part, open questions were asked based on an interview scenario crafted around themes relevant to respondents’ activity and experience, professional trajectory, financial situation, the organization of work and the role of technology in it, relations with colleagues and shop floor managers, and the terms of employment. Diversity was not an explicit topic of the questionnaire, but socio-demographic categories rather emerged organically from the conversations. Each interview was fully recorded and transcribed without any modification for analysis. Common ethical guidelines including obtaining respondents’ informed consent and anonymization of data were followed to preserve confidentiality.

To mitigate the limitations of reconstructing organizational practices mainly from data collected with workers, we complemented them with three interviews with trade union militants and observed three workers’ fora on Facebook from 2018 to 2022: AMAZON PL (ca. 17,000 members), Amazon Fulfilment Center POZ1 (ca. 7,000 members) and Amazon Fulfilment Poznań POZ1 (ca. 8,000 members). The observation of the fora was announced to members, following common ethical guidelines in netnographic research (Jemielniak, 2020; Whiting & Pritchard, 2017). We moreover followed the (inter)national press and conducted desk research (Amazon Workers and Supporters, 2018).

Data analysis

The interview material was initially read and discussed between the authors in multiple meetings to create a shared sense of the case. In these initial conversations, all other data sources were also used to reconstruct the work environment including the origins of the warehouse, the type of labour contracts used by the company, the (estimated) demography of the workforce, the spatial organization of the warehouse, the shift system, the work process and the algorithmic management of all operations. We then systematically coded the interview material using the qualitative data analysis software MaxQDA. We first identified all fragments in which respondents referred to socio-demographic categories. Of the 340 fragments, 73 were on gender, 61 on educational level, 44 on poverty, 41 on age, 38 on urban-rural background, 28 on sexual orientation, 22 on disability, 19 on ethnicity/migration background, and 14 on conviction. Jointly examining these fragments, we realized early on that, to our surprise, many respondents narrated Amazon not only as an extremely exploitative and precarizing work environment, but also an exceptionally diverse, desegregated workplace where all workers were treated equally.

In a second step, to answer our research question informed by a critical theory of borders (How does global capital (un)make socio-demographic borders to govern labour?), we systematically looked for descriptions of Amazon making or removing socio-demographic categories in shaping its relations to workers, the labour process and the relations between workers. Through multiple iterations, we reconstructed four organizational practices through which the company treated workers equally irrespective of their socio-demographic profile: the algorithmic management of deskilled labour, non-selective hiring, enforced social norms of interpersonal respect, and universal casualized employment. Although these practices were identified inductively, they reflect HRM practices – hiring, job allocation, compensation and diversity policies – which are considered key in the organizational literature in the reproduction of inequality (Amis, Mair, & Munir, 2020; Janssens & Zanoni, 2014). In this phase, we also noticed that workers often stressed how these practices differed from what they had experienced in other Polish workplaces.

In a third step, we attempted to make sense of the observed contradictory effects of the organizational practices of unmaking of borders. The far-reaching equalization of workers’ treatment micro-emancipated some from oppressive relations and at the same time enabled the employer to enforce competition among all workers leading to highly exploitative and precarious working and employment conditions. Then, turning to the multiscalar dimension of borders, we further interpreted Amazon’s unmaking of organizational borders in the light of its use of the proximate national border. This latter sanctioned deregulated Polish labour legislation, integrating POZ1 labour into global capitalism at unequal, more precarious conditions, while allowing the free flow of commodities under a unified European market.

In each step of the analysis, we triangulated the emergent results with the other data sources. Triangulation with the social media fora data and the website information did not generate contradicting insights, as workers used the Facebook largely to exchange practical information. This however allowed fine-tuning the results. For instance, we had not initially spotted the urban/rural and poverty categorizations, which are not commonly used in critical diversity studies, yet which are meaningful socio-demographic categories in the Polish labour market.

POZ1 in the Polish and European Context

POZ1 was established in 2014 in response to union activity in Germany. Amazon’s sustained growth is based not only on highly effective online marketing, a ‘prime membership’ offer allowing clients unlimited orders with free postage and quick delivery for a flat fee, and a ‘digital Taylorist-Fordist’ organization of work, but also, crucially, on the availability of precarized labour (e.g. Boewe & Schulten, 2019; Briken & Taylor, 2018; Cattero & D’Onofrio, 2018; Delfanti, 2021). Indeed, POZ1 is located to take advantage of the integrated EU market and a relatively inexpensive workforce, capital-friendly labour regulations, cheap land and tax bonuses offered by the Polish state. The labour market in post-socialist Poland has been characterized by precarious, low-skilled and low-paid work (Bohle & Greskovits, 2007) and continued work migration, especially of youth, mainly to other European countries (White, 2016). Structurally high levels of unemployment have weakened workers’ bargaining power, rendering precarious forms of employment endemic (Boulhol, 2014).

For instance, temporary employment is, at 27%, among the highest in the OECD countries (Boulhol, 2014). Informal labour market practices are diffuse and reflect a socialist past characterized by low trust to officials and institutions and resourcefulness in overcoming bureaucratic obstacles and making ends meet (White, 2016). Recent figures of the National Labour Inspectorate (2021) indicate that 68% of the employers controlled in 2020 were in breach of labour law, including failing to comply with health and safety measures (48.9%) and work time regulations (6.2%) and not paying workers on time (19.2%). The Polish labour market also remains deeply segmented by gender and age (Lewandowski & Magda, 2018). Only 59% of women and 25% of individuals aged between 55 and 64 are active on the labour market (Boulhol, 2014). Since 2015, the government led by the highly popular right-wing Law and Justice party has adopted a series of patriarchal (Gwiazda, 2021), anti-immigration (Krzyżanowski, 2018) and homophobic (O’Dwyer, 2018) laws. This political turn has also affected businesses, as diversity policies are increasingly seen to be in conflict with conservative Polish values (Gryszko, 2009).

Mainly serving the neighbouring German market, POZ1 today comprises 60 truck docks and 123,000 square metres, of which 75% is storage (Dembińska, 2016). Warehouse operations are divided into ‘inbound’ and ‘outbound’. Inbound refers to the registration of all goods into the warehouse management system database, through the receiving of the goods, the unpacking, quality control and labelling through an Amazon standard identification number linked to a location in the warehouse, and the placing of the product on the shelf (stowing). Outbound includes the picking of the products from the shelves, the packing and the setting into boxes (totes) to be delivered to clients by external couriers. Workers are organized in teams of about 20, led by a team leader reporting to an area manager who in turn reports to an operation manager. They work on 10-hour shifts including two paid 15-minute breaks and one unpaid 30-minute break for lunch, and alternate day and night shifts monthly. Those on a full-time contract work four days a week, with the possibility of one additional day a week in peak times. Part-time workers, a very small share of the workforce, work two days a week, usually at the weekends.

On average, POZ1 employs about 6,000 workers, who are called ‘associates’. They are recruited from the main town and the surrounding rural areas in a radius of about 120 km, with a majority of workers travelling a distance of over 25 km, reflecting the low local unemployment rate of 2.2% (Statistical Office in Poznań, 2019). Although official data is not available, like all Amazon warehouses world-wide (Amazon Workers and Supporters, 2018; Briken & Taylor, 2018; Delfanti, 2021), POZ1 employs a highly socio-demographically diverse workforce. Based on our interviews, we estimate that about 60% of the workers are women, about 40% are under 30, 30% are between 30 and 40, and about 30% are over 40 years. Workers range from high-school and university students (for whom the job is a way to gain income to study) to adult workers with families or other financial and social responsibilities (for whom this job is the main income and who look for stable employment) and retirees (who often work to supplement other sources of income). At the same time, the POZ1 workforce is almost exclusively Polish, reflecting the ethnically homogeneous population in the Poznań region.

Unmaking Borders Through the Algorithmic Management of Deskilled Labour

Unlike most organizations, which match workers to jobs they consider suitable for them, Amazon allocates work in the warehouse without any consideration of individuals’ socio-demographic profile, competences, experience or preferences. It manages them as unbordered, fully interchangeable units of labour:

The work is randomly assigned. You never know . . . On the day of recruitment, even the [temporary work] agency [which handles the hiring] doesn’t know what jobs they offer. Just when you sign up at the beginning they tell you: ‘You go to this department, your work is to pick’. There were people who wanted to drive forklifts because they had done it before, but no one listened to them. (male picker, 26 years old, #16)

Workers are moved around depending on where they are needed in the warehouse, often without prior communication. Respondents called these internal transfers przerzucanie, a term referring to an indistinct mass that is moved, typically coal shovelled to another place, pointing to the lack of consideration for individuals.

This random allocation of workers is made possible by the algorithmic management of labour in a deskilled labour process that has been described as ‘digital Taylorism-Fordism’ (Cattero & D’Onofrio, 2018). All workers carry barcode scanners telling them the next task and continuously registering their single acts of labour to be input, as data points, into the asset management programme (cf. Briken & Taylor, 2018). This system synchronizes these data with the orders placed by customers online and inventory data in real time, algorithmically optimizing warehouse operations. The automated assignment of tasks to interchangeable workers keeps ‘lead times’ to handle the goods as short as possible and largely eliminates the need for interpersonal coordination among them and with managers.

Many respondents elaborated at length on the mind-numbing nature of the work steered by the algorithmic system and how they developed strategies to keep up with the competition with all other workers, which they called ‘the norm’. This term referred to the automatically calculated, dynamic numerical benchmark derived from continuously aggregating the work data of all workers, used to identify underperforming ones for termination:

The system registers all your achievements, so when you have a bad day, it’s also on the disk. And when one day you don’t meet ‘the norm’, a leader approaches you, gives you a piece of paper to sign stating that. It does not matter that, let’s say, that for a month you were good and above ‘the norm’, this one sheet means that with the next paper or the next, you can get fired. And that’s why some people really work hard, no matter what. (female packer, 22 years old, #61)

To keep workers in a state of uncertainty and reduce the possibility that they enacted strategic behaviour lowering productivity, the actual norm was not made visible to them. Those who wanted to know how they were performing compared to others needed to find and ask a shop floor manager, a time-consuming action that would likely result in lower performance or a shorter break. Also, how the norm was actually used by the company remained unclear to workers, as some told us that they were transferred and terminated without prior warning that their productivity was too low.

Amazon’s lack of consideration of workers’ socio-demographic profile in work allocation regularly featured in interviews. Some respondents mentioned for instance how older workers who were not able to keep up with the heavy work and work pace were sacked:

Some older people were directed there. . . it wasn’t easy to receive there, the boxes were often heavy. Older ladies complained: ‘Why the pick again, my legs hurt, I have no more strength to walk. . .’ these people should be moved around, two days today on the pick, two days to receive, change, let her stand. Because you really need to walk on a pick, run, a lot. (female stower, 39 years old, #27)

There was a gentleman, 80 years old. . . Seriously, he was on the pack, big items. He always called someone to log him on because to scan you need a few clicks and he had no idea how to do it. But once logged in, he could work. He was fired in November. . . after two [negative] feedbacks. (male picker, 24 years old, #40)

While these respondents questioned the ethics of letting all workers do the same work and compete, independent of age, not everybody did. Others rather complained about having to compete with ‘different’ workers with a too strong work ethic and who were too productive, increasing the norm and letting them no chance to stay in work:

Some ladies, Jesus, I did not like to work next to them. I said, God, move me somewhere else, because grandmothers of 50+, they are just like robots, really. I was impressed, one did not take the products from the wall, she ran for them. One lady was even reminded by the manager [not to run], because she did not walk, a real massacre if you think about working like that the whole day. But this pace is because she is a person who has worked so hard all her life, without thinking. I had no chance to keep up with her. . . (female packer, 25 years old, #8)

Along the same lines, some groups’ excessive zeal was depicted using disparaging terms such as working like ‘monkeys’ or ‘little robots’ and being ‘brought up in a village’, etc. (female picker, 24 years old, #33), alluding to lower intellect and lack of understanding. These fragments suggest workers’ awareness of how Amazon’s unmaking of borders within the workforce increased competition and, consequently, intensified and precarized work for everybody. Abandoning any socio-demographic characterization of workers, one respondent vividly described Amazon as a jungle populated by animals fighting against each other to survive:

There is a primal rage, like in the jungle. Amazon is a typical jungle. Either you are meat, or you are going to eat meat. You have to act like a lion or rather like a panther and you have to be able to camouflage yourself to the herd and just not show anywhere, on anybody’s screen, act at your own pace but within the limits. [. . .] That’s the key, being fast, quiet, do your job and decrease your chances of being taken down [fired]. (male stower, 29 years old, #22)

In Amazon’s algorithmic management system, socio-demographic categories no longer operate as organizing principles. Workers are randomly assigned to deskilled jobs and re-constituted as data points collected by the scanner registering their single acts of labour. So measured, their performance is continuously processed and compared to the aggregated performance of all other workers, operationalizing competition as a dynamic, automatically calculated ‘norm’. This opaque algorithmic management system establishes a numerical border between workers based on their relative performance to identify those to be terminated, keeping all in uncertainty (cf. Alaimo & Kallinikos, 2020). Workers are so constituted as undifferentiated, deskilled and disposable labour to be continuously optimized to ensure the efficient flow of commodities and capital accumulation.

Unmaking Borders Through Non-Selective Hiring

To sustain a constant flow of undifferentiated labour to keep warehouse operations running, Amazon hired through practices that unmade socio-demographic borders within it. Many of our respondents told us that the temporary work agencies handling all of Amazon’s recruitment and selection hardly did any selecting. Job applicants were expected to meet only very minimal health requirements and could quickly start work after filling in a few administrative forms:

They don’t check the background or the education [. . .]. All that matters is that enough people get in. A person has two arms, two legs, the ability to work – it is enough: ‘Thank you. Next, next, next’ [. . .]. They ask for your PESEL [tax identification number], birth date, and to enter your personal data, but don’t examine you. Even the medical staff issue a quick certificate and you move on, just stand in line to go to Amazon. It’s automatic. (male stower, 34 years old, #17)

I remember [. . .] I arrived [at the temporary work agency] and then I called my mother to tell her: ‘Mom, I don’t know what it is. It was not an interview, they took everybody, believe me: all. Only one person. . . had some blisters on his hand and they told him he was not suitable, because of the mycosis. . . It was the only person, and there were hundreds of people.’ (female packer, 23 years old, #32)

Many fragments from our interviews centred on how Amazon hired all kinds of people, including ‘obese’ people, older people in their 60s and 70s, illiterate people, ‘mentally ill’ people, ex-convicts, people with a disability or a prosthesis, individuals with coronary bypass, etc., even if some of them eventually turned out not to be able to carry out the heavy warehouse work:

I’m surprised they accept elderly and obese people. But you see, they just accept everyone, because there is a constant problem [shortage]. There’s always someone coming and going, right? They accept older people and they really cannot do it at times. When I go back home by bus after work, I see those tired people, even young people are tired. . . We are talking about 60+ even 70. (female stower, 26, #21)

While this respondent emphasized the company’s need to have sufficient labour – any labour – in a system that continuously expelled workers, others acknowledged how Amazon’s non-selective hiring gave them opportunities:

I left prison [after 14 years], I went to a center that helps convicts [. . .] and they told me to look for a job. I walked to different Adecco offices and the other ones, and was offered work at Amazon. . .. They did not ask me [about his criminal record], and I did not say anything because they had a policy that they didn’t care [. . .]. These things [conviction] are hard to hide, but no one ever told me anything. Nobody cared, because the truth is that everyone was f*** up and they didn’t have time to look at who slept with whom, who sat and who was the altar boy. (male packer, 33 years old, #26)

A few young women workers narrated how this hard, exploitative work allowed them to escape unwaged work and soften oppressive relations at home or afford necessities, such as school for children or leisure activities.

Other respondents reflected more generally on how non-selective hiring gave work to individuals from marginalized groups who would not be hired by anybody else, and how this practice was favourable to Amazon, as it produced a particularly compliant, motivated, hard-working labour force to carry out high-pace, deskilled jobs:

Well, nobody wants to take these [older] people, Amazon accepts everyone. I once talked to a lady who was half a year away from retirement. [. . .] She said that she would keep working here, because nobody else would hire her at this age. . . in a store or physical work, because they want young, dynamic people. (female stower, 38 years old, #9)

Amazon is big [. . .] it gives them a chance. In my opinion, this is a plus. There are about 250 people who wouldn’t be able to find other work. You do your work: chop, chop, it doesn’t hurt you, that’s it. They keep such people, they know they can’t go anywhere else, because they are good at really hard work, but I don’t know if they would be good at another job and if this job would be easy for them to learn, like in Amazon. (male packer, 42 years old, #36)

Non-selective hiring was experienced less positively by workers from socio-demographic groups which had historically been more advantaged in the Polish labour market and society. They were surprised and disappointed that their professional attire, educational qualifications, work experience and aspirations, which had played a key role in accessing work in the past, were here of no relevance:

I was a bit disappointed, because they accept everyone, right? So, for example, I had a so-called ‘job interview’, but it was more of an information meeting, not a conversation. There were people who could not fill out a simple form. . . Three signatures. A bit upsetting, because I graduated from university and I had expected a warehouse, manual work, but didn’t think it would be like that. It blew my mind. (male packer, 31 years old, #4)

I always worked in the office. I’m used to dressing up to go to a job interview. [. . .] There were 150 people in this small agency and they didn’t even ask about your competences or expectations. Only: ‘Who wants part time? There! Who wants a full-time job? There!’ They welcomed everyone, distributed a questionnaire. [. . .] It was actually the employment contract itself. (male stower, 41 years old, #48)

For them, the unmaking of borders between workers and their treatment as interchangeable units of labour was demeaning and shocking. A few respondents even portrayed this hiring practice as potentially dangerous:

I’ve heard of people with a coronary bypass that work there, I’ve seen a few with prostheses, lots of physical disabilities, including being hunchbacked to a degree that they couldn’t see up. Somebody is folded in half but can work in Amazon, where I [able-bodied young man] can barely make my norm, is there any logic? And think about mental diseases, how many deviants, potential psychopaths might be getting in, not to mention a criminal record and that kind of thing, right? They are all in Amazon. (male stower, 34 years old, #17)

The company’s denial of borders and negligence to select was described as putting not only the workers belonging to these categories themselves in danger, but also other workers who could become victims of their acts of deviance or violence (female worker in her forties, forum post, January 2020).

Taken together, these fragments show how Amazon’s non-selective hiring practice turns workers into an undifferentiated mass of labour, unmaking socio-demographic categories commonly used to allocate workers to jobs in the Polish labour market. This renders them interchangeable from the outset, an essential condition to sustain competition and to dispose of those workers who do not perform sufficiently well. Non-selective hiring differently affects different categories of workers: it gives unprecedented work opportunities to workers of historically subordinated socio-demographic categories, reducing the relative historical advantage of others. In so doing, this organizational practice enhances competition and the precarization of labour as a whole.

Unmaking Borders by Enforcing Social Norms of Interpersonal Respect

In order to unmake the borders within labour in the everyday functioning of the warehouse, Amazon also enforced social norms of respectful interpersonal interaction between all workers. This practice is important as dominant social norms play a key role in excluding certain groups and sustaining segregation along socio-demographic lines. Like many multinationals, Amazon Poland officially commits to equality, diversity and inclusion, as indicated on the Polish company website:

We believe that building a friendly and inclusive culture is essential for people to do their best and what we can achieve as a company. We actively recruit people from various backgrounds and support the creation of an inclusive workplace. We take steps to ensure employees a sense of belonging, a sense of value, and the opportunity to achieve success and development. (Amazon Polish website, own translation)

This formal policy is reflected in the enforcement of respectful behaviour on the shop floor. Employees are instructed to always behave in a friendly manner, including greeting and thanking one another. Workers generally abide by these rules, we were told, as failure to comply could lead to disciplinary feedback and even termination. According to our respondents, instances of swearing or sexual harassment were rare, and when incidents did occur, managers intervened.

Our respondents’ narratives featured situations in which co-workers treated them without respect and were disciplined. For instance, a worker self-identifying as gay told us how his manager asked one of his co-workers to apologize to him for his rude behaviour:

I don’t know if it [the policy] sometimes protects. For example, once a guy was quite rude to me. A few offensive words. Well, a friend persuaded me to go [to the manager], and this guy came and apologized to me. I didn’t want it to be a big issue, quite embarrassing for him, it was all I needed. This person is still working to this day. We avoid each other. . . Everyone has the right to their own life, I don’t care about his opinion, it’s enough [that he behaves respectfully]. (male stower, 42 years old, #57)

In another case, a woman worker in her 50s recounted how a much younger colleague who was impolite and unhelpful to her was removed from the job next to her:

[In the beginning] I knew nothing about computers and was getting confused with some stupid popup. So Marta, she was maybe in her twenties, was next to me and I was looking at what she was doing. And she would hide her screen from me. [. . .] Later I heard her talk about me as ‘that old woman’ and I got really mad. I went to her and I kind of made a scene (laughs) [. . .] The team leader, he was maybe her age, asked if we had a problem, so I told him. And guess what? I never saw Marta again. They sent a trainer who spent about 3 days kindly showing me everything. (female packer, 55 years old, #18)

Amazon enforced social norms of respectful interpersonal interaction despite the fact that these norms contradicted prevalent ones in the Polish context. This was particularly visible in the case of social norms of respect towards LGBTQI+, as homosexuality remains a highly controversial topic and homophobic behaviour diffuse and tolerated:

[A co-worker] went to the manager, reported me. I told the manager that he was too much into me and I could not tell whether he fancied me, trying to date, or what? [. . .] The manager said that I spoke impudently, intolerantly or something, that I violated the rules of Amazon, that you have to be tolerant. And I told him: ‘Listen, it was because I felt embarrassed by how he was looking at me and I don’t know what he wanted.’ I said: ‘He raped me with his eyes.’ And you know, he started to laugh, it turned into such a comical situation and we were doing tolerance [being ‘politically correct’]. (male stower, 42 years old, #57)

Some of our respondents reflected on how working at Amazon exposed them to a wider variety of ‘different’ people than they would normally run into and this contact made them aware of their own prejudices, changing how they saw people and even appreciating differences:

[In Amazon] I worked with people [who looked like] real freaks but they turned out to be great. I became more tolerant of [tattoos and behaviours]. I was also surprised that I could just work with these people, because they turned out to be very hardworking. Instead of focusing on how somebody looks or behaves, I started looking at how they work. (male stower, 44 years old, #65)

[Nickname of worker] is a typical ‘Woodstock person’, he looks more like Satan: black pants, long hair, guitar, heavy metal. So here’s a tattoo, here’s a tattoo, long hair and aaah, right? And our grandmothers at packing hated him right away and talked behind his back. I started to help him, persuading people, to make it easier for the guy. So we became buddies [. . .] [In Amazon there are] cool people, very different people, different personalities, from different environments, ages, etc. I went to work thinking: ‘Who will I meet today?’ Although everything else is so mechanical. That made it fun even though the work would not really allow you to get to know anybody better. (male packer, 31 years old, #4)

Other respondents were less self-reflective, yet mentioned that they ‘did not mind’ having homosexual people in their department as their sexual orientation was ‘none of their business’, suggesting that the company succeeded in imposing social norms of universal interpersonal respect.

A trade union representatives explained to us how enforcing social norms of respectful interpersonal interaction sustained the exploitation of labour:

People at POZ1 have experienced exploitation in the form of some foreman coming to you and telling you: ‘Faster, you idiot, don’t be a prick!’ At Amazon, this kind of behaviour is unacceptable. If some manager told you anything like that, they would fire him [sic] right away. Violence only takes place with the feedback, but maximally you can get it once a week and they come and say: ‘I’m sorry, I have to give you a reminder in the form of feedback and we need to know how we can help you, because you do not meet the norms.’ But this type of violence is even greater, in my opinion. It hurts people, after 3 years of hard work in anxiety of being terminated, a long time, this hurts you directly. I have boys in the hall who worked at Samsung, a typical ‘work camp’ [military culture]. When the manager harassed them, they locked him in a closet for 12 hours, and beat another one up. I do not support that behaviour, but you can see how it’s different [from Amazon]. (female trade union representative, #TU 1)

Comparing the relations between workers and shop floor managers in Amazon with those in other companies in Poland, she argued that such social norms imposing good behaviour hid the profound violence of employment relations resting on workers’ uncertainty and fuelling anxiety. This organizational practice contained the resistance of male youth, who in more segregated, masculine contexts would take overtly violent action to contest the exploitative and abusive working conditions.

Amazon’s enforcement of social norms of universal interpersonal respect undoes socio-demographic borders as principles of organizing social interactions unequally inside the workplace. Against social norms prevalent in Polish society, this organizational practice equalizes interactions between workers from different socio-demographic groups and even raises some workers’ awareness about their own prejudices. Norms of universal interpersonal respect are essential to dedifferentiate socio-demographically diverse labour, so that the relations between workers are equalized and workers from subordinated groups do not leave due to harassment and discrimination, but solely on the basis of the algorithmic assessment of their productivity by the company. This practice thus contributes to maximizing competition and the exploitation of labour as a whole.

Unmaking Borders Through Casualized Employment for All

Next to these organizational practices, Amazon also unmade socio-demographic borders within labour through the employment of all workers at the same casualized conditions. Temporary work agencies carry out all initial recruitment and employ two thirds of the workers (the ‘green badges’), the remaining third are Amazon’s direct employees (the ‘blue badges’). Workers initially receive a one-day contract for the introduction and then contracts of two weeks, one month and three months. While workers employed directly by Amazon typically have longer contracts, they never have permanent employment and are paid slightly less than workers employed by the agencies. Although many respondents had a general aspiration to become Amazon employees, in the interviews with us, respondents did not categorize or associate with others based on the colour of their badge, suggesting that precarity was unrelated to the formal employer.

Like all other Amazon warehouses worldwide, POZ1 continuously hires and fires to respond to fluctuations in commodity flows. Workers go on and off work as the company needs them, largely unaware of how employment decisions are made. The lack of transparency of the criteria used to fire generated much speculation: many workers posted on Facebook fora that they had been simply informed they could not come back to work the next day. This practice kept them in structural uncertainty and highly aware of their precarity:

What would give a sense of stability, would be a feeling that this company respects your work, but it does not. They say it does but that’s bullshit: ‘Well, Mr. [name], you are missing 0.8% and your numbers have been fluctuating lately’. . . From the HR perspective, I was redundant, because I was 0.8% below the norm. [I am] only a number in Excel, added or removed. It does not matter to them. Only the norm counts. (male stower, 34 years old, #55)

Despite Amazon’s treatment of workers as anonymous, fully interchangeable units of labour, the very harsh working conditions, the algorithmically generated norm enforcing unremitting competition and the structural precarity of employment, many respondents pointed to how Amazon was a better employer than most:

I quit my job in gastronomy after nine years of working for small crappy businesses. It was time to find a larger corporation that will, so to say, give ‘normal’ contracts, pay on time and take care of employees’ work and safety conditions. We went with a friend to Manpower to look for any job. We were looking for something reliable, and we got it. We got an offer to go to Amazon, and we said: ‘Well, let’s take the risk.’ (male packer, 31 years old, #4)

Newly recruited POZ1 employees received a wage of 20 PLN (about 4.40 euro) pre-tax per hour, as well as monthly bonuses of 7.5% depending on individual performance and an additional 7.5%, based on the overall warehouse performance. They also received one meal a day in the canteen and a free bus to commute. Altogether, POZ1 workers thus earned well above the Polish minimum wage of 17 PLN (3.74 euro).

The endemically precarious employment conditions in other workplaces in the area were often mentioned: substantially lower pay rates, the risk of not being paid at all, the expectation of full flexibility, lack of benefits, informal employment, abusive behaviour and physical harshness. They were however particularly prominent in the narratives of workers belonging to historically subordinated groups, such as women and migrants, and/or in particularly vulnerable situations, such as ex-convicts:

People complain that it’s such hard work, we work like ants and the salary is low. But when I got hired, for me the money was: ‘Wow!’, because my first job was in a small grocery store, were I earned 8 PLN an hour. Compare 8 to 16 PLN, and you see how I felt about my first paycheck. (female picker, 24 years old, #33)

[When I got out of prison] [Amazon] made my life stable. The pay was really great, above three grand [3000 PLN]. Amazing meals for 1 PLN. I don’t know how cook so that was really a big help. I had to buy everything, was not sure how to do things and had no home to return to. I had dinner at Amazon, I managed to rent a room and had a bus to work. Otherwise I would have been on the street. (male packer, 33 years old, #26)

[I started working here] because generally financially. . . [Amazon] was better. I came back from Germany after 3 years. . . I had worked in retail there and had almost no money, it did not pay off, nobody makes a fortune there anymore. (female picker, 34 years old, #6)

Along the same lines, some women workers were appreciative of POZ1’s compliance with Polish maternity legislation. Others told us how the highly predictable shift system allowed them to combine warehouse work and care at home, even though the long shifts, extremely fast-paced work and long commutes exhausted them and undermined their health.

While comparison with other local employers emphasized the perks of employment at Amazon, the comparison with the conditions offered by Amazon to workers in Germany and other western countries pointed to substantial inequality:

Amazon moved the warehouse here for one reason. What they used to pay in euro, now they pay in [Polish] złoty. It is four times less than in Germany. And we work harder because we are still behind [compared to western European countries]. (female picker, 38 years old, #70)

Amazon gave us the terms of the December bonus [for peak time] for consultation. They proposed 600 Polish złoty gross (135 euro) for full-time employees. Amazon employees in Germany, Spain, Italy and France would receive 300 euro, 300 US dollars in the United States, and 300 pounds in the United Kingdom. I think that we have a big problem, we are really doing our best to show that to workers. (male trade unionist, #TU 2)

Despite this visible inequality, trade union representatives explained that workers’ mobilization was difficult, due to the high turnover and Poland’s legislation requiring a ballot with a turnout of at least 50% of the workforce and a majority vote in favour of striking, a quorum that had till then never been reached.

The practice of casualized employment conditions for all workers unbordered labour in POZ1, equalizing the employment of historically subordinated and advantaged socio-demographic groups, creating the legal terms to enforce universal competition, uncertainty and precarity. This practice rested on a capital-friendly Polish legislation only minimally protecting workers, dividing them from workers employed under German legislation. It thus reaffirmed the national border as essential to Amazon’s global strategy of delocalization to low-labour-cost countries with weaker workers’ mobilization. At the same time, these casualized employment conditions drew a border between POZ1 labour and labour employed by other companies in the Poznań region, which often offered even more precarious employment. Our analysis shows how multiple borders are simultaneously mobilized to create and juxtapose multiple ‘normalities’ (Lazzarato, 2009) to enforce precarity and valorize labour for capital accumulation.

Discussion

This study has drawn on a critical theory of borders (Lazzarato, 2009; Mezzadra & Neilson, 2013; Neilson, 2012) to empirically investigate the governmentality of labour in POZ1, an Amazon warehouse in Poland employing a diverse workforce. Our analysis has unveiled what we call a ‘post-diversity’ labour governmentality. Such governmentality is based on the (un)making of socio-demographic borders along the lines of gender, age, sexual orientation, criminal record, etc. within the workforce, despite their widespread use to fragment and unequally exploit and oppress labour in Polish workplaces and society.

These insights advance the critical diversity literature in multiple, related ways. First, we show how global capital unmakes socio-demographic borders to govern labour. We have identified the organizational practices – the management of deskilled labour through an algorithmic system, the non-selective hiring of workers, the enforcement of social norms of interpersonal respect and a universal system of casualized employment – through which diverse workers are turned into indistinct, equal and equally disposable labour units (Mezzadra & Neilson, 2013). Such unbordering, our analysis shows, is essential to let all workers compete with each other, independent of their socio-demographic profile, in a ‘race to the bottom’ fuelling uncertainty and keeping them in a condition of precarity, as disposable workers (Lazzarato, 2009). Exploitation and precarity are here no longer solely defining of the condition of workers belonging to specific socio-demographic groups, marginalized and segregated in less protected and less rewarded jobs. Rather, they define the condition of the workforce in its totality (cf. De Coster & Zanoni, 2023; Neilson & Rossiter, 2008). Amazon does not solely take advantage of the existing (institutional) vulnerability of certain socio-demographic groups of workers, such as migrant workers and the unemployed (Boewe & Schulten, 2019; Briken & Taylor, 2018; Cattero & D’Onofrio, 2018). By unmaking borders within labour, it more fundamentally enacts a form of labour governmentality that turns it into a structural ‘reserve army’.

Second, we show that, different from critical diversity studies, in ‘post-diversity’ labour governmentality, power no longer operates through an ‘ideal worker’, a normative ideal that cannot be attained (Acker, 1990, 2006; Ashcraft, 2013; Benschop & Dorewaard, 1998). On the contrary, it is equality between workers that facilitates the extraction of value: each worker is integrated into the process of capital accumulation under the same conditions as any other unit of labour. Amazon mobilizes the principle of equality to undo the heterogeneity of discrete, embodied labour. Abstracted from any normative ideal in its relation to capital, labour ceases to be ontologically defined by its symbolic and embodied difference, it ‘overcomes’ its own diversity. Once so abstracted, it can be assessed against an algorithmically calculated productive norm that dynamically evolves with the aggregated productivity of all workers at any given time.

Importantly, equality governs historically subordinated and advantaged workers in distinct ways. For the former, equality elicits consent to exploitation, as they experience equality as countering the structural disadvantage they commonly encounter in the labour market and society (cf. Fraser, 2013; Peticca-Harris et al., 2020). It carries with it a micro-emancipatory dimension relative to the forms of oppression along predominant patriarchal, heterosexual and ageist norms (cf. Zanoni & Janssens, 2007), against the background of conservative social norms in Polish society (Gryszko, 2009; Gwiazda, 2021; Krzyżanowski, 2018; O’Dwyer, 2018). For the latter, equality entails the disavowal of their historical advantage, to let them compete with all and contain their expectations and demands vis-a-vis capital. Often overlooked in the critical diversity literature, these governmentality effects are key to understanding widespread resistance to equality by historically advantaged groups of workers (cf. Bohonos, 2021).

Finally, our study advances critical diversity studies by unveiling the multiscalar nature of (un)bordering characterizing the ‘post-diversity’ governmentality of labour for capital accumulation. While our focus was the unmaking of borders through practices inside the organization, relying on critical border theory, our analysis also shows how those practices cannot be understood independent of other borders multiplying labour at different scales on which they are predicated (Lazzarato, 2009; Mezzadra & Neilson, 2013; Neilson, 2012; see also Andrijasevic, Rhodes, & Yu, 2019; Zanoni & Pitts, 2022). Most visible is the national border between Poland and Germany. By dividing labour within the EU, this border operates as a selective ‘filter’ that keeps labour protection outside Poland – a prerogative of labour in Germany – while at the same time allowing the unlimited circulation of commodities (cf. Mezzadra & Neilson, 2013; Neilson, 2012). As other studies of Amazon have shown, so also in POZ1, this border allows capital to impose casualized employment on labour, leveraging the country’s relatively weak labour legislation, endemic precarity, lower wages and limited workers’ mobilization (Boewe & Schulten, 2019; Cattero & D’Onofrio, 2018).

Yet our analysis reveals that post-diversity labour governmentality rests on multiple additional borders. The identified practices establish a border between the organization’s own labour and the remaining labour in the region, by offering Amazon’s own workers somewhat better employment conditions than what they expect from other employers. This border is essential to ensure a continuous inflow of labour into a labour process that, to sustain high exploitation, continuously expels the least productive workers. This labour governmentality further dynamically establishes a border within its labour force based on workers’ productivity relative to each other. Algorithmically enforced competition first fragments individual workers into single acts of labour captured as data points. It then benchmarks them against an aggregated ‘norm’, a numerically expressed border that identifies the least productive workers to be disposed of, keeping all in uncertainty about future employment and precarizing their lives.

Conclusion

Taking the perspective of the most exploited and oppressed workers, critical diversity studies have to date largely overlooked the possibility of ‘post-diversity’ organizing, where socio-demographic categories no longer operate as principles of organizing. This paper suggests that employers might no longer deploy such categories as principles of organizing when equality can be more effectively used to exploit the diverse workforce as a whole. More specifically, employers might enact practices that constitute workers as undifferentiated, interchangeable and equal to enforce competition, making labour disposable and precarized, turning it into a reserve army in its totality. A critical theory of borders advances the critical diversity literature by offering a novel vocabulary to ground difference in contemporary global processes of capital accumulation. If, as Neilson (2012) holds, capital’s ‘logistical fantasy of creating a frictionless world is [. . .] intimately wedded to the desire to eliminate the gap between [diverse] living and abstract labor’ (pp. 335–336), then Amazon appears to come closest to its realization.