Abstract below, please use the links above for the full detail.
Abstract
Scholarship on “algorithmic management” focuses on how employers use algorithms and digital devices to routinely control workers. It also looks at how workers resist this control. But how does algorithmic management affect the ability of employers to respond to collective action? The answer is important because algorithmic management is the “new contested terrain” of labor struggles. Drawing on 42 interviews with workers and court documents obtained through a Freedom of Information Act request, the author examines Amazon’s antiunion campaign in Bessemer, Alabama. The findings reveal that employers can weaponize elements or effects of algorithmic management against unions via repurposing devices that algorithmically control workers, engaging in “algorithmic slack-cutting,” and exploiting patterns of social media activity encouraged by algorithmic management. These findings demonstrate that the labor process can shape counter-organizing opportunities for employers, not just organizing opportunities for workers. They also reveal that algorithmic management has upgraded the antiunion arsenal, while shedding light on other aspects of algorithmic management that have escaped notice. The discussion section presents a framework for researching how other workplace variables shape counter-organizing. The conclusion discusses implications for our understanding of workplace regimes and the future of labor control.
Much of the world has entered an age of “surveillance capitalism” (Zuboff 2019), in which a “society of algorithms” (Burrell and Fourcade 2021) increasingly shapes social, political and economic life. Nowhere is this trend more pronounced than in a growing number of virtual and physical workplaces. These are the laboratories in which employers can turn de facto “captive populations” (Zuboff 2019:157) of workers into, as Amazon’s algorithmically powered warehouses have been described, “unwilling subjects in a sophisticated encompassing experiment in digital surveillance” that is having “disastrous impacts on their bodies and lives” (Delfanti, Radovac, and Walker 2021:4) Following in the tradition of labor process theory (Braverman 1998; Burawoy 1985; Edwards 1979; Thompson and Smith 2010; Wood 2021b), scholars have tracked the solidification of such efforts into a control technique that involves the use of algorithms and supporting surveillance and communication technologies to automate managerial tasks and control workers (Kellogg, Valentine, and Christin 2020; Lee et al. 2015; Wood 2021a).
Recent research has focused on how employers use this technique, known as “algorithmic management,” to direct, evaluate and discipline workers on a day-to-day basis (Kellogg et al. 2020; Lee et al. 2015; Vallas, Johnston, and Mommadova 2022; Wood 2021a). It has also looked at how workers experience and resist algorithmic management (Kellogg et al. 2020; Lei 2021). But what happens after workers take collective action against employers that use algorithmic management? Can algorithmic management also affect the ability of employers to repel organizing by workers, not just control their routine activity?
The answers are important because algorithmic management is the ascendant control technique and the “new contested terrain” of intensifying labor struggles (Edwards 1979; Kellogg et al. 2020; Schaupp 2023), as well as a hothouse for the development of new surveillance and algorithmic manipulation techniques more broadly (Zuboff 2019). Forms of algorithmic management are rapidly expanding far beyond their birthplace in the gig economy, reaching into logistics, retail, manufacturing, hotels, customer assistance, banking, law enforcement and many other industries (Wood 2021a:1). The outcome of the escalating struggles over algorithmic management therefore could shape the future of work, capitalism and political systems by helping revive or further erode one of the most equalizing forces in history: the labor movement (Burawoy 1985; Edwards 1979; Reuschemeyer, Stephens, and Stephens 1992; Rosenfeld 2014; Thelen 2019:16).
This article reveals how algorithmic management can shape employer resistance to collective action by examining a key early battle in the new labor upsurge: the first large union election at an Amazon warehouse, which took place in Bessemer, Alabama, in early 2021 (Brook 2021). Drawing primarily on 42 in-depth interviews with Amazon workers and analysis of court records obtained through a Freedom of Information Act (FOIA) request, I do so by highlighting how Amazon exploited constitutive elements or effects of algorithmic management to resist the union drive. Specifically, I show that Amazon (1) weaponized workplace devices that algorithmically direct and discipline workers, (2) engaged in what I call “algorithmic slack-cutting” to partially restore consent, (3) leveraged an app that automates human resources (HR) management to disseminate antiunion messages, and (4) exploited distinctive patterns of social media activity encouraged by algorithmic management.
Taken together, these findings suggest that algorithmic management generally enhances the antiunion arsenal of employers by guaranteeing the presence of additional capacity to intimidate workers, shape their preferences and rapidly restore consent.
These findings contribute to literature on labor studies, labor process theory, algorithmic management and the sociology of work. The empirical contribution of this research is to show that algorithmic management strengthens the antiunion arsenal—the weapons used by employers to repel a union drive after one has begun—and expands the power of employers to influence their workers more generally, including through penetrating and flexible communication capacities and through “algorithmic slack-cutting.” The main theoretical contribution of this research is to provide a framework through which scholars might investigate how other control techniques or dimensions of firm architecture (Lei 2021) can shape counter-organizing opportunities for employers, not just organizing opportunities for workers. In doing so, I demonstrate the value of bridging labor process theory with the employer counter-organizing literature. The findings also have implications regarding analysis of workplace regimes, “flexible despotism” as a macro regime and future uses of algorithmic management, which could include influencing the political and social behavior of workers beyond the workplace, not just their technical and political behavior within the workplace. Overall, these contributions recommend further inquiry into how employers scan weaponize elements or effects of the labor process to undermine union organizing and shape economic, social and political behavior more generally.