The algorithm’s middlemen: resistance, compliance and managerial contradictions in Amazon’s warehouses in Poland

Purpose The purpose of this paper is to examine how middle managers in Amazon’s Polish warehouses navigate the tensions between algorithmic control, corporate expectations and their own managerial agency. It addresses a gap in existing literature by shifting focus from frontline workers to the overlooked experiences of middle management within digitally automated environments. The study investigates how these managers perform their roles under algorithmic oversight while simultaneously engaging in subtle forms of resistance, offering a deeper understanding of managerial work in technologically governed labour regimes. Design/methodology/approach This study is based on 26 in-depth interviews with middle managers across two Amazon warehouses in Poland, conducted during two research phases. A qualitative approach was used to explore how algorithmic control influences managerial roles and resistance. Using emergent thematic coding and Scott’s concept of hidden transcripts, the analysis identifies patterns of covert resistance and compliance within the context of digital labour governance. This interpretive methodology allowed for the capture of nuanced, informal managerial practices that would remain inaccessible through quantitative or surface-level investigation. Findings This study finds that middle managers at Amazon’s Polish warehouses occupy a contradictory role as both enforcers of algorithmic control and covert resisters of its limitations. While publicly adhering to corporate protocols, they privately reinterpret data, bend rules and engage in informal practices that challenge rigid performance metrics. These infrapolitical acts allow them to maintain operational functionality and uphold ethical commitments without direct confrontation. The findings reveal that algorithmic governance does not eliminate human discretion but instead reshapes it into hidden, strategic behaviours, highlighting the persistence of managerial agency within automated and highly controlled work environments. Originality/value This paper offers a novel contribution by foregrounding the experiences of middle managers − an often-overlooked group − in the context of algorithmic management. By applying James C. Scott’s theory of hidden transcripts, it reveals how managerial resistance can exist within compliance, challenging dominant assumptions about automation eliminating discretion. The study provides original insights into how algorithmic systems are mediated, negotiated and occasionally subverted by those tasked with enforcing them. It is valuable to scholars of digital labour, organisational studies and critical management, offering a nuanced understanding of how power and resistance operate in highly surveilled, data-driven workplaces.