A systematic review of algorithmic power, economic dependence, and the epistemology of freedom in platform work

Abstract

Platform companies present the freedom to choose when to work as the defining virtue of gig labor, and labor economists have priced that flexibility. Yet workers’ income and access to tasks are governed by algorithmic assessments they can rarely see or contest. This article asks whether temporal flexibility suffices to call the platform–worker relation free. Following SPAR-4-SLR and PRISMA 2020, we analyze 95 studies (2015–2026), identified through AI-assisted retrieval from Scopus, citation chasing, and targeted searches, using critical interpretive synthesis. The evidence base is expanding rapidly but is epistemically segregated: psychometric studies, all published since 2024, recast algorithmic power as perceived job characteristics; legal scholarship proceduralizes it as information and review rights; and econometric valuations of flexibility are conditional on rules that platforms write and revise. We therefore distinguish first-order discretion (choosing when to work and which tasks to accept) from second-order control (the capacity to know, contest, exit, and collectively shape the rules). Republican non-domination, the capability approach, and Islamic contract ethics converge, from different foundations, on four requirements: knowable rules, contestability, viable exit, and collective voice. Islamic contract ethics contributes the concept of relational gharar, excessive uncertainty located in the rules governing a relationship rather than in single transactions. We use this framework to assess ILO Convention No. 193, the EU Platform Work Directive, and Southeast Asian regimes, and we propose an agenda for measuring second-order control.