Studies on online labor platforms (OLPs) have revealed that OLPs can have extensive managerial control over independent workers, which affects their autonomy and precariousness. The permeability of the management makes some OLPs' roles as neutral intermediaries in labor exchanges questionable. While there are several platform work studies on the effects of human resource management (HRM) activities, earlier studies have focused more on certain types of OLP companies. Earlier OLP classifications did not make systematic distinctions between HRM activities either. This paper offers a classification to view how HRM activities manifest in OLPs. The study utilizes terms of service and webpage data from 46 multinational and Finland-based OLPs. Based on these data, OLPs have been classified into six models with five governance principles and institutional logic. The study uses the idea of institutional complexity and claims that OLPs balance their operations between the complexity of two institutional logics, market, and corporation, by using varying strategies with their HRM activities. Differently managed OLPs are also often marketed to different worker groups. This indicates that workers' levels and quality of autonomy differ between OLPs. Hence, could be expected that platform workers' expectations toward OLPs, perceptions of fairness, and experiences of wellbeing may be influenced by the HRM activities in which they engage. The results contribute to the ongoing discussions of power asymmetries between OLPs and platform workers, and thus OLPs' roles as either marketplaces or hierarchical corporations. Formed models can be utilized to enrich studies on key issues of platform workers' autonomy, precariousness, and experiences in different types of OLPs.

Much of the research on platform workers has focused on individuals involved in low-skilled and highly standardized tasks. However, platform workers are not a homogeneous group. Utilizing a classification system that makes a distinction between different layers of platform control and grouping platforms according to how they divide decision rights between platforms and workers, we examine how and for what purposes platform workers operating in three types of control contexts have practiced and developed their digital agency for making out. The study, based on an analysis of platform webpages and 32 semi-structured interviews of food couriers, freelancers, and interim managers, shows that workers can exercise their digital agency on all three types of platforms, but different platforms create different conditions for this depending on their special forms of control. In addition, the forms of control also affect to what extent workers are motivated to direct their agency for making out. Instead of regarding platform work as just another layer of a periphery segment in the labor market, our analysis suggests that platforms exercising algorithmic control are new types of arenas for work, which seem to reproduce, or even amplify, the inequalities found in the offline world of work in the digital world.