Is “compromised algorithmic control” equivalent to “compromised performance”? The effect of algorithmic control on the performance of crowdsourced workers

Abstract

In the rise of online labor platforms, algorithmic control has a profound impact on numerous crowdsourced workers. But although algorithmic control aims to improve performance, it also can easily cause harm to workers. To explore a reasonable range of algorithmic control, based on the job demands-resources theory, three studies were conducted with crowdsourced food delivery riders and ride-hailing drivers to test the influence of algorithmic control on their performance. Our findings demonstrate three key mechanisms: (1) algorithmic control reveals an inverted U-shaped relationship with job engagement in which moderate levels optimize worker role integration; (2) algorithm familiarity moderates this curvilinear relationship by amplifying the effects at both extremes of excessive and insufficient control; and (3) job engagement mediates the influence of algorithmic control on performance outcomes, and such a mediating role has been further moderated by workers’ familiarity with the algorithm. The findings facilitate a comprehensive understanding of the impact of algorithmic control, offering practical guidance for algorithm developers and enterprises in formulating reasonable control strategies.