Previously we discussed the power that a platform has over its participants. We can turn this question on its head and ask how platform participants can organize so as to exercise power over the platform.

Recently, we studied this question from the perspective of collective action in the gig economy. The goal of algorithmic collective action is for a group of participants on a platform to strategically modify their data so as to steer the behavior of the predictor learned by the platform toward a common good.

Our work provides a first theoretical framework for algorithmic collective action. Our theoretical and empirical results demonstrate that even small collectives can exercise leverage over a platform’s learning algorithm []. Subsequent work by Baumann and Mendler-Dünner extends these results to the case of state-of-the-art recommender systems as can be found on popular streaming platforms [].

Unfortunately, when the platform uses not a predictive algorithm, but rather a more combinatorial matching mechanism as is common in ride hailing platforms, the results are quite different. Here, worker oversupply arises easily and workers have little to no leverage in situations where work is in excess supply []. On a positive note, these results point to effective strategies, such as worker segmentation and taking shifts.

Computer science and economics have focused predominantly on a model of individuals that act rationally but in isolation of others based on their own data alone. Algorithmic collective action breaks from the mold by focusing on the power of collective strategies. In doing so, this new area has the potential to improve working conditions in digital economies by providing increased leverage to workers.

Algorithmic collective action charts a path for individuals to meaningfully contest platform power. Our work provides a scientific backbone to the many documented organizing efforts of platform workers in the gig economy. We have also made an effort to document and systematize these efforts [].

The growth of generative AI has recently lead to increased calls for collective action. Many fear that companies have absorbed large swaths of copyrighted material from the internet, thus disenfranchising content creators from economic rewards for their labor. In response, scholars and activists have called for better ways that content creators can leverage their data. This development underlines the importance and future potential of this new research area.