CAREER: Recasting Algorithmic Management in the Gig-Economy
This project studies the ways that algorithmic management, using digital tools to automate and remotely manage workers, may negatively impact workers and their rights. Ride-hailing platforms, which are rapidly replacing traditional taxi services, are a canonical example of algorithmic management, where the software platform uses a variety of opaque means to automatically assign and evaluate work. The research community has had difficulty studying and improving these platforms, due to the platforms' proprietary and closed nature. This has exposed both drivers and passengers to biased or unfair outcomes, such as passengers poorly rating minority drivers or drivers declining rides from certain categories of customer. At the root of these problems are the opaque mechanisms for algorithmic management, and the several research and community efforts to make algorithmic management more palatable have largely impacted isolated functionality. These isolated cases point to a larger and urgent need to re-imagine these platforms as equitable workplaces, where we hold our algorithmic managers (and the people that develop them) to the same standard that we hold human managers. This work has the potential to extensively inform and redefine the standards and policies around how platforms that algorithmically manage work should be designed to form more equitable work environments.
To accomplish this goal it is necessary to build a more open platform, so that we can directly investigate these mechanisms. This project will develop an experimental ride-hailing platform that gives drivers and passengers control over parameters that impact algorithmic outcomes, as a means to understand and interact with the platform. It will serve as a testbed to conduct a series of mixed-methods studies that progressively increase in size and scope. These studies will be focused on three major themes: (1) understandable individual interactions with algorithmic managers, (2) equitable group interactions through algorithmic managers, and (3) self-governance for organizations with algorithmic management. In partnership with independent drivers and experts in labor relations, evaluation studies will use the platform to generate foundational knowledge about giving workers more control over the algorithms that manage them. Due to the open source nature of this work, any group that wants to begin their own cooperative platform will be able to utilize the application, tools, and resources developed in this research. These contributions will have a wide impact on how work is managed by algorithms and how to create human-centered algorithms and platforms for work.