Algorithmic accountability and digital justice: A critical assessment of technical and sociotechnical approaches.

Abstract

The concept of digital justice is intended to open up discourse about strategies for bringing relief to those who believe they have been discriminated against or harmed by algorithmic decision making. Digital justice has depended on algorithmic accountability, a means by which entities can be held accountable for the consequences of algorithmic decision making. This paper critically examines the concept of algorithmic accountability to assess its utility as a ground for digital justice and argues that it is fraught with difficulties. After discussing digital justice and algorithmic discrimination, algorithmic accountability is decomposed into two types, technical and sociotechnical. These approaches are critically assessed and a cautionary note is struck about the difficulty of enacting algorithmic accountability. If this argument is persuasive, it implies that the concept of digital justice also has difficulties. The paper concludes with suggestions for moving forward that do not use either version of algorithmic accountability.