The Algorithm Agenda: A Discourse Analysis of the Dutch Algorithm Register.

Abstract

This paper examines accountability and transparency in the Algorithm Register (AR) in the context of the Netherlands government's algorithm use. This paper aims to explore how transparency and accountability are implemented within the AR's functions, especially when considering criticisms of their application in both technical and political spheres. They are explored using Critical Discourse Analysis on government letters and AR development forum discussion and documentation. Additionally, the paper employs Category Analysis for the AR's information categories. The study highlights the AR's emphasis on transparency, accountability, and clarifying governmental algorithms. Finding that accountability is focused on administrative aspects, which deviates from the conventional three-stage accountability process encompassing informing, debating, and consequences. Transparency, rooted in government discourse, is selective. The AR adopts open-source practices and platforms but centers on efficient governance. By centralizing algorithms, it aims for efficient supervision and increased transparency. However, there are gaps: the AR promotes transparency but isn't wholly transparent, and it prioritizes feedback over enforcing consequences. The AR leans towards technical issues over societal concerns, and AR development discourses may validate governmental algorithmic development biases, equating accountability through transparency with responsibility