Maranke Wieringa | Approaching algorithmic account(-)ability: developing tools to foster formalized and practical transparency in municipal data projects

Maranke Wieringa | Approaching algorithmic account(-)ability: developing tools to foster formalized and practical transparency in municipal data projects |  overarching project: Governing the Digital Society | Utrecht University, Department of Media & Culture | Supervisors: prof. dr José van Dijck, prof. dr Albert Meijer, dr Mirko Tobias Schäfer | September 2018 – August 2022 | m.a.wieringa[at]uu.nl

This research aims to tackles the question: What is the current view on algorithmic accountability within municipalities, (how) do they practice it, and how can this practice be improved? The first part of the investigation will thus sketch a general overview of algorithmic accountability, in collaboration with VNG Realisatie. Aside from this national picture, the project will also look into a concrete case: the algorithmic fraud detection system implemented by several municipalities to screen those who receive benefits (Hijink 2018), which were the topic of recent parliamentary discussion (Security.nl 2018).

The second part of the project will establish a practical intervention in the field: a toolkit will be developed in close collaboration with municipal data professionals (e.g. data scientists, decisionmakers). This collaboration will reveal unique insights into professional perspectives and current practices regarding algorithmic accountability. Developing the toolkit provides ‘a seat at the table’, and ‘anthropologicalesque’ insight into the data professionals’ attitudes which will feed back into the former part of the research.

The aim of this toolkit is to help these professionals reflect on the algorithmic decision making process, to formally embed this reflection into municipal data projects, and to create a valid method to provide testimony to the working mechanisms of algorithms and the public values which inform them. In other words, to give more insight in the ‘black box’ (Pasquale 2015) of municipal data projects, conform the GDPR.