Governance of Algorithms, Governance by Algorithms: Are 'Decentralised Autonomous Organisations' a Blueprint for Participatory Digital Organisation?
17 Pages Posted: 4 Apr 2022
Date Written: December 12, 2021
Algorithms are inherently centralised processes, from coding, to training, to deployment and maintenance. Meanwhile, blockchain communities are experimenting with “Decentralised Autonomous Organisations” (DAOs) as a participatory institutional framework for individual autonomy to organise outside organisations. DAOs are an attempt at decentralised organisation to self-govern, using algorithms, for autonomy from third-party mediation. This piece explores if DAOs can teach us anything decentralised approaches to the governance of algorithms. With the rise of algorithmic decision-making systems in public administrative processes, this research seeks to uncover the dynamics of DAOs as participatory ways to organise outside of organisations in the digital age. I explore the case study of “GitcoinDAO” as a decentralised organisation governed by algorithms, whilst simultaneously seeking to collectively govern algorithms to manage a machine learning process to detect fraudulent “sybil” attacks. With algorithms as peers in decentralised organisations, algorithms emerge as new political actors in how people organise outside of traditional organisations in the digital age. DAOs provide this institutional framework in the articulation of shared objectives, codes of conduct, and “constitutions”, to guide algorithmic governance design. This locates humans and algorithms as peers in organising, establishing algorithms as political actors in shaping and determining the outcomes of decentralised organisation and human autonomy. This research provides insights into the social outcomes of algorithmic governance for others seeking to explore participatory digital institutional infrastructures.
Keywords: DAO, blockchain, algorithmic, governance, infrastructure, ethnography
JEL Classification: D02, D80, O33, O35, O30, Z10, Z18
Suggested Citation: Suggested Citation