The machine of muddling through: Machine learning, evidence and epistemic changes in policy science

• Fernando Filgueiras

  Federal University of Goiás , Escola Nacional de Administração Pública , Instituto Serzedello Corrêa

  Author

  https://orcid.org/0000-0001-9570-8113

According to Lindblom, public policy relies on limited analysis, is formulated under conditions of bounded rationality, and operates with little or no explicit programme theory. Policies are shaped by trial and error and achieve only incremental results. In this article, we examine the epistemic changes that arise in the theory and practice of public policy when machine learning tools are introduced into the policy process. Machine learning algorithms embody Harold Lasswell's vision of a scientifically grounded policy enterprise sustained by robust evidence about social conditions. These algorithms accelerate and optimise data production. Yet they also create new challenges for policymakers, including questions about the transparency of algorithmic architectures, the quality of the results produced, and persistent problems of reproducibility. We argue that the industrial character of these algorithms and their uncritical adoption do not support a policy process founded on scientific knowledge. Machine learning algorithms carry significant implications for the epistemology and practice of policy science. Their outputs are shaped not by scientific evidence but by processes of trial and error and experimentation conducted within opaque design processes. The design of algorithms thus converges with the design of public policies, generating new practical dilemmas for policymaking. We conclude that the incorporation of machine learning into the policy process produces what we call machines of muddling through: systems that alter the epistemic foundations of policy work without resolving its fundamental uncertainties.

1. Fernando Filgueiras, Federal University of Goiás, Escola Nacional de Administração Pública, Instituto Serzedello Corrêa

   Fernando Filgueiras is an associate professor at the School of Social Science, Federal University of Goiás (UFG). Professor of Doctorate in Public Policy at the National School of Public Administration (ENAP). Professor of Instituto Serzedello Corrêa at the Federal Court of Accounts. Affiliate faculty at Ostrom Workshop on Political Theory and Policy Analysis, Indiana University. Researcher at the National Institute of Science and Technology (INCT)–Quality of Government. Fellow of the National Council of Scientific and Technology Development (CNPq). Filgueiras has a Ph.D. in Political Science from the University Research Institute of Rio de Janeiro (Iuperj).

• PDF

There is no data available.

Copyright (c) 2026 Policy Making – Latin American Public Policy Review

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.