Algorithmic governance in public organizations: socio-technical reconfiguration, paradoxical tensions, and capability development in four ecosystems

Purpose This study examines how algorithmic governance reshapes workflows, decision rights, accountability boundaries, and professional roles within public organizations. It investigates the socio-technical mechanisms through which algorithmic systems reconfigure organizational processes, generate concurrent bright and dark outcomes, and support the emergence of dynamic capabilities for adaptive governance. Design/methodology/approach The research adopts an exploratory multiple-case design involving four Italian digital ecosystems: a national chatbot initiative, a civic digital twin, a healthcare artificial intelligence platform, and an educational ecosystem. Data were collected through 20 semi-structured interviews and more than 40 documentary sources, and analyzed abductively following the Gioia methodology. The coding combined inductive analysis with theoretical sensitization to socio-technical systems, paradox theory, and dynamic capabilities. Findings Algorithmic governance emerges as a process of socio-technical reconfiguration that redistributes decision rights, reshapes accountability boundaries and introduces new hybrid roles. This process produces simultaneous bright outcomes (efficiency, transparency, predictive insight, learning) and dark outcomes (role ambiguity, dependency, accountability gaps). Their coexistence varies according to governance clarity, data and algorithmic maturity, and participatory practices. Over time, organizations develop interpretive, integrative, and adaptive reconfiguration capabilities that enable them to transform tensions into structured learning and enhance organizational resilience. Practical implications The study provides actionable guidance for policymakers and managers to design clear accountability frameworks, strengthen hybrid analytical-professional skills, and institutionalize reflexive evaluation routines that support responsible algorithmic governance. Originality/value The paper advances theoretical understanding of algorithmic governance by integrating socio-technical, paradox, and dynamic-capability perspectives, and by identifying the mechanisms through which public organizations learn and adapt within evolving digital ecosystems.