Administrative Discretion versus Algorithmic Governance in Smart City Policies

• In Event: Digital Transformation in Government:  Policy Diffusion, Administrative Capacity, and Discretion

Abstract

Three successive waves of digital transformation have progressively reconfigured the exercise of administrative discretion in smart cities. The first, electronic government, digitized back-office operations. The second, smart government, introduced 24/7 online services to citizens. The third wave is algorithmic governance, in which artificial intelligence, predictive analytics, and optimization algorithms now allocate resources, enforce regulations, and trigger sanctions with minimal human involvement. Algorithmic governance shifts discretion from visible frontline bureaucrats toward invisible system designers, data architects, and model developers—fundamentally altering the locus of administrative power in smart city policies.

This paper examines the emerging tension between administrative discretion and algorithmic governance across four policy domains: dynamic mobility pricing and management, automated regulatory enforcement, predictive policing, and public service deployment, drawing on comparative cases from the United States and Europe. We ground our analysis in the two classical narratives of bureaucratic discretion identified by Maynard-Moody and Musheno (2000): bureaucrats as state agents, who enforce rules consistently and impartially, and as citizen agents, who exercise moral judgment responsive to individual circumstances. While AI systems can effectively automate the state agent as a consistent policy enforcer, the citizen agent resists algorithmic substitution since it is context-dependent, relational, and ethically complex.

To resolve this tension, we propose a framework organized along a spectrum of discretion. At one end, low-risk, routine operations—parking management, traffic signal timing, and utility meter readings—are well-suited to algorithmic governance. At the other end, high-stakes, value-laden decisions—evictions, intensive policing, emergency triage, child protection—require that human bureaucrats retain meaningful authority to decide and justify outcomes. Between these poles lies a middle range where algorithms may recommend, and humans adjust policies. Mapping smart city policies onto this spectrum enables more principled governance design than blanket automation or reflexive resistance.

We introduce two institutional concepts to operationalize this framework. Discretion-aware algorithms are systems deliberately designed to leave room for context, exception, and human override in sensitive domains, incorporating structured review points that preserve accountability. Algorithm-aware administrations are organizations that invest in digital and ethical literacy for public servants, involve street-level bureaucrats and citizens in system design, and embed algorithmic tools within clear legal and organizational accountability chains. Together, these concepts chart a path beyond the false dichotomy of automation versus discretion, providing a pathway toward a nuanced integration that serves both efficiency and democratic legitimacy.

The central argument of the paper is that the future of smart city governance depends not on algorithmic sophistication alone, but on whether meaningful, accountable discretion is preserved where normatively required. This is fundamentally a governance and institutional design challenge, not a technological one. The paper provides a set of design principles for smart city policymakers. We also outline possible avenues for interdisciplinary research that would integrate insights from public administration, computer science, law, and public policy and develop frameworks in which smart city algorithms enhance efficiency without undermining legality, fairness, rights, and democratic trust.

Maynard-Moody, S., & Musheno, M. (2000). State Agent or Citizen Agent: Two Narratives of Discretion. Journal of Public Administration Research and Theory, 10(2), 329–358.

Author

• Milena I Neshkova, Florida International University
  Presenting Author