Algorithms and Decision-Making in the Public Sector

Article Source: Annual Review of Law and Social Science, Vol. 17, pp. 319-334, 2021
Publication Date:
Time to Read: 2 minute read
Written By:


Kyla Chasalow

 Sarah Riley

Sarah Riley



This article provides a road map for the study of algorithmic systems used by local governments in the United States, including issues relating to procurement, bias, transparency, and regulation.


Policy Relevance:

The design of algorithms used by government should support accountability.


Key Takeaways:
  • Increasingly, local governments within the United States use algorithmic systems to make decisions relating to criminal justice, benefits, and education; "algorithm" refers to technologies that use machine learning or programmed rules to inform or execute government actions.
  • Municipalities are the leading provider of public services in the United States; vulnerable constituents, including people of color, are disproportionately affected by municipal algorithms.
  • Algorithmic tools are increasingly central to local government.
    • Local governments struggle to raise revenues, and algorithmic tools promise efficiency.
    • Local government services are highly visible and involve critical decisions.
  • The availability of algorithms may focus attention only on aspects of the problem that algorithms can solve; for example, attention might shift from addressing the root causes of poverty to screening ineligible citizens out of benefit applications.
  • The integration of data collected by different entities (such as education and law enforcement) can improve efficiency and access to services, but raises the prospect of far-reaching surveillance.
  • Governments often acquire algorithmic tools from commercial vendors, leaving it unclear which entity is responsible for bad outcomes; systems that affect rights, values, and social outcomes should be subjected to close scrutiny in the procurement process.
  • Transparency and accountability of algorithmic systems may be proactive (by design, as required by law) or reactive (a result of information requests after a problem has developed).
    • Algorithmic impact assessments call for analysis before a system is developed.
    • Citizen groups support accountability and oversight.
  • Government workers may have discretion to accept or reject an algorithms’ output, either correcting or introducing bias; workers resistant to algorithmic decisions may ignore recommendations or obfuscate data.
  • Algorithmic systems need to be repeatedly evaluated, to ensure that bias or a shift in function over time is detected and addressed.



 Karen Levy

About Karen Levy

Karen Levy is an Associate Professor in the Department of Information Science at Cornell University, associate member of the faculty of Cornell Law School, and field faculty in Sociology, Science and Technology Studies, Media Studies, and Data Science. Professor Levy researches the legal, organizational, social, and ethical aspects of data-intensive technologies. Her work explores what happens when we use digital technologies to enforce rules and make decisions about people, particularly in contexts marked by conditions of inequality.