Technology
The Politics of Algorithmic Decision-Making in Public Administration
Quick fact
Many governments use algorithms to determine welfare benefits, yet a 2023 study found that over half of the municipal algorithms surveyed in Western Europe were not publicly documented, making them essentially invisible to citizens.
Why this is interesting
Your eligibility for public benefits might be decided by a computer that you have never heard of. But who decided what that computer should value?
Read the full explanation
Understanding The Politics of Algorithmic Decision-Making in Public Administration
Picture a government office that used to employ case workers to review applications for benefits. Now, imagine a computer program that processes those applications automatically. The shift seems efficient, but it is not just a technical change—it is a political one. Every algorithm is built with human choices: what data to collect, which factors to weigh, what thresholds define eligibility, and even what constitutes an error. These choices reflect values and priorities. For example, an algorithm that flags suspected welfare fraud might be designed to be very sensitive to potential fraud, but that sensitivity could cause many legitimate recipients to be flagged and investigated. This is a political choice about where to place the balance between protecting the budget and protecting citizens' rights. In essence, algorithms in public administration are new arenas where politics happens, but without the public debate that typically surrounds policy decisions.
A deeper explanation
The core of the politics of algorithmic decision-making lies in the fact that these systems are not neutral tools—they encode value judgments. Developers and administrators make choices about how the algorithm weighs different factors, how it deals with uncertainty, and how it handles errors. For instance, when designing a child welfare risk scoring system, the choice of which variables to include (e.g., parental income, criminal history, number of prior calls) and how they are weighted can inadvertently embed racial or socioeconomic biases present in the historical data. Furthermore, these systems can create a 'black box' where even the agency itself cannot explain precisely why a decision was made. This raises profound questions of accountability: if an algorithm makes a wrong decision, who is responsible? The agency, the programmer, or the data? The political nature is also visible in the lack of transparency—many such algorithms are proprietary, protected as trade secrets, so even legislators cannot access them. This undermines fundamental democratic principles like due process and the right to contest a decision. Understanding this politics is crucial because it affects the fairness and legitimacy of government actions in the digital age.