Ben Green, assistant professor at the University of Michigan, presents a grounded framework for data scientists aiming to create algorithms that support social justice outcomes, drawing on real-world challenges encountered in urban planning efforts.
- Algorithms should be measured by societal impact, not just technical metrics.
- Practical barriers such as data availability and institutional capacity shape outcomes.
- Social justice-focused data science requires alignment with policy resources and constraints.
What happened
Ben Green, assistant professor of information and public policy at the University of Michigan, has authored Algorithmic Realism: Data Science Practices to Promote Social Justice, which proposes a method for data scientists to develop and evaluate algorithms through a social justice lens. His ideas stem from direct involvement in a Memphis urban policy project where he confronted multiple real-world issues that go beyond pure technical challenges.
During the Memphis project, Green encountered difficulties obtaining reliable data on property conditions essential for urban revitalization decisions. Moreover, he found that the local government lacked the resources and organizational capacity to maintain technical tools or implement the actionable recommendations generated by data models. This experience shaped his perspective on what it means to build algorithms that truly contribute to social good.
Why it matters
Green’s approach challenges the common tendency in data science to prioritize technical perfection, such as accuracy scores and fairness metrics, over practical effectiveness in the real world. By prioritizing how algorithms perform once deployed—accounting for institutional, political, and resource limitations—data scientists can better ensure their work leads to meaningful social improvements.
This perspective is especially critical for public policy and urban planning contexts, where the success of data-driven interventions depends not only on algorithmic design but also on the ability of organizations to act on insights. It highlights the necessity of interdisciplinary collaboration and continual evaluation beyond the development phase, ultimately advocating for a more realistic and justice-oriented view of algorithmic impact.
What to watch next
The widespread adoption of Green’s algorithmic realism framework could reshape how institutions approach data-driven social justice initiatives by emphasizing feasibility and ongoing practical impact over theoretical model performance. Observers should note how governments and organizations integrate these principles into their tech development and deployment strategies.
Additionally, as data science for social good matures globally, attention will likely increase on addressing structural barriers like data accessibility, funding, and political will. Monitoring how future projects manage these challenges and whether they sustain operational capacity after initial deployment will provide valuable insights for advancing this emerging ethics-driven field.