iSchools Book Talk
Algorithmic Realism
presented by Ben Green
The Book
Algorithmic Realism provides bold new vision for data science—and a practical, step-by-step guide for how data scientists can contribute to social justice.
Despite the desires of data scientists to develop algorithms for good, algorithms regularly produce injustice in practice. Given these persistent harms, the field must reflect on difficult questions about its identity and future. Can data science be a force for promoting social justice in the world? What practices should data scientists follow to achieve this goal?
In Algorithmic Realism, Ben Green presents a bold and interdisciplinary approach to data science. Drawing on his experience practicing data science in the public interest, he argues that improving society with algorithms requires transforming data science from a formalist methodology focused on mathematical models into a practical methodology focused on addressing real-world problems. By providing an expanded framework for the “data science workflow”—the steps that characterize the algorithm-development process—he offers a practical, step-by-step guide describing how data scientists can apply their skills in service of social justice. Through these contributions, the book reveals a vision for a renewed, but realistic, optimism about data science’s potential to foster a more equitable world.
The Author
Ben Green is an Assistant Professor of Information and Public Policy at the University of Michigan and a Faculty Associate at the Berkman Klein Center for Internet & Society at Harvard. Through his research, writing, and advocacy, Ben contributes practical lessons for achieving a world in which AI improves society rather than exacerbating inequality and oppression. He has published op-eds in outlets including The New York Times, The Wall Street Journal, and The Hill, and has been quoted in newspapers such as The New York Times, The Washington Post, and the Los Angeles Times. He is the author of two books from the MIT Press: Algorithmic Realism: Data Science Practices to Promote Social Justice(2026) and The Smart Enough City: Putting Technology in Its Place to Reclaim Our Urban Future(2019).
What makes this book special?
Algorithmic Realism presents a comprehensive methodology for practicing data science in the public interest. For years, researchers have called for data scientists to act more ethically, think more about power, and work more closely with communities. Many data scientists are receptive to these calls but are unsure what to actually do differently. My book provides a bridge between critical scholarship and data science practice, showing how data scientists can promote social justice in rigorous and practical ways.
I present this new methodology of “algorithmic realism” through an expanded version of the data science workflow. Rather than focusing just on the technical steps of modeling data, I situate that work within a larger process that spans five stages: problem scoping, project scoping, data analysis, implementation, and evaluation. The book goes through these stages in step-by-step detail, with principles, practices, and case studies for each.
Why should people read this book?
People should read this book to gain a new conceptual frame for thinking about data science. Underpinning its practical guidance, Algorithmic Realism offers a broad reimagination of what data science should be as a field and how to conceive of the purpose of algorithms. At the heart of algorithmic realism is the “pragmatic maxim for data science”: The ultimate test of an algorithm’s attributes is the impacts it generates in practice. In other words, algorithmic realism treats algorithms as instrumental tools. What matters is not an algorithm’s technical attributes, but the algorithm’s role and impacts within the particular social setting of its use.
This view of algorithms can shake data science from set ways of conceiving itself and the obstacles it encounters. In cases such as algorithmic fairness and AI explanations, data science has become stuck, pursuing narrow formalist approaches that fail to produce ethical and useful tools. Many in the field perceive intractable dilemmas, such as the incompatibility of different fairness definitions, that leave the field stuck with unsatisfactory solutions. Algorithmic realism gets data science unstuck—not by revealing novel ways to optimize algorithms, but by revealing novel ways to envision how algorithms can tackle injustice and encourage equitable decision-making.
What will people learn from this book?
Here are some of the lessons that readers will gain:
Why “algorithmic formalism” is a flawed approach to developing algorithms.
How pragmatist philosophy can improve data science.
How data scientists can take a solution-agnostic approach to building algorithms.
Why fairness metrics fail to advance social justice, and what data scientists should do instead to challenge injustice and discrimination.
Why explanations don’t improve human-AI collaborations, and strategies to design algorithms that augment human decision-making.
How data scientists can future-proof their career prospects from AI replacement by becoming “data strategists.”
Initial steps that data scientists can take to embrace more interdisciplinary and justice-oriented projects.
iSchools Book Talk is a new iSchools feature series spotlighting books that members of iSchools would recommend to the iSchools community. These books can be presented either by the authors themselves or by any other iSchools member.
Please contact admin@ischools-inc.org in case you would like to feature a book as well.
Comments