Values in the Stack: Challenges and Levers for Building Ethical Data Infrastructures

This lecture will not be live streamed. We apologize for any inconvenience.

An important focus of human-centered data science is encouraging developers who build data systems to reflect upon the ethical choices they make in their work. This talk describes a participant-observation study of the team developing Named Data Networking (NDN), an internet-scale data infrastructure. The study explored the values embedded in NDN, and discovered both challenges and facilitators for building ethical infrastructures. Findings revealed that infrastructural technologies challenge traditional methods for values-oriented design.

Many infrastructure design decisions have to be made before use cases, user needs, stakeholder values, and even technical protocols are well understood, complicating traditional approaches to value-oriented design. In addition, an even more fundamental barrier was a clash in perspective: values-oriented design assumes that technologies have politics and values. But network architects consider infrastructural neutrality fundamental to the technical success of internet-scale systems. Despite these obstacles, this study found that particular work practices — termed “values levers” — can engage values beyond neutrality, and help make ethics discussions relevant to the technical work of design.

Evidence from the study supports the larger project of empirical data ethics: improving data science by incorporating evidence-based understandings of how people and systems work together to produce, confront, and solve ethical challenges.