The Case of the Strangerationist
Vera D. Khovanskaya, Maria Bezaitis, Phoebe Sengers · 2016
We describe a method for critically informed development of new technical systems by combining analysis of historical discourse with critical technical practice. We take the case of social recommender systems, a class of algorithms that calculate which people should be recommended to whom. We demonstrate similarities between limitations of "social network" rhetoric in contemporary social matching algorithms and discourse on planning in Artificial Intelligence. We develop an algorithm for social matching that recombines "lost" ideas from the history of AI, orienting around situated behavior and algorithmic transparency. By implementing this approach in a functioning prototype called "the Strangerationist", we examine directly how conceptual commitments inform low-level technical decisions, and how available technologies shape conceptual vision. Our goal is not to design a "better" algorithm but to explore the challenges and opportunities of weaving together historical discourse and critical analysis of values embedded in technology with the experience of designing it.