Implicating visualization in the rationalization of algorithms

Abstract

Once operationalized in practice, algorithms are generally understood as removed from human interpretability and control. In this thesis I examine interactive data visualizations and contend that, to the contrary, the interactive mediation of algorithms is increasingly implicated in their development and control. Accordingly, the field of visual analytics suggests that user perceptions of algorithms are highly informed by interactive mediations, whether or not algorithmic operations are visible to users. Accordingly, this thesis takes a publicly accessible dataset and uses interactive visualizations to support a theoretical examination of how interactive visualizations inform the perception and operationalization of algorithms.