Painting pictures to augment advice

Abstract

I present an approach to designing decision support systems. The approach is to dissect a decision from both a normative and a cognitive perspective, and then to design a diagram that helps bridge the gap between the math and the mind. The resulting diagram is ultimately implemented as a visual interface in a support system. I apply the approach to two prototypical problems in "Command and Control" and highlight two practical principles that were used to guide the interface designs. One principle is that the system's interface should be informative, i.e., it should show users the underlying reasons for algorithmic results. The other principle is that the system's interface should be interactive, i.e., it should let users see and set the inputs that affect outputs. I discuss how interfaces designed by these principles can help users understand system recommendations and overcome system limitations.