Toward a Calculus of Confidence

Programmers, and end-user programmers in particular, often have difficulty evaluating software, data, and communication components for reuse in new software systems, which effectively reduces the value programmers derive from those components. End-user programmers are especially ill equipped to exercise the customary high-ceremony means of evaluating software quality. We seek effective ways to use low-ceremony sources of evidence, such as online reviews and reputation data, to make components’ quality attributes easier to establish, thereby facilitating more effective selection of components for reuse. Achieving this will require identifying sources of low-ceremony evidence, designing the meta-information required to track the differing sources and levels of credibility of various sources of evidence, and developing a method for combining pieces of disparate information into overall estimates of component value.