Using conditional promises, the reliability of promises can be measured without considering the reliability of the whole agent, by defining notions of when conditions operate and when promises are expected to be fulfilled. This inspires an analytical method based upon “incomplete conditions” that attributes failures to incomplete knowledge rather than operational faults. This analysis allows agents to choose offers based upon statistical measures of the completeness of stated conditions.
Alva L. Couch, Hengky Susanto, Marc Chiarini