Evidence-Based Software Processes

Many software projects fail because they commit to a set of plans and specifications with little evidence that if these are used on the project, they will lead to a feasible system being developed within the project’s budget and schedule. An effective way to avoid this is to make the evidence of feasibility a first-class developer deliverable that is reviewed by independent experts and key decision milestones: shortfalls in evidence are risks to be considered in going forward. This further implies that the developer will create and follow processes for evidence development. This paper provides processes for developing and reviewing feasibility evidence, and for using risk to determine how to proceed at major milestones. It also provides quantitative result on ”how much investment in evidence is enough,” as a function of the project’s size, criticality,