Trusted Software's Holy Grail

In this paper, I discuss what I believe is the grand challenge facing the software quality research community: the ability to accurately determine, in the very earliest stages of development, the techniques that will be needed to achieve desired levels of non-functional attributes such as: reliability, availability, fault tolerance, testability, maintainability, performance, software safety, and software security. I will further consider the associated technical and economic tradeoffs that must be made in order to: (1) achieve these desired qualities, and (2) to certify that these qualities will be exhibited when the software is deployed. And I will also take into account the fact that satisfying a particular level of each attribute requires specific expenditures, some of these attributes conflict with each other, and when the environment or usage profile of the software is modified, all guarantees or claims of quality should be viewed suspiciously until additional evidence is provided.