Features Yes. Reliability Maybe?

This article explores questions around whether and how software and systems’ dependability can be achieved within current development environments primarily focused on feature functionality. Despite decades of research and practice in reliability methods, industry experience demonstrates that typical projects are shown to deprioritize these approaches or simply have no familiarity with their usage. Agile methods and artificial intelligence (AI) technology are reviewed in this light. This discussion highlights that methodological popularity does not alone guarantee systems’ dependability or reliability. The adoption of sound engineering practices to specify and design for required quality attributes is presented as an effective path forward. By combining the best of both standard dependability practices and commonly promoted development models, improved results can be achieved. Recommendations for refocused research and education in this area are also presented.