An Evolutionary Theory-systems Approach to a Science of the Ilities

Abstract For system engineers to effectively document requirements for non-functional properties (ilities), to reason about tradeoffs, and to implement and verify such properties, the language used in these endeavors must be precise enough to support rigorous engineering activities. Yet the language used today is often ambiguous and imprecise. Moreover, many past attempts to improve it with natural language definitions and explanations have not converged. We propose an embedded theory-systems (ETS) alternative approach. It replaces natural language with theories (models) in an expressive formal language; mechanically derives software from these models to foster community engagement with the theories; and uses feedback based on interactions with the software to drive theory evolution and validation. We hypothesize that this approach can accelerate convergence on models that are precise and validated enough for rigorous systems engineering. We present an early case study on applying this method to the Ross et al. semantic approach to defining change-related ility terms. Results include a clarifying formalization of their informal model, its evolution through four stages of feedback, insights into key remaining shortcomings, and evidence that the approach can promote engagement with theories in ways that drive convergence toward shared, precise, useful language for engineering system ilities.