Towards Efficacy Hypotheses for Safety Cases

Abstract

Safety cases have been produced, reviewed, and written about for decades. Experts point to growing popularity and opine that their use helps to reduce major accident risk, but their history includes noteworthy accidents. Given the potential consequences of regulatory failure, it is crucial that regulatory practice be evidence-based. However, testable hypotheses about the efficacy of safety cases are rarely given, let alone supported by compelling empirical evidence. Moreover, different authors have used the term 'safety case' to mean different things. To help put safety argument practice on a sounder evidential footing, we conducted an analysis to identify potential efficacy hypotheses for future study. Our analysis considers the kinds of value arguments might bring, forms of safety case and safety argument, stakeholders, and plausible alternatives serving the same purposes. In this paper, we present our analysis and findings and discuss potential research directions.