Evaluating a Corpus of Root Causes and Measures to Guide RCA Processes in Critical Software

Reducing defects in critical systems is a challenge faced every day by industry experts and researchers in the dependability and quality assurance fields. However, the state of the art and practice of Root Cause Analysis (RCA) is largely ad hoc. In this paper we develop a corpus of root causes that are evaluated across a broad survey in the industry and academia. The results provide several key insights including: 1) a broad schema of root causes and actions, leveraging our prior work on real defects in critical systems: 2) understanding on the relevance and value of potential root causes, measures and the RCA process as seen by industry experts across aerospace, safety and academia, 3) detailed insights into the different causes, development measures and their acceptance and importance, 4) survey results providing an invaluable input baseline for further analytical work, and 5) recommendation for an RCA process and data collection schema that could be more widely used and shared across the critical software community.