Abstract

Disasters in sociotechnical systems are caused by many factors. For example, the tragic chemical disaster in Bhopal, India, was caused by a leaky valve and a maintenance error. Safety scientists and investigators call these proximate factors. But the tragedy was also caused by what experts call systemic factors, which include the facility’s design, the organization of its workforce, and the safety culture of the plant and surrounding community. Other disasters share a similarly complex causality, depending on a multitude of proximate and systemic factors. Investigators and scientists distinguish between these different types of factors because they think each has distinctive causal features and also because they often think some factors are more significant causes of disasters than others. In this dissertation, I analyze these distinctions and the reasoning experts use when selecting important causes of disasters in cases like the Bhopal gas tragedy and the Challenger shuttle loss. Using James Woodward’s interventionist theory of causation, I explicate three causal concepts that can be used to distinguish among causal factors: amplifying/damping, causal inertia, and causal delay. I employ these concepts to elaborate and clarify the causal features safety scientists identify to draw distinctions among different types of causal factors, and their reasons for thinking some causes are more important than others for understanding disasters. My dissertation takes a pragmatic approach to causal selection. Although philosophers have emphasized John Stuart Mill’s remark that selecting individual causal factors is “capricious,” I base my approach on Mill’s deeper thinking about the logic of causal selection and the philosophical challenge it poses. I show that selections of causal factors in investigations of disasters and the surrounding discussions of safety are motivated primarily by the dual aim of preventing future disasters and making systems safer. Considering these purposes is essential to understand why investigators single out causes with certain features. Analyzing causal selection in terms of why certain types of causes with certain kinds of features are more useful than others for particular purposes clarifies vital reasoning about the important causes of disasters and enriches our understanding of causal knowledge.