Prof Philip Dawid "Decision theory good, causality bad"

I argue that much of the enterprise commonly pursued under the banner of "statistical causality" is better understood as a branch of statistical decision theory. As soon we realise this, and formulate our queries and our methodology accordingly, we avoid being sidetracked into irrelevant metaphysical dead ends, and can concentrate on the questions of real interest. Examples include making inferences from observational longitudinal data about the effects of dynamic treatment strategies.