"In theory, Bayesian data analysis provides features that are of practical importance for scientific research. In practice, Bayesian theory, like any statistical theory, is underspecified for coping with the many modular decisions involved in any real-world project."  -- "Bayesian Workflow" (book forthcoming in June 2026)

Bayesian workflow is our attempt to close this gap and make applied Bayesian analysis tractable without abandoning what makes the theory worth using in the first place. In this seminar, I'll give a high-level overview of the components of a modern Bayesian workflow and address two recurring objections: that iterative model building is either not "really" Bayesian or that it inevitably leads into the garden of forking paths. I will say why I think both of these worries miss the mark.

I'll also discuss the motivation behind certain design choices in Stan and the broader Stan ecosystem that shape how applied Bayesians actually work, including some choices we regret. Decisions that seem trivial at the time to software developers can have large implications for a whole generation of Bayesian practitioners.