Machine learning has a reproducibility crisis

Machine learning is often characterized as much an "art" as a "science" and in at least one regard, that's true: its practitioners are prone to working under loosely controlled conditions, using training data that is being continuously tweaked with no versioning; modifying parameters during runs (because it takes too long to wait for the whole run before making changes); squashing bugs mid-run; these and other common practices mean that researchers often can't replicate their own results -- and virtually no one else can, either. Read the rest

Psychology's reproducibility crisis: why statisticians are publicly calling out social scientists

Princeton University psych prof Susan Fiske published an open letter denouncing the practice of using social media to call out statistical errors in psychology research, describing the people who do this as "terrorists" and arguing that this was toxic because of the structure of social science scholarship, having an outsized effect on careers. Read the rest