Jeff and I talk about the recent Reinhart-Rogoff reproducibility kerfuffle and how it turns out that data analysis is really hard no matter how big the dataset.