I added a small contribution to the density plot debate to the analysis of my crossword solving times: https://jdonland.github.io/posts/crossword_times/crossword_times.html

tl;dr: LOESS-based density plots like the ones produced by `geom_violin` have some problems that are ultimately the result of neglecting to specify and fit your own model.

#geom_violin #loess #bayesian #crosswords