Here's a nit to pick:

Have you ever seen a box-and-whisker applied to a geometric distribution before?
It strikes me as a bit scandalous to use a B&W on an open data set.