In the course of Statistics for Data Analysis I give every spring to PhD students in Physics I spend some time discussing the apparently trivial problem of evaluating the significance of an excess of observed events N over expected background B. This is a quite common setup in many searches in Physics and Astrophysics: you have some detection