1. prove that a regression technique that people are using to argue things are empirically true can be fit to anything
2. illustrate by spelling out HUMBUG as a plot, fitting it to that, and "empirically proving" things about it

http://jstor.org.sci-hub.cc/stable/1927538

video interview with the author explaining how he came to write that paper: https://www.youtube.com/watch?v=4BeWBy8gYHA