The only thing this experiment could show is that if the ratings changed, but the writers didn't, then the writers are not responsible. It can't show the reverse.

Everything in the article is just properties of arithmatic averages in sorted sequences.