> https://www.evanmiller.org/how-not-to-sort-by-average-rating...

  phat = 1.0*pos/n

There is something disturbing about that line. While my wish to believe the article was strong, it makes me want to check the claims.