I remember when this was first released, everyone clowned on it (including me) for a few reasons the biggest one being the classic problem with these sorts of Bayesian analyses. That being, the individual weights per piece of evidence are pretty arbitrary.
Meanwhile, it seems quite likely to have produced the correct results so ¯\_(ツ)_/¯
Meanwhile, it seems quite likely to have produced the correct results so ¯\_(ツ)_/¯