> But we haven't been biased against A. [...] It was just their prior distribution that was different. This is why that isn't a counter example. Right at the b

> But we haven't been biased against A. [...] It was just their prior distribution that was different. This is why that isn't a counter example. Right at the beginning, the article clearly states that this method is only applicable if the prior distribution is equal: | You can use this technique whenever (a) [...], (b) [...], and (c) the groups of applicants you're looking at have roughly equal distribution of ability.