I can't tell for sure, but it doesn't seem like this analysis normalizes for total submissions in each category. For example, if there are very few Ask HN submi

I can't tell for sure, but it doesn't seem like this analysis normalizes for total submissions in each category. For example, if there are very few Ask HN submissions, but all of them make the top 3%, then his charts would say that Ask HN is not very popular. What you really want to know is whether posts with a particular attribute reach the top out of proportion with their representation in the total population.