I did normalize by number of favorites too, so each person had 1 vote shared over all their favorites. The idea that a person who only had one favourite was a r

I did normalize by number of favorites too, so each person had 1 vote shared over all their favorites. The idea that a person who only had one favourite was a really strong signal for that single link, whereas people who have lots of favorites are using it differently to what I want to select for. At some point I had a similar selection of Ask HN dominance. One problem with my data is I only grabbed the first page of favorite results. EDIT: SHOOT, you exposed a bug, my parser skips things things without URLs! Hence I ended up filtering out most of the Ask HN's! EDIT: Fixed it Now my list looks a bit more similar to yours but not quite. Currently "Ask HN: What books changed the way you think about almost everything?", your #1 is my #50. If I turn off normalization, it goes to #13 on my lists, so that has quite a significant impact on the rankings. sidenote: Lol this post is the 20th most favorited article now.