I noticed that a few pieces of the algorithm were not released, and the algorithm has surely changed since arc3.1, but the published code explains at least the overall behavior of Hacker News ranking.