Hacker News, like Reddit, is a community-powered news site where users can submit links they find noteworthy. By default, new submissions go under the ‘new’ tab, but some get promoted to the much more visible ‘news’ tab (the front page), generating a substantial amount of traffic to promoted sites. Submissions under the ‘new’ tab also have a ‘no follow’ tag:
Promoted stories lose the tag, in addition to the traffic, help for SEO purposes.
Given these benefits to having links promoted to the front page, many theories have arisen has to how the ranking process works:
However, ranking is not the same as promotion. In order to rank on the front page, it would seem you must first get the link promoted from the ‘new’ tab to the ‘news’ tab.
A common theme is that there is an algorithm that regulates the rankings. However, the front page isn’t that ‘big’, with only thirty entries, whereas the number of submissions is vastly greater. At peak hours, there is a new submission every 20-30 seconds. If sorting were by votes, there would be much more activity under the ‘news’ tab, but there isn’t, and it isn’t uncommon for promoted stories to stay up on the front page for many hours. Thus algorithms, are only a part of the picture, perhaps a very small part of it.
So what is the missing ingredient to promotion?
What prompted this question was seeing two entries (#25 and #26) on the ‘news’ tab that have conspicuously few votes relative to the others:
These two stories were submitted two hours before the screenshot.
Also, on the ‘new’ tab there are two entries (#34 and #44) that have five votes and are two hours old:
So if #25 and #26, with five and three points and are two hours old, can get to the front page, shouldn’t #34 and #44, which both have five points each and are also two hours old?
However, neither of them are visible on the front page:
But one of them (#44) makes it to #62 on the front page, but being so far down it won’t generate much traffic:
But then,20 minutes later, why is #22, which only has three votes, already on the front page:
Meanwhile, entry (#34) about ‘data science for math/phys background’ is still stuck on the ‘new’ tab despite getting six votes:
This seems inconsistent with an impartial ‘vote-based’ algorithm. Obviously, not all votes are created equal. In fact, I would wager that most votes don’t count at all. By my estimation, there are a handful of accounts that have sufficient privileges to promote links from ‘new’ to ‘news’. These votes count and the rest are ignored or discounted substantially. I imagine there are also votes that can demote links off the front page, which could be why some entries vanish. The problem with vote counting algorithms is that they can be gamed by spammers using new accounts to pad vote counts, but seniority is much harder to fabricate. Ken Shirriff offers additional commentary about penalties in the ranking algorithm.