Reddit, Stumbleupon, Del.icio.us and Hacker News Algorithms Exposed

For example the HN algorithm is outlined in several places e.g. http://news.ycombinator.com/item?id=38704 and on the pg Arc site, and the reddit algorithms is

  def hot(ups, downs, date):
  s = score(ups, downs)
  order = log(max(abs(s), 1), 10)
  sign = 1 if s > 0 else -1 if s < 0 else 0
  seconds = epoch_seconds(date) - 1134028003
  return round(order + sign * seconds / 45000, 7)

Using a virtual clock, such as a tick count based on traffic (number of other submissions, votes, pageviews, etc.), could adjust for this.

At the very least, even reddit's blog post announcement of the code mentioned they're not open sourcing the spam detection code. Surely that's part of the ranking algo in that it determines if a submission gets ranked or not.

For the SU one, I think there is more to the story, but that's for another post...

it is just some variables that they tweaked until they fit.

Joshu, i think the del.icio.us formula is right, or they might just be using a different time constraint than 1 hour(maybe 2hrs or anything else). Because to findout what's popular they dont require any complex stuff. Simple math as specified in Danny's post is enough for their task.

Go Danny! Go!!!