Obviously (and understandably) the vote fraud detection algorithms and heuristics are not disclosed in order to keep the scripts effective.

However, to analyze a recent case of relatively significant reputation loss, it would be helpful to know whether I deduced the process mechanics from these posts correctly like so (confirming this hopefully shouldn't disclose anything that could be abused):

  1. Presumably the process is running fully automatic, i.e. both detecting the suspected fraud as well as reverting the votes involved is done by the script itself?
    • Or are the suspicious voting patterns only detected and do require a moderator to manually audit/approve/apply this?
  2. The votes are silently removed from the respective posts and nobody is notified, but the voter will be able to notice his votes are gone in case he revisits the respective posts and be able to recast the vote eventually therefore?
    • In case this is correct, will it only apply after a reputation recalc on the voters own account eventually (i.e. the silently removed votes will still be visible on the posts until after a recalc on his account)?
  3. There is no way for moderators to manually overrule these mechanics and decide that reputation is removed for the recipient but votes are left in place for the voter to prevent recasting?

1 Answer 1

  • Moderators cannot override, nor are they involved in the abuse script process.
  • The script does a rep recalc on every user affected for that run automatically once the votes are deleted.
  • It is fully automated and runs every 24 hours.
  • The votes are deleted. I.e. As far as the system is concerned they never existed. There is not track record kept of how many or when they were deleted.
  • 1
    are you sure they aren't logging any 'detected anomalies'? I mean at a DB or physical file level.
    – devinb
    Commented Aug 24, 2010 at 11:40
  • @devinb It could be a possibility, but from my understanding no they are not, however, Jeff and the team would be able to give a more accurate answer. Commented Aug 24, 2010 at 11:41
  • Thanks much, that's almost entirely conclusive already - just to nail down one last thing: they never existed implies that as long as I can see my vote on a post, my vote has not been considered fraudulent and still generates reputation for the recipient (eventually, depending on CW/cap etc.), right? Commented Aug 24, 2010 at 11:42
  • @Steffen Correct. Commented Aug 24, 2010 at 11:44

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .