I propose to detect double postings automatically by the system and perform a minor penalty like a 24h ban of posting content.


There are numerous cases in which users duplicate their own question to get it back into the latest questions feed. This eats up considerable amounts of moderation time and is a clear violation of etiquette. As with the implementation of the quality check before posting, a temporary ban of offending users would encourage to use the encouraged way of posting bounties to add attention to a post.

Logic Implementation

As duplicate detection is a very resource intensive task, the system would have to work very localized. The detection time frame could be limited to a few days - the exact amount could be determined by evaluating already existing duplicate closes, checking these questions for real exact duplicates (not only by meaning but by Levenshtein distance and performing a statistical analysis of the time, in which 80% of exact duplicates occur (from my current SO experience, I'd say ~72h).

Most duplicates are also posted by the same user, as most users do not go through the hassle of registering a new account for reposting the question, so we could also limit the search to posts by the same user. This lowers the amount of questions to search through to a few at maximum.

The algorithm to use needs more evaluation, as there are several working examples out there. The easiest implementation would be a Levenshtein distance, but it is not the most efficient one.

UI Implementation

As we want to positively encourage users to stick with the rules, the offending user should be warned once with showing the consequence of a temporary ban before the question itself is posted, similar to the quality warning of very short posts.

If the question is posted nevertheless and a duplicate is detected, the system should automatically tag the question as reposting for moderation and set a low vote count (around 2 instead of 5) to close it. If the reposting-flag is acknowledged by two positive votes, the new user should be informed and banned for 24h from posting and the reposting should be merged or deleted with the original.

  • 4
    Are you talking about duplicates on the same site, or cross posts as well? I think people copy pasting their question in multiple sites is far more common than re-posting it on the same site, if you didn't have cross posts in mind you probably should include them.
    – yannis
    Nov 2, 2012 at 11:20
  • As well as posting duplicate answers. As a (very) naive implementation, the system could create hashes of their posts on submit, and just store hashes of the posts they've made. If anything matches up exactly, then you know it's a duplicate. I don't see that being really that resource intensive, unless like Jon Skeet, you have thousands and thousands of answers. I don't think they should be banned for doing it, just don't allow it. Nov 2, 2012 at 13:18
  • 1
    @George_Stocker: The hash system is not very secure and very errorprune if used for duplicate detection. We should research more in the field of plagiarism detection instead of exact matching.
    – Lars
    Nov 2, 2012 at 13:21
  • @YannisRizos That is also true - maybe the Developers can shed some light about the possibilities of implementing a cross-site feature.
    – Lars
    Nov 2, 2012 at 13:21
  • 1
    @Lars errorprune was a funny freudian slip. Or was it deliberate?
    – itsbruce
    Nov 2, 2012 at 14:09
  • @Lars The way I see it, those are two separate things. 1) A user posts his exact same answer to multiple questions. The Hash mechanism catches that nicely. 2) a user posts an answer that was copied from another user. That's tricky, and I'm not even sure the cure would be better than the disease. Nov 2, 2012 at 14:26
  • @itsbruce I'm sorry, I am no native speaker. I meant error prone - no pun intended. =)
    – Lars
    Nov 2, 2012 at 16:26
  • @GeorgeStocker No, because by changing one character, the hash mechanism will fail. Plagiarism detection (also plagiarising yourself) is a different field with other requirements and doesn't play well with hashes.
    – Lars
    Nov 2, 2012 at 16:28
  • @Lars We have a problem now with people posting exact duplicate answers to multiple questions. That's the problem I'm trying to address. Nov 2, 2012 at 16:43


You must log in to answer this question.

Browse other questions tagged .