I noticed that questions often have similar or duplicate answers, i.e. one answer is contained within another. As far as I can tell, there is currently no structured approach to cleaning up such mess.

There are many possible use case scenarios here, but I think that a user normally wants to compare answers in only two ways:

  1. Answer A is contained within answer B. (positive)
  2. Answer A plagiarized from answer B. (negative)

Option 1 acknowledges writer(s) of A in B. If enough users mark this relationship, answer A is removed, and writer(s) of A are treated as co-authors of B.

Option 2 states that the author(s) of A simply copied B (with possible minor modifications) in order to steal votes. If enough users mark this relationship, answer A is removed, and writer(s) of A lose some reputation. This would prevent problems like this.

In both cases, the algorithm involved should check the timestamps on the answers, and possibly also the diff-s.

EDIT (added due to Tom Wijsman's response):

Because, in the end, it's about the readers, not the authors, and the readers don't want to read 5 similar answers, but rather 1 answer that concisely includes information from them all --- if many readers think an answer is contained within another, then their opinions should be respected. My proposal strives towards this goal, while also trying to make it fair from the authors' perspective.



Please consider option 3, where both users do their best to try to get the better answer. From the reader's viewpoint, that yields a answer that is of a higher quality, more pleasant to read and useful.

Most answers that are posted can always use some form of improvement, whether it be explaining a piece of code, explaining what a word of a specific jargon means or even improvements like better content/formatting.

For an example, I just take the current hottest network-wide question. It asks how to shorten repetitive code. We tend to see that the most upvoted answers go into explanation, while the least upvoted answers only have some code without any explanation around it. Note how the highest voted answer explicitly edits his answer, to explain why his answer is the right choice in this situation.

An answer of sufficient quality is rarely a duplicate because people tend to explain things in different ways; deleting these disallows them from improving their answer further and disallows them from gaining reputation for their hard (but slightly duplicate) work.

I can't really tell something from your example since the evidence was removed, but looking at the top answer pointed there I can tell that it can still be improved; by educating the user that such information can be found by searching/reading documentation, or why a regular expression is unnecessary...

In any case; if it is an obvious intended copy that includes detailed explanation, then downvote / delete.

  • Both options I wrote about are compatible with the fact that the authors want to get a better answer. I edited my question to make my underlying philosophical point clearer. – Rok Strniša Nov 26 '11 at 15:07
  • @Darthenius: Do you have a concrete example? Other than those where the answer is outright removed? – Tamara Wijsman Nov 26 '11 at 15:08
  • I have now added three examples to the question. – Rok Strniša Nov 26 '11 at 15:44
  • @Darthenius: In example 1, your answer has an explanation while the other answer is just code, that would result in a loss of quality. In example 2, the code of the original answer has been copied minutes after so they should be deleted like in my last sentence. In example 3, the OP is just trying to ride some reputation by posting his actual solution that's just a small adjustment to your code; this last thing happens a lot and I would say that it is a problem by itself. Example 2 can be solved with a "low quality" flag or message that says it's a duplicate, example 3 with "not an answer". – Tamara Wijsman Nov 26 '11 at 16:07
  • Your arguments don't hold, since the readers matter the most, and it is the readers who have the opinion of when one answer contains another. The approach would improve self-moderation of the site, which is what SO is good at and successful for. – Rok Strniša Nov 26 '11 at 18:45
  • @Darthenius: It's not readers that matters, but quality. Time is better spent on improving content than moderating content. Note how duplicates are a good thing, they help people improve their post (FGITW) as well as readers find there things (in the case of questions) as well as the problem explained in a different way (answers). We have a voting system to bring the most working solutions towards the top. Having similar answers on the page is really not a problem; exact duplicates can be deleted and OP's self-answering dupes is a different problem... – Tamara Wijsman Nov 26 '11 at 19:06

The problem you describe isn't really a problem and its solution is mighty complicated and not really needed. Yes it would be nice to always have one canonical correct answer, but it doesn't always happen that way.

There is no harm in multiple correct answers because this can provide an affirmation to the OP of the right thing to do to solve his problem. i.e. multiple users posting similar answers help to provide a consensus.

Yes it would be nice to see some collaboration to merge partially correct solutions into a single mega answer, or for users to notice the answer the provided has already been suggested and withdraw it, but it's not enforceable.

I see no problems with the answers on any of those examples such that they breach Stack Overflow's etiquette and code of conduct.

We will as moderators remove answers that are blatant plagiarism, but there has to be very good evidence to support this.

With regard to Tom's comments about flagging these, don't waste your flags because these would be declined. Whilst there are multiple similar answers, they are answers all the same. Moderators are not arbiters of which correct (or wrong) similar answer gets to live, especially in FGITW scenarios such as this.

It's up to you the community to use your votes to do this, not by flagging.

There is one scenario where we might consider acting on an exact duplicate answer. This would be where an old question has an accepted answer and perhaps multiple similar answers provided at around the same time. We sometimes see low quality answers provided months or years afterwards that are exact duplicates that don't add any new value. If we think it's of no additional value then we may remove that post and leave a comment as to why. But these are exceptional circumstances.

  • I don't think it's that complicated, especially for the SO crew. Also, I disagree with your argument that having multiple identical/similar answers confirms to the reader what is the right thing to do more than a single answer with combined votes them all. – Rok Strniša Nov 26 '11 at 18:41
  • @Kev: In my experience moderators do handle on flags in the case of example 2. I agree that it would be useless on example 3; as I said, it's a problem on it's own... – Tamara Wijsman Nov 26 '11 at 19:00

You must log in to answer this question.

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