This QA site doesn't quite take the form that others do. Here, we have fairly specific questions. But I already see two Project Euler #1 questions that are similar. Are these duplicates?
2\$\begingroup\$ Well, the second question (codereview.stackexchange.com/questions/7/…) asks nothing about the algorithm, more of structure/good practise? \$\endgroup\$– ZolomonJan 19, 2011 at 21:25
\$\begingroup\$ What if there were tags associated with a code snippet so that questions could be grouped together? e.g. recursion, dynamic-programming, etc. That might help figure out if two questions are similar. \$\endgroup\$– David WeiserJan 19, 2011 at 22:48
\$\begingroup\$ I was about to post a duplicate question of this one. I think this problem might come up a lot in searches and in the feature which pops up possibly duplicate questions as users make their posts. There would likely be a flood of "how do I make this better" title posts, with too many hits based off specific language keywords or functions in code excerpts. IE: information overload. \$\endgroup\$– bob-the-destroyerJan 30, 2011 at 2:26
1\$\begingroup\$ This is an interesting problem in the sense of an unsolved challenge. Is it a problem in the sense of a current difficulty that has a demonstrable negative effect on the site? \$\endgroup\$– itsbruceDec 10, 2014 at 1:17
Four years later, we now have over a dozen questions on Project Euler #1, including multiple implementations in the more popular languages. We also have hundreds of linked-list questions, many of which solve the same problem.
Considering the rule that Code Review answers can cover any or all facets of the code, the threshold for considering two questions as duplicates should be set quite high. I propose the following criterion:
Are two questions similar enough that a course instructor would accuse a student of copying the solution?
That is, are the differences all trivial ones, such as tweaking whitespace and renaming identifiers? If so, we can reasonably say that the question has already been answered.
In practice, duplicates are likely to occur only when
- a user posts the same question more than once due to confusion (in which case we can just delete one)
- plagiarism has occurred (in which we can also close the question as Not Your Own Code)
- a cross-posted question is migrated to Code Review
- a user posts a follow-up question after making changes that are entirely trivial
I think we're going to have define duplicates based on an answer/response instead of a question (Like Stack Overflow).
We're going to have lots of very similar questions if people are posting code for feedback. But we can say that it is a duplicate if your typical response would be "use a specific pattern for this problem".