Academic honesty: where to draw the line?

Question

Originally, when I joined the site, I was under the impression that users users are supposed to indicate the sources of their answer. That is, if a given answer makes use of something developed in another answer, this should be acknowledged. In fact, the discussion under this thread appears to indicate that many users think so, too.

However, more recently I got more and more messages that essentially said

even if an answer is copied, it can still be a great answer.

So, according to this logic, there is no point in acknowledging the posts one got the inspirations from.

However, even more recently Joseph Wright wrote

However, when it comes to questions about the content of answers, life can become more difficult. Straight-up copy-pasting is easy to address, but where there are questions about the overlap of ideas, it can become more difficult.

This statement only makes sense if it is not irrelevant where the information came from.

And I got a private moderator message (concerning some sock puppets) saying

They've been clever enough to 'reference' the sources in each case, which might mean they don't get flagged.

Again, such a statement only makes sense if there is a notion of giving credit to others.

The purpose of this is only to clarify what the standards are.

1. Literally copying code seems not to be OK. Is that right?
2. If you rename a macro in a code from another answer, do you have to mention the original code?
3. If you copy the strategy of another post, do you have to mention it?

In case there is a certain form of expected behavior, I'd like to know what the consequences are if one does not meet these expectations.

Of course, I understand that sometimes it is hard to prove that someone "adapted" things from others. This question is not about ways to prove misbehavior (if there is any), it is more about what the general consensus is.

Answer 1

The StackOverflow model encourages re-use of posted material, with the CC-BY-SA license applying to all posts. As such, the issue comes down to what is reasonable re-use.

First, there's the question of whether re-used material is a significant part of a previous post. Of course, this is a judgement call and involves how much material is posted for example in a question: it's entirely reasonable when posting an answer that code from the related question can be re-used without attribution. I think we can broadly say that all of a shorter piece of code (say up to a dozen lines) or a similar-sized block from a larger post would be 'significant'.

There are two cases where such re-use would be problematic. The first is where no attribution is given: it's required by the license. Wholesale copying of an answer without any credit is of course likely to be easy to spot and pretty rare. The second is where re-use on the site doesn't involve some non-trivial change. Here, things like just renaming macros and making no other changes would be trivial. Adding to another answer's code ideally will credit it, and if there is a large amount of overlap I would anticipate action might be taken if that was not the case.

Life is more complicated when changes are more substantial: certainly from a moderation point of view, action is only likely when un-credited reuse is pretty clear. In particular, it's very hard to prevent similar implementations using the same approach, even where it might appear that one person has used the idea from a second answer.