Etiquette on more than one correct answer

Question

Here's an etiquette question where I feel that my default MO behaviour might lead me astray.

There are two answers to this question. Mine, and caramdir's. I think it's fairly indisputable that mine is the better answer (at least, now that I've edited it to contain the explanation). But caramdir's adds a small, but significant, modification that I overlooked. It's actually not directly referred to in the question, but the mathematician in me screams out that the true answer ought to include caramdir's addition.

As it stands, both have gotten about the same number of votes. My gut feeling is that my answer ought to be accepted (assuming that an answer is to be accepted) and there the matter should stand. My reasoning being that someone reading the answers should read all the answers and notice caramdir's addition. However, I have a feeling that the SO route is that the accepted answer should be complete and that it should not be necessary to read the other answers (though of course it's a good idea).

Is that correct?

If so, what is the correct etiquette here? I could incorporate caramdir's addition (with attribution) in to my own with the simple addition of a '%' (it appears that the debate as to whether "blank comment lines" are evil or not does not involve caramdir and that caramdir, from the first comment, regards the two answers as synonymous at least as regarding the implementation of the solution). If I do this, should it be CW as, technically, more than one person contributed to it (I would regard this as somewhat extreme since tracking down and properly crediting the true source of many answers would be a Herculean task!)?

I actually would prefer to be able to leave things as they stand (possibly with mine accepted). I would read that as saying "Here's the answer, but the other highly voted answers might add some additional information that is worth reading.". Thus answers are not in competition with each other, but each is voted as to its merits in and of itself.

(While I'm talking about that particular question; I really hope that the two answers are not being used as a contest to decide whether it's best to remove blank lines or insert %s. That really would be detrimental, particularly as the question is (supposedly) by a newbie.)

Answer 1

Just fix your answer and add a short note (in the end of the answer) in which you give credit to Caramdir.

I think we should aim at providing the best possible answers, give credit to those who deserve it, and stop worrying too much about the reputation score. With newbies it's more important that you give them the "fair" share of reputation (to make them feel more welcome and to enable them to use features such as commenting more quickly), but in this case both of you have > 500 points, so I wouldn't worry about that. Earning more reputation isn't that critical for either of you any more, you can already use most features of the site.

Answer 2

Another comment as the one who asked the original the question, for a long time I was actually reluctant to accept either answer because (please don't shoot!) I didn't liked either.

I didn't like Caramdir's answer because it's not complete, and I didn't like Andrew's because of the series of Edit and Second Edit's. A pet theory I have is that so many people like Caramdir's answer because it just gets to the point, even if Andrew's answer is better people just don't want to read the whole saga of how the answer was improved over time.

I was planning to eventually ask this as another question but here it goes, do we really need to keep visible the history of how the question has been edited/improved? Isn't it fine to just edit the answer to keep it complete but concise? (Of course with a small disclaimer at the end with attributions to the people who helped improving it.)

Answer 3

The original and "official" SO policy is like you say, to make sure the accepted answer is as complete as possible. It is pefectly ok to edit in information taken from other answers (preferably with a note crediting the author).

Alternatively (if the amount of content you'd have to insert is significant), you might just add a note suggesting the reader also check caramdir's answer. But there is nothing wrong with adding information to your own answer, even if it is taken from a "competing" answer. It results in a better answer than we'd have gotten otherwise, and that is what matters.

Ideally, a reader shouldn't need to read every answer. The accepted one should be the best, and most complete one. Reading the others is often a good idea as well, but readers won't always do that, or they may find it hard to distinguish between good and bad information if they have to correlate 8 contradicting answers.

Answer 4

However, I have a feeling that the SO route is that the accepted answer should be complete and that it should not be necessary to read the other answers (though of course it's a good idea).

Both your post and Caramdir's post provide different ways of solving the problem. Caramdir could edit his and provide more information about why the lines need to be removed, but that's his choice. I don't think that this is the SO route - except for FAQs, which should have an official answer.

For example, see this hour-old SO thread about unit testing in python. There three answers which provide a package, a tutorial, and a book, respectively, which are more highly voted than several others. These are three equally valid and useful answers. The questioner will select the answer which is most applicable to his problem and learning style, and the others will be highly voted alternatives - honorable mention, if you will. If this were a mature question, I'd read these three and ignore the 0- and 1-vote noise below. In contrast, this question on MO which has a single answer which is provably correct. Perhaps that kind of question has determined MO behavior?

On this site, some answers may be provably correct. However, this is likely to be the exception, rather than the rule. IMO, questions which (1) have answers which cannot be compiled by a single user ("big-list"), or (2) which should have official answers (FAQ and policy-determining questions) should have one community-wiki answer including all agreed-upon points, and maybe some dissenting opinions if appropriate. Most answers should not include all the possible answers.

Answer 5

I'd like to respond to some of the points raised, particularly in Juan's answer, as I think that this issue has exposed some of the underlying assumptions of those coming from SO as opposed to those of us from Elsewhere.

Regarding the "edit history" in my answer, I put this in for two reasons. On MO, we have had the following situation: someone asks a question that has a trivial counterexample; someone else posts an answer pointing out the counter-example; the question is then modified to remove that case; whereupon everyone who looks at the question afterwards votes down that answer since it is now explicitly excluded by the reformulated question. This can be more subtle than I've laid out, so I've gotten in the habit of essentially keeping the edit history visible when I'm making substantial changes.

The underlying assumption that Juan makes, backed up by ShreevatsaR, is that someone taking a look at the question-and-answers is going to go to the trouble of looking at the history, checking the order of events, and voting or commenting accordingly. This also assumes that they know how to do this!. I remember that it took me quite some time before I realised that I should click on the word "edited" to see the history - this is not at all obvious.

The second reason that I kept the history is that the exchange felt like a conversation. Juan (playing a role), was asking for clarification. At that point, I have two options: to fold the extra information in to the answer or to ensure that the extra part is clearly signposted. The point that I wish to make is that the latter is the most useful to the original questioner, since that person knows what I originally wrote and is only interested in seeing what has been added (or altered).

The underlying assumption here is that when answering, one should always have an eye on the future. What will be most useful to someone else reading this question-and-answer? For me, this isn't even of secondary importance. The only thing that matters when answering a question is: does it help the original questioner? Everything else is a side benefit. Of course, we anticipate that there will be such benefit - that's the basis of this framework, but to alter our behaviour to maximise it is wrong, in my opinion.

This site is for asking questions and getting answers. The fact that we learn from watching each other is what brings answerers back to the site again and again. But what draws questioners is the prospect of getting their questions answered. The questioners are our primary resource, and so we must make sure that nothing we do hinders the process of questioners getting answers. And I really do mean questioners, not questions.

Once a question has been satisfactorily answered, then one can ponder whether or not it could do with tidying up. I agree that a new person coming to that question doesn't want to read through the history. However, that cleaning up is the responsibility of the community after the questioner is satisfied. Ideally, the original answerer (me, in this case) should clean up their answer, but anyone with enough rep could do it.

So I feel that when Juan says that he didn't like my answer, I think he's not being quite fair! The Juan that doesn't like my answer, is the real Juan; but the Juan who asked the question is not that Juan, but a pretend one.

We, as the ones committed to this site, should have an eye on building up a good resource. But we should not expect Joe Public to actively take part in that. We want Joe to answer the question and go away again with a warm glow of having helped someone who needed it, so that he keeps coming back to see if anyone else needs help.

I think I've gone on long enough. Let me close with this thought: for me, for this site to be successful, it needs to be more than just the 'tex' tag on SO. It needs to attract people that aren't used to SO at all. I said elsewhere, that between the MO model and the SO one, I prefer the SO one for here. But we need to remember that many people coming here won't have any experience of that model (as I don't, really). Although we can lay this out in the faq, what will really make the difference is how we react to people. If we say, "That's not how things are done.", then they'll leave. If we say, "We've found it best if we do X not Y." then that's alright. But best is to say, "That's a great answer (or whatever)! Do you mind if I edit it a little to make it clearer for people stumbling on this question in 6 months' time who haven't been following the discussion?".