# What is a reference question?

What is a ? These are my concerns to tag one question as a reference:

1. Question is about fundamental problem, e.g How does computer program works?
2. Question contains information that will be repeated many times and one or some good answers can help future owner of problems: e.g Why this language is regular?
3. Question itself is not very good, but answer/s to the question are satisfying 1 or 2, e.g: Graph searching, BFS vs DFS.

In the about we have a rough description, but is not bad to be more specific. One interesting usage of reference questions is finding duplicate questions as easy as possible.

P.S1: I cannot remember but I think I was a creator of this tag, actually there was some reference questions and the related discussion on meta, I tag some of them as reference question to make them easy to search, specially when we want to mark a new question as duplicate. (Some other users like Raphael and frafl done this more preciser than me).

P.S2: May be it's good to merge answers to this in the previous meta post.

• I don't like the tag as it is easy to confuse with reference-request. There is no fixed definition of a "reference question" that I am aware of; compiling the list you link has been a project of mine. The goal is to compile a list of good answers to common, basic questions that can be helpful in many situations. These are rough guidelines, though, and there is no clear process to determine whether a new question is a reference quesiton. – Raphael Sep 9 '13 at 7:29
• @Raphael, Sure the project that you start, is very good, and tagging is easy task (that I usually do it). But suppose newbie is coming to our community, with high probability (more than 50%) he never reads FAQ, also with very high probability he never reads your post on meta, but with high probability he will see reference question tag after few days of contribution (specially if community sometimes bring one of them up). Also Suppose somebody who is member for a year everytime (when he want to reference it) should search a meta post, ... which is not easy task. – user742 Sep 9 '13 at 8:24
• The reference-question tag has so few questions that it is all but invisible. It is true that the meta question is invisible, too, but I have been linking to it in many comments. We'll have to to that for every suitable question, but that's the game we are playing, isn't it? The advantage of the meta site is that we have an explanation and a structured list there, so it's more suitable as a "landing site" for newcomers than a question list. – Raphael Sep 9 '13 at 9:20
• Specifically for (3), I wouldn't tag it as a reference question. An argument such as "answers are covering some future questions" is not a good one, I think. Every answer is possibly covering a future question, and any new question can be a duplicate of any older question. The question is quite specific asking about differences of BFS/DFS. I'd be more willing to tag it if it was "how do I choose what graph search algorithm I use, given X", or something like that. The question doesn't have a "broad scope", nor is it about "general methods and concepts". – Juho Sep 9 '13 at 9:49
• @Juho, Broad scope is depend to your interest, for example how to decide whether a language is regular, has broad scope in automata and languages, but in graph theory, DFS and BFS are widely used and answers to that question are also showing their most appropriate usages. As Is clear by some future question I mean that the answers are broad, not something like $n^2 \in O(n^3)$, for example read the answers there, some of them are good for even far questions (like Suresh's or Raphael's answer). Also tags are for making something easy. Not dealing hard with them. – user742 Sep 9 '13 at 10:17
• @SaeedAmiri I disagree. Just like there's a bunch of ways of showing a language is not regular, there are many more ways of searching a graph than BFS/DFS. Just like the pumping lemma is not the only tool available. Maybe someone else can chime in on this too :-) – Juho Sep 9 '13 at 10:20
• @Raphael, May be is good to link the tag description to meta post to make both of them more visible. Also IMHO, using tag is easier than meta post, once you see the tag one time, then you just need to remember some part of a tag, then you can find it easily and filter questions by tag. Also if we consider this tag as simple way to find related answers to a new question, then is easy to mark a question as duplicate. – user742 Sep 9 '13 at 10:21
• @Juho, May be someone can provide something new, but as my understanding, at least up to now base of nearly all known graph search algorithms is BFS or DFS and they widely used, even if somebody asks how to use pumping lemma (for regular or pushdown automata), in my opinion is still general question and answer to that covers many future questions, but currently, in this question, I mostly like to clarify what we should tag it as reference question, or is this tag good at all ? (e.g Raphael is not agree with this tag), after that we can go more in depth. – user742 Sep 9 '13 at 11:17
• @SaeedAmiri As mentioned above, I don't consider the tag reference-question to be useful but in fact harmful. In particular, it will never make the correct question show up as related. (Gilles has made multiple statements against meta tags, which this one clearly is.) If I were to decide alone, I would remove it completely. – Raphael Sep 9 '13 at 12:03
• On the specific question, I agree with @Juho. A question/answer thread comparing difference graph searches might be a good reference post, one comparing two specific algorithms is not. – Raphael Sep 9 '13 at 12:03
• @Raphael, So you don't like tag, bit what do you think about purpose of a tag? If this tag is correct then what is the role of tag (I said my opinion in the question). – user742 Sep 9 '13 at 12:38
• @SaeedAmiri Tags are good if they describe the content of a question, thus grouping similar questions. reference-question clearly does not do that: a wide variety of topics share this tag. See also here. – Raphael Sep 9 '13 at 12:42
• @Raphael, What jeffe mentioned is applicable to subjective tags, but this is objective one, I mean this case is totally different, most of us know that what are the fundamentals in CS, so if we see this tag we can say ok is about something which is very common that we know it, e.g by your grouping, Algorithm tag is also useless, but this are totally different in nature (compare to subjective useless facts). – user742 Sep 9 '13 at 12:52
• Also about that specific question, In my opinion DFS and BFS are broad enough to be consider as one topic in CS (They are usually book chapter). Specially in the case of graph search algorithms, I don't know if you or Juho knowing any other algorithm which is not descendant of BFS or DFS and widely useful in graph search algorithm, then I want to know it. (We talking about now, not something that may be exists, which is not helpful for now). – user742 Sep 9 '13 at 13:07
• I just saw that you added reference-question to another question that's not on The List. I think this one qualifies to join the club (I think there are a couple of near-duplicates already), so I added it to the list. – Raphael Sep 10 '13 at 22:16

Two questions have emerged here.

## Which questions are suited to be a reference?

I think we agree that a reference question should have

• general character,
• cover lots of potential questions (in particular exercise problems) and

Whether any given question fulfills these criteria is obviously a subjective matter. I don't think there is a shortcut to discussion on Meta for contentious cases.

So far, I have been adding question to the list at my own discretion (with some help from Ran G.); you may decide for yourself how good a job I have done. I certainly invite everyone interested to propose new questions and/or monitor the list (and, if we decide to keep it, ) in order to help ensuring the quality of this list (and the tag).

## Should we have a tag for reference questions?

I think no for several reasons.

• It's a meta tag. It does not aid in searches, it fools the "related questions" mechanic and takes up a valuable tag slot.
• It easily can and is being mis-applied. I have seen (and fixed) several posts using instead.
(This could be amended by renaming the tag; has been proposed.)
• If the tag is to be useful, it has to be only on high-quality questions. This is hard to control without looking at every single question posted ever.
• Newcomers won't know about it, so it does not give automatic benefits.

All good things the tag can possibly do are also achieved by the existing list of reference questions with none of the disadvantages. It's a nicer place to point newcomers to, too, given its structure and introductory text.

• I don't think the tag is necessary either because we have the list of all "reference questions" anyway. Usually the reference questions have a lot of votes too, so looking at relevantly tagged questions means seeing them, pretty much. On the other hand, the list requires manual updating (whereas using a tag doesn't), but that doesn't seem like too much work right now at least. – Juho Sep 11 '13 at 12:44
• Reference questions should also be listed in the relevant tag wiki(s). – Gilles 'SO- stop being evil' Sep 11 '13 at 12:44
• Bad points for meta post: 1. Hard for maintenance, 2. Few number of people can remember it, 3. Hard to search, 4. Just very few people are care about it. By the way if most of people are agree to remove this tag, I'd suggest to remove all tags and create a meta post for each of them. The only disadvantage of this tag is just mistaking with the reference-request tag. But even in this case, still CS.SE is safe and has no problem at all: either somebody will see this and fix it (with 99% chance, and editing tag is easy), or nobody see this (nobody understand it), so there is no problem at all. – user742 Sep 11 '13 at 15:42
• @SaeedAmiri You say "editing is easy" but at the same time "[meta post is] hard for maintenance". Huh? I'd rather edit one question, not dozens. – Raphael Sep 11 '13 at 20:49
• @Raphael, editing tag is easy and because is very easy there is an extra functionality to do this. – user742 Sep 12 '13 at 7:59

I find the tag useful. Here's how it is useful: it is useful for me to identify resources that I can point new users to. I realize it may not be super-useful to new users, since they probably won't know about its existence.

Here's the benefit of , from my perspective. We've identified a handful of topics that often seem to get asked (how do I prove that a language is/isn't regular, how do I compare functions using big-O notation, etc.). Folks have written one very good question and answer on that topic: a reference question. Then, whenever anyone asks about that topic, they can be pointed to that reference question; if their new question is already answered by the reference question, we can close their question as a dup of the reference question.

So far, so good. No special tag needed. But here's where I find the tag useful. I don't have a list of all reference questions in my head. Often, I recall that there's an excellent reference question on the topic, and I want to search for it so I can point someone to it (or close a question as a dup of it), but finding it is a pain. Because we have the tag, I can just search on that tag, and then I will quickly find the reference question.

So, yes, the tag is useful, and we should not nuke the tag. The meaning of is basically: this is a FAQ (a frequently-asked-topic); here's a question that we think gives an excellent overview.

If the tag exists, we should be a bit selective about how the tag is used. We should only use the tag where we think it fits the criteria above.

The BFS vs DFS question is, in my opinion, not a good example of something that the tag should be used for. There is no evidence that this is a frequently asked question, no evidence that we'd want to use the BFS vs DFS question as a frequent dup target, and it doesn't provide a good overview that will answer many questions all in one go. It's just totally different from the other actual reference questions.

That said, if others want to get rid of the tag, I'm fine with that. I'd prefer to keep it because I find it handy and I use it frequently, but I won't stand in the way of removing it. But in any case, let's not pollute it by adding the tag to questions that don't fit the criteria.

• For the first part I'm agree with you, but for the second one, First of all, I am the creator of this tag, for the reasons that I wrote it in my question, Second I cannot understand who is random user? e.g are you a random user? if yes, how one random user can classify others as random user or not? If no, why? Third, for that specific question, I didn't marked my own question as reference, I marked someone else question as reference question because of a reason that I wrote. Finally I didn't get point of your answer, can you show me any logical argument for rejection in your answer? – user742 Sep 9 '13 at 19:15
• +1 Agreed with this! – Juho Sep 9 '13 at 20:51
• @Juho, Can you describe what is a logical point of criticism part? Also if you agree with this argument means you already know what is D.W.'s definition, then would you answer the questions that I asked from this account? (I'm just curious if you know his reasoning). – user742 Sep 10 '13 at 8:56
• @SaeedAmiri Can you clarify what you mean by "a logical point of criticism part"? And D.W's definition of what? – Juho Sep 10 '13 at 8:59
• @Juho, Please read my previous comment: 1. D.W. defined random users, I asked him/her, what is the random user (see details in 2 comments ago)? 2. In the second part, he just repeated somethings that we already discussed, but my reasons for reference questions had 3 parts, and s/he didn't talk about that (for criticism), also s/he just said some sentences like: "no evidence that we'd want to use the BFS vs DFS question as a frequent dup target", while I said why this is important, actually my question and comments on question are much more detailed than this claim. – user742 Sep 10 '13 at 9:09
• @SaeedAmiri Perhaps one could take a "random user" as "non-moderator". Alternatively, whenever a question is tagged as a reference-question and someone disagrees, a meta post can be opened (as was done now). The point is: we get tons of questions like "how to prove L is not regular", therefore a reference question on proving non-regularity makes sense. We don't get tons of question on the differences of BFS/DFS, therefore it doesn't make sense to have a reference question on it. This is why I agree with what was written. – Juho Sep 10 '13 at 9:15
• @Juho, Being Moderator does not bring privileges to do something with tags, and random user in this interpretation makes no sense at all. I think this way of talking is cheap but e.g this:cs.stackexchange.com/questions/13669/… is non moderator question and answer and is reference even in meta question. And more than this, nobody asked this until know, means is not frequency asking (is not comparable with BFS and DFS).--> – user742 Sep 10 '13 at 9:23
• Also if right know students are like to do their homework in formal language, this doesn't mean being regular or not is more important than knowing what is the usage of DFS and BFS and their difference. And last thing but not least, I really expect that when I discuss with someone, he/she show me some new points or details that I didn't see them before, repeating what already discussed (specially forgetting the answers) is not really interesting, and I try to avoid this myself, if you see this from me let me know it. – user742 Sep 10 '13 at 9:27
• @SaeedAmiri Perhaps it would be the best if one has a problem with a question being tagged as a reference-question to take it up on meta. For your last point, sure, wouldn't that be ideal? :-) – Juho Sep 10 '13 at 9:32
• @SaeedAmiri I beg to differ. Moderators do have some handles on tags nobody else does, as far as I know. – Raphael Sep 10 '13 at 10:38
• I agree with this more than with Saeed, but I have two comments. 1) What's wrong with this list as reference for the initiated? Favorite it so you can find it easily, and rest assured that it is a curated list with a proper introductory text (note that changes are visible and can be controlled, other than tag edits that may easily be missed). 2) If we want to keep the tag, we should rename it to frequently-asked-question or similar to avoid confusion with reference-request. – Raphael Sep 10 '13 at 10:40
• @Raphael, If you can do something that we cannot do, doesn't mean on same thing you should have more privilege, tagging a question is one of a basic things that many users should be able to do it, it's not moderator task at all, or creating new tag, again is by community members, it's not moderators work to create a new tag, ... e.g if you can delete question it doesn't mean someone else cannot edit it. Here tagging a question is basic, and in my opinion it has nothing to do with moderator at all. And I suggest you to never think in this way (that you have more permission for tagging). – user742 Sep 10 '13 at 11:48
• P.S: here concept is clear: we talk about tagging or creating tag, and related stuff, not some magic that we are not aware or not related to this discussion at this moment (e.g maybe deleting tag). – user742 Sep 10 '13 at 11:51
• About your idea on rename, etc, I'm not agree with you (current name is exactly what it is), if somebody make mistake with reference request, he should be more careful, and if even happens, is easy to edit the question, also as I said before, dealing with tags is much more easier than meta post (which needs extra work for updates). But is better to listen to others opinion. – user742 Sep 10 '13 at 11:59
• @SaeedAmiri I don't understand what you are trying to imply by saying "But is better to listen to others opinion.". I listen (otherwise I wouldn't be talking to you at all), but I hope I am still free to disagree regardless. What is more, you'll note that I have not killed reference-question even though I don't like it. So I really don't get what you are complaining about. – Raphael Sep 10 '13 at 13:34

I think we should keep the tag (or a similar tag).

• Most of the questions now tagged as reference questions are likely to be asked by students of an automata theory/formal languages class. So if they find a question about regular languages, it's not bad pointing them to our reference question "P vs. NP", because they're likely to ask such a question a few weeks later. So our current reference questions are, in some sense, semantically related. However, we could just introduce a focused "beginners" tag for this purpose.

• Tagging is an automatic way to point from one reference question to the list of reference questions. IMHO, the tiny text with a link to the reference list is rather ugly and only some reference questions even have this text. The best solution for this purpose would be a special check box like the one for community wiki and a slightly different design for reference questions. However, this violates to some extend the philosophy of SE (content shall be featured by votes, not by authorities) and may be useless on larger sites (30k featured questions is like no featured question).

By the way: The reason why I started tagging questions was, that I erroneously thought this question was tagged as "reference-question" by Raphael and he thus changed his mind on meta-tagging.

• We can easily find reference questions without the reference-question tag: they should always be linked from the relevant tag wiki. A beginner tag would be a terrible tag because it's subjective and doesn't really classify the question, SE has gotten rid of such tags. I don't see the point of having a pointer from reference questions to the list of reference questions: when is this ever useful? – Gilles 'SO- stop being evil' Oct 2 '13 at 8:11