39

Recently a new pattern is emerging that is worrying me.

  • A new user asks a question that may need some clarification. Well, it's the first question in a new forum, often by a TeX novice, often by a non-native English speaker.

  • S/He goes away, as new users usually do not expect that there is an immediate reaction. (I remember well that after my first question I was quite surprised to see how fast the first responses came.)

  • Some hyper-active members of our community start to close the question shortly after as being too broad, unclear or whatsoever, sometimes even without a single attempt to clarify the issue, or without waiting for the user to return to answer.

I think it is rather frustrating to see ones question closed so quickly, without any communication taking place.

Can we agree to (try to) talk to people first and then to close such questions only with some delay, say two days (maybe more if it's weekend)? (I'm not talking about obvious duplicates, clear off-topics, or spammers. But even then, apart from spammers, some comment seems appropriate.)

  • 13
    Sadly, a perennial problem. Close voting is not a sport and Reviewers what do you eat for breakfast. – Alan Munn Apr 14 '17 at 23:52
  • We had this wish/request/discussion a 'few' times before, here on Meta, in the comments below the questions that are endangered to be closed or in chat... – user31729 Apr 15 '17 at 0:06
  • 1
    I think new reviewers don't know it. It's better if you tell them when you see such cases, like @ChristianHupfer did with me some time ago... I thought that closing an answer at once was correct, reopening it only if the OP gave some new info. – CarLaTeX Apr 15 '17 at 3:08
  • 9
    There's another closing trend lately which annoys me even more: closing questions as “off-topic because solved in the comments”. The right thing to do would be to post an answer… – clemens Apr 15 '17 at 9:13
  • 2
    @ChristianHupfer I think what's wrong is the reason "solved in comments", because this is not a valid reason. It should be something like "uncharacteristic/insignificant syntactic error", "unclear what's been asked" etc. Whenever it is unlikely that anyone else is able to find the post and/or use the answer (like these ambiguous missing brace errors) it should be closed. Otherwise, if the solution in the comments may be of value to someone else, it should be turned into an answer. So let's avoid the argument "because solved in the comments" in the future. – gernot Apr 15 '17 at 10:15
  • 2
    actually this site is noted amongst the stackexchange network for not closing stuff early (and not downvoting) so while I'd agree with the sentiment, the question would be easier to understand if you linked to specific examples. In particular is it a general trend or is it (as happens sometimes) a single reviewer who's over active. (I agree that solved in comments isn't a good close reason) "solved by OP as noted in comments" might be, but that's rather different. – David Carlisle Apr 15 '17 at 10:30
  • 2
    @ChristianHupfer but when the post itself is a mess then closing as unclear is the better option, no? – clemens Apr 15 '17 at 10:41
  • 2
    @ChristianHupfer I am not against closing. I am against “closing as off-topic because the question is solved in the comments” – clemens Apr 15 '17 at 10:42
  • 3
    @clemens -- i've had the experience where a wild guess was confirmed by the op, who asked me to post that as an answer. when asked, i do try to provide an answer that attempts to explain why i made that guess. that way it could potentially help someone else. but it's true that some questions are such a mess that i can't reasonably come up with a search argument that would find them. there was once a reason something like "too localized, won't help anyone else" that was removed site-wide. i miss it. – barbara beeton Apr 15 '17 at 12:22
  • 2
    @Braiam well true but the first vote is often decisive, once a question picks up a close vote it gets into the review queue and often gets closed, if it was left longer it might have been answered, but as I say, i don't see it much of an issue here (try posting a poorly worded question to stackoverflow, you'd be glad the question was closed before your rep reached -100:-) – David Carlisle Apr 15 '17 at 14:09
  • 3
    I think closed because solved in comments is shorthand for something longer and more off-putting. If you see questions closed because people have got solutions, that is a lot less off-putting than seeing them closed because unclear etc. Closed because solved in comments and the question is so bad we can't figure out how to make it or an answer to it useful is criticism with no useful purpose. And, after all, that is often why I vote to close: if it weren't solved, I wouldn't suggest closing. Why go out of my way to be rude when I can say something polite and true, if not the whole truth? – cfr Apr 16 '17 at 1:35
  • 1
    Maybe we should also have a similar question to this one, about answering questions before the OP has clarified their question :-) (Example) – ShreevatsaR Apr 19 '17 at 19:25
  • 2
    There's a reason the wording is "on hold", and you do have reopen votes. Note that having a question closed doesn't prevent the OP from doing anything at all, it only prevents answers from being added. So it's not entirely clear what the issue is here. The OP can still react regardless of whether the question is on hold or not. – Jason C Apr 25 '17 at 13:50
  • 2
    @ShreevatsaR "Maybe we should also have a similar question to this one, about answering questions before the OP has clarified their question :-)" -- There's a solution for that already: Put the question on hold. Answering is the only thing that putting a question on hold actually prevents. :) – Jason C Apr 25 '17 at 15:48
  • 1
    Going through unanswered question, i sometimes see questions that have absolutely no comment; sometimes i see questions with a long comment thread, where users had to pull all the information by the OP piece by piece, ending with a two day delay comment "Yes, this is exactly what i need". He never got an answer, because nobody was pinged by the comment. How to handle those questions? Where the OP didn't know what to do to make the question better? – Johannes_B Apr 26 '17 at 5:28
9

I was tempted simply to agree with the comments; but I do feel this deserves (another!) actual answer, and I don't 100% agree with the one that has been given.

Summary: Except in the very clearest cases, no question which might in due course be answered should be closed until at least 48 hours have passed. And unless the question obviously doesn't deserve an answer, we shouldn't normally vote to close a question that we wouldn't fee competent to answer ourselves (whether or not we would be inclined to do so).

In my opinion, there are only only a few clear cases where a question should be closed before an answer has been given or a reasonable time (at least 48 hours) has elapsed:

  • If the question is unequivocally spam.

  • If the question is an absolute duplicate question (see below) to one that already has a good answer and, preferably, the OP has confirmed that it is, i.e. that the answer given to an earlier question has worked.

  • If the question is clearly off-topic, or opinion-based, so that it will never deserve an answer.

  • If, as often happens, the questions was the result of some truly "one-off" problem or snafu on the part of the OP and has been resolved in comments. As we know, there's not pat category for this, and the practice is to close it as off-topic, IIRC. Anyway, the key point is that the question doesn't need an answer. But we have to be careful because people often make the same mistakes, and so these questions can be valuable.

In nearly every other case, it is right to wait before closing a question, even if it seems very likely that it will in the end be closed:

  • The question is not clear. Give the OP a chance to clarify it. Be as specific as possible about what is not clear.

  • The question lacks truly essential information to enable it to be answered (e.g. as to a package or class that is being used). Give the OP a chance to provide that information. Be as specific as possible about what is needed.

  • The question is probably a duplicate to another similar but not identically worded question. In that case, point the OP to the question and ask if it is really a duplicate.

But we have always to remember these things:

  • A duplicate question is not the same thing as a duplicate answer. You know that an answer that tells you how to change a title to, say, italics, will also probably tell you how to change it to bold. A new TeXer won't. Experts are cursed with knowledge, and can't see that what is obviously the same problem for them looks quite different to others. The practice of referring posters to a selection of related answers as if they were duplicates is counter-productive. We are answering questions, not writing a manual.

  • Manual-shaming is shameful. Manuals vary in quality. Even good ones are often dense, and not always well-organised. In the case of some of the "mega-packages" (KOMAscript, biblatex, fontspec, tikz) the manuals are hugely forbidding and not at all structured for quick "task and finish" reference. RTFM is the pedagogical equivalent of observing that a language is Turing complete -- it really doesn't help in practice. Much better either to ignore the question, or answer it (with a reference to the manual, if you feel like it, or a reference to other more discursive material that may be more suitable for a new TeXer).

  • Saying "welcome" is not the same thing as making someone feel welcome. Saying "welcome, now please take off your shoes, don't touch the objets d'art, don't sit in my favourite chair, read this set of rules for the house" ... it's not actually welcoming. This stuff needs to be done with tact, remembering that people come here with vastly different background knowledge and sometimes not very good English.

  • New users have real problems creating MWEs, which we should be much kinder about than we often are. How often do we see the inconsistent criticisms: "you left out some package you are using" vs "you have listed a load of packages that don't matter". Experts have a hunch, usually, what packages are likely to matter and which are not. New users often come with weird "templates" which load up all manner of deprecated cruft; they don't know about things like \filecontents, standard bibliography examples they could use, or example graphics. These things need to be handled quite delicately. It's often better and easier simply to edit the example when answering the question than to complain.

  • It is nearly never right to close a question because it is a "bad" question. One downvote and no answers is the way to deal with those.

  • It's just as important to upvote well written questions. There's a lot of stick around here sometimes, and not enough carrot.

I'd also venture a suggestion: those who do the valuable work of triaging questions (and it is valuable) should try to focus on questions that they feel they might conceivably answer, i.e. they understand enough to answer or at least to know how they would work an answer out. Just searching for similar questions if you don't understand the area, or demanding MWEs if you aren't personally going to use them, often doesn't help. Sometimes a person who could answer the question finds it quite clear without a fully compilable MWE, for instance. I'd say the same goes for voting to close. One should be pretty cautious about voting to close questions that are quite outside one's ability to answer, if the substance matters.

  • You cannot realistically enforce a 48 hour policy, even by posting it on meta, mostly because most users don't read meta (only 4% of your active main users have visited meta in the past 2 weeks) and because the workflow for this is not feasible (requires bookmarks or user scripts and possibly reviewing 10 pages [~48 hours] of questions). Instead, if you want to consider something like this, first identify if most of your close votes come from review vs. drive-by votes. Then, if they come from review, you can ask to have your queue delay increased. – Jason C Apr 25 '17 at 19:54
  • (Also it's not realistic to expect people to check the ages of posts in the review queue then skip ones > 48 hours, which also removes it from their queue forever so if they didn't bookmark it before skipping and check back 2 days later, the high level systemic effect is a possibly large % loss of overall reviewers. Nobody's going to do that [or, more accurately, at least 4 people won't do that]). – Jason C Apr 25 '17 at 19:57
  • 2
    I'm not interested in "enforcing" anything and I pretty much deprecate that language. I'm certainly against adopting a policy because it fits the tools I happen to have. That's back to front. Anyway, if questions don't get prematurely flagged/proposed for closure, they won't get queued. If they slip into oblivion unclosed, I doubt it matters. – Paul Stanley Apr 25 '17 at 20:54
  • Well, I use "enforce" very loosely. Thing is, I'm also against adopting a policy because it fits the tools in spirit, but in reality, users tend to operate within the constraints of the tools. The most effective path forward, then, is to push to change the tools, and SE does respond to that sort of thing, quite well in fact, but the basis must be a compelling argument, which means identifying the real issues and then identifying how the tools contribute to those issues. – Jason C Apr 25 '17 at 21:01
5

I'm not an active user anymore but it's strange to see that this is still an issue. I can give a summary as I have tried to convince many people and failed miserably back in the day. It's funny to see that not much has changed:

  1. Some go for badges and some go for the review queue. But some think this place have rules (they weren't around when these alleged rules supposedly being set at all so mostly they make stuff up). They also think that new users are mostly leeches who are trying to take advantage of extremely worthy users and their efforts. As if TeX is something really important in any line of business they borrow most of this new user jargon from Stackoverflow where professionals really try to slack off by asking questions. But that's just nonsense for TeX. Thus they don't see the obvious logical fallacy of educating others if the number next to the nickname is not high.
  2. Other users just ask and answer and enjoy working with TeX.

You just need to learn to filter out the first and stick to the second type. If you vote for reopen the close-voters will also vote for opening even though they voted for closing already in the first place. It's just a stupid system.

  • 2
    I do not agree fully. I'm here for fun and to answer interesting questions, but when I see the n-th question about some stupid template without MWE, I just vote to close because I don't want anyone to waste time on this. – Henri Menke Apr 23 '17 at 9:05
  • 6
    @HenriMenke It's not your call to design how others would like to spend their time. If you don't like a question ignore it unless there is significant problem with it. – percusse Apr 23 '17 at 11:45
  • 1
    @HenriMenke I think being active on facebook is a waste of time. Still I don't go around trying to get all facebook accounts locked. – sgf Apr 23 '17 at 13:28
  • @HenriMenke Some people recognise template snippets, know where to look for to get testable code and can post an answer within minutes. There was a question i would have been able to answer without a MWE, three others requested an MWE. Bernard posted an answer based on assumptions, which still answered the question. So, why should the question be closed? – Johannes_B Apr 23 '17 at 14:06
4

So here's the thing. Most importantly, bear in mind that a question being placed on hold does not prevent the user from doing anything that they could do otherwise. The only two effects it has is it prevents answers (which won't usually be possible if the question is unclear, anyways), and, of course, it makes the question a candidate for deletion later.

With that in mind, I believe the wrong questions are being asked here, and also the premise is slightly off:

  1. You keep saying "closed" but, at least for certain reasons (like "unclear"), you mean "on hold" (see e.g. the wording in the box). This is significant: Long ago this wording was changed specifically to highlight the fact that the state is often temporary. (It remains "on hold" for 5 days until it is labelled "closed", although that is only a wording thing, the possible actions a user can take are not affected.)
  2. As mentioned, the user can still do everything they could do before.
  3. The "on hold" box says "edit the question" -- that's the reaction the user should take, and they can still take it.
  4. Putting a question on hold isn't bad.
  5. What you should really be doing is pushing for people to reopen more quickly and stop treating a question being put "on hold" as some sort of showstopper, which it isn't.
  6. You should be pushing for polite explanatory comments from existing users when a question is put on hold, explaining to the user how they can clarify and that their question can and will be reopened once they do clarify. This is the most valuable thing here: In addition to what is displayed in the box, friendly, constructive comments are what makes the difference between "on hold" being perceived as good vs. bad.
  7. If you believe the "on hold" instruction box is unclear to new users thus leaving them not knowing what to do, you should push for better wording.
  8. More subtle of course, your post may unintentionally contribute to a larger problem: By making a post like this you cast "on hold" in a negative light, thus increasing its undeserved bad rap, and it's a bit of a spiral.

Check out What is a "closed" or “on hold” question? for further info about the intent and process, as well as http://tex.stackexchange.com/help/closed-questions.

There are two MSE feature requests you may be interested in supporting:

The "on hold" system isn't bad. It's good. It's good because it prevents wandering answers to unclear questions, and that's the only thing it prevents. Having a question be put on hold is independent of whether or not the user can react, and therefore it does not matter if it's placed on hold before or after the user has reacted.


There is exactly one situation where it actually can create a bit of friction, which is the relatively rare situation where 5 people deem the question is e.g. unclear, but then somebody comes along who is actually able to make sense of it but can't post an answer. This does happen sometimes. If this happens and you are that potential answerer, what you should do is:

  1. Edit the question to make it clearer for everybody, since you are able to make sense of it.
  2. Nominate it for reopening if you have the privilege.
  3. Leave a comment explaining what you did.
  4. As a last resort you can always solicit reopen votes for a question you've legitimately improved on meta or in chat.

The above is just for that one edge case.


Btw, some data, if you want it, limited to the past year:

Total Questions:        27001 
Open Questions:         21203  (includes reopened)
Closed Questions:        2704  (not deleted)
Deleted Questions:       3094    
Closed Unclear/Broad:     387  (not deleted)
Reopened Unclear/Broad:    35                   
Reopened Total:           144      
Avg. Time to Close:     21810 minutes (~15 days, all reasons)
Min. Time to Close:         1 minute  (all reasons)
Max. Time to Close:    492441 minutes (~11 months, all reasons)
Median Time to Close:     245 minutes (all reasons)

Of non-deleted questions, 387 out of 27001 were put on hold as unclear / too broad, and of those, 35 were reopened. This data is not meant to sway anything one way or the other, it is only data, interpret as you will.

  • 1
    If there are no chances of getting an answer from the OP's point of view it's just administrative noise no matter you justify as a reviewer. That edge case is our daily issue. – percusse Apr 25 '17 at 19:16
  • In your answer I cannot find any argument that makes it reasonable to close a question (spam aside) quickly. On the other hand, it is reasonable to leave a comment asking for clarification and giving the OP time to react. No need to confront the OP with a banner "post on hold". This is not part of normal conversation. And in the cases that finally triggered my post nobody cared to leave any comment, but went for closing, or there was a clear answer but the closers were just ignorant. All these cases can be handled by a simple rule: Do not close earlier than after 48 hours. – gernot Apr 25 '17 at 19:19
  • @percusse All right, so, if that is your daily problem, it's still the wrong question being asked. That problem has nothing to do with the OP or whether they can react or not. If that is your problem then what you should be saying is "Please do not close questions as unclear simply because you do not understand the subject material." That's a valid problem, and entirely different than the way gernot framed it. See, it's all about identifying the real problems and tackling those. Ask the right questions. – Jason C Apr 25 '17 at 19:21
  • @gernot Well, you misunderstand. While I personally see a lot of value in closing questions, that's actually moot and I don't want to spend time on that. Arguing that is not my intent. The first point is: This has nothing to do with "before the OP has a chance to react". The second is: You must consider reality: Questions will be closed more quickly than some people (like you) want, whether you think they should be closed or not, and you simply cannot change this because ... [cont'd] – Jason C Apr 25 '17 at 19:33
  • 1
    @gernot ... only 4% of your users are active on meta which is the upper bound of people that you could hope to reach with this and the workflow to delay closures by 48 hours is simply not feasible. There is no way to do that smoothly other than bookmarks and frequent checks, user scripts, etc. (E.g. 48 hours covers the first 10 pages or so of questions on the site, too much to casually review.) You need to find the real problems and address them in a realistic manner, using the existing system and capitalizing on the good parts. – Jason C Apr 25 '17 at 19:35
  • @gernot Btw you may be interested in meta.stackexchange.com/questions/284710/… (you should identify, btw, if your closures are coming from review vs. driveby votes). And no that FR did not go over well (long story, there was drama involved and also I was sort of taking one for the team, I don't actually support that FR even though I proposed it... don't ask, heh) but it could be a starting point for inspiration. – Jason C Apr 25 '17 at 19:37
  • @gernot It may also help to keep in mind that SE's purpose is not just to help the original asker and get them out the door, its purpose is to curate good Q&A content with long term value. Future readers are as important as, if not more important than, the original poster. – Jason C Apr 25 '17 at 19:41
  • (Another way to phrase all this is: Stop trying to change user behavior. You can't, especially if it's for the wrong reasons. Instead, work with the site mechanics. Identify real problems, come up with realistic solution that can even reach people who don't check meta [such as feature requests for site mechanic changes, close reason rewording, etc.] and go with that. Furthermore, if you believe you've identified a problem, do the legwork to verify that it is really a problem, hit up SEDE if you have to. Think carefully and you can have an effect. Otherwise you won't solve anything.) – Jason C Apr 25 '17 at 19:44

You must log in to answer this question.

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