13

There have been a few questions flagged recently for deletion (for example pgfplots: I do not know the key '/tikz/nodes near coords' error, https://tex.stackexchange.com/questions/44051/cross-referencing-with-amsthm-and-cleveref#question and https://tex.stackexchange.com/questions/35189/cleveref-package-produces#question). These have all been closed as 'too localized', and the flag for deletion says something like 'of no interest to other people'. I can see this argument, but I can also see that these questions cover common pitfalls. Do we want to keep them, or should they be deleted?

8
  • I think the package-update-related ones can be deleted but is it going to effect the site stats?
    – percusse
    Commented Jul 1, 2012 at 9:16
  • @percusse What stats are you thinking of?
    – Joseph Wright Mod
    Commented Jul 1, 2012 at 9:36
  • I remember the recent discussion about branding the sites which was based on how many questions is asked and what's the unanswered percentage etc., site size in general.
    – percusse
    Commented Jul 1, 2012 at 9:42
  • 1
    @percusse I have no idea whether closed questions count as part of the 'size' of a site or not, while deleted ones presumably don't. That said, we don't really need to worry about this: we are top 10 across the network by various measures, and I think are quite happy with that.
    – Joseph Wright Mod
    Commented Jul 1, 2012 at 9:44
  • 4
    Then for what it's worth, my vote is with the deleting them as they often confuse more then they teach.
    – percusse
    Commented Jul 1, 2012 at 9:48
  • 1
    If we want to cover some common pitfall, than we should explicitly add a well-written question+answer about that pitfall, instead of keeping some badly written questions around.
    – Caramdir
    Commented Jul 1, 2012 at 15:54
  • 2
    I personally think they should be left as other users who come across the same issue may find the question and realize to check for package updates. Sure, if there is a really badly written question this might make sense, but otherwise I don't see any upside to removing those questions. Commented Jul 1, 2012 at 18:51
  • 1
    @JosephWright: I broadened the scope of your question slightly to also include 'not a real question' questions, because I want to write up a response, as kind of a general statement about deleting, in which I also want to cover 'not a real question' questions. Since this question is basically going in that direction, I thought it'd be better to keep it in this thread rather than asking another question. Hope that's ok.
    – doncherry
    Commented Oct 23, 2012 at 20:39

5 Answers 5

7

Removing content is something we've in general been very cautious about. In general, it's been used only for things which are either completely unreadable or which are identical duplicates (for example where a question is posted to StackOverflow and TeX-SX, and the SO one gets migrated to us).

The comments here are split, so I suggest that we only delete this type of question if there is a good question on the same topic. That probably requires appropriate questions/answers writing: writing the questions correctly will not be easy.

2
  • 1
    Are these question really a burden to the site? If you delete posts that are exactly the same, rather you merge them as admins and explain there similarity or something like that and what lead to there deletion so that new comers can refer to them to learn from others mistakes.
    – azetina
    Commented Jul 4, 2012 at 15:19
  • As I've said, my answer here mainly reflects the fact that there was no better one from anyone. There are now two answers which take different tacks. I may well delete mine if it appears that the other two cover the cases well.
    – Joseph Wright Mod
    Commented Jan 13, 2013 at 9:52
6

I would like to make a case for deleting questions a little more liberally than we currently do.

I’m pretty sure it was flags by me that initially triggered this question, when I didn’t have the 10k privilege of voting to delete yet. Most, if not all of my flags back then were approved by a diamond mod and the questions deleted. Since I reached 10k in October 2012, I’ve cast 24 delete votes, all on closed questions, but none of them reached the three delete votes necessary for deletion; in most cases, my vote remained the only one.

Here are some specific reasons why I propose to delete certain (types of) (closed) questions – after they’ve been inactive for a little while:

  1. It’s impossible to determine what the problem is and the OP isn’t responding
  2. The problem was something totally unrelated to the question
  3. Identical duplicates (as mentioned by Joseph in his answer)
  4. The question is really unspecific, probably no proper MWE, and (thus) possibly several issues in one question
  5. Question is extremely localized and a terrible fit for our question format
  6. Question contains several questions and is a terrible fit for our question format
  7. The problem is extremely trivial (typo)
  8. Question is off-topic and wasn’t migrated

My general goal in deleting such questions is to uphold the overall quality of tex.sx and to foster the aspects that make stand out from “normal” forums. I don’t really think we want to help users googling their problem just randomly, but rather through specific and well-formed questions.

If a problem is generally relevant, the question should either be edited to fit our format (and not deleted), or the question should be asked again in a new post, either as a dummy question or by the next user coming across this problem. Most of the advice given to the OPs of the questions I linked to is pretty general and could be given again any time by many of our experienced users.

Questions containing several problems/questions might keep good questions, containing only one problem, from being asked.

And finally, in exchange for helping them with their problems, we expect users to make the effort and ask proper questions with a MWE etc. – if they don’t they shouldn’t be “rewarded” with the Stack Exchange “currency” of reputation. However, if the question was voted below zero, deleting it gives the user a chance to keep going and ask better questions without the “burden” of bad questions in their profile.

Summing up, I’d be happy to see a little more activity on the delete vote page and us as a community weeding out just really the worst of the closed questions.

2
  • Maybe I misread your “identical”, but in my eyes for duplicates it is better to only close them. It happened to me several times, that a web search showed me a closed duplicate, where I then found the link to the original, and I would not have found this original with my search items.
    – Speravir
    Commented Jan 13, 2013 at 21:43
  • @Speravir I intended "identical" to mean something else than "exact" (in SE terminology): if you take a look at my example question for that reason, you'll find a comment of mine linking to a question that is a literal duplicate, word by word (modulo edits); this is usually the result of cross-posting and migration. For normal "exact" duplicates, I agree that we should only close them (and sometimes merge).
    – doncherry
    Commented Jan 14, 2013 at 1:49
6

Unless something is offensive or an exact duplicate (or I wrote it) I would never vote for deletion. Taking a random example the list in another answer

Which IDE/Editor and "compiler" do you use for TeX/LaTeX? - How do macros work?

The question is badly structured (mainly because it is lots of questions) and it got closed as NARQ (that already is probably more than I'd do) but parts of the question are reasonable and there are some good points in the answers. Unless it is being suggested that a someone with mod powers manually edits the question to multiple questions and re-distributes the answers to the split up format, deleting the question just throws away the work people put into answering it.

3
  • 3
    Of course, one issue here is that on the main site (StackOverflow) such questions get closed quickly so they do not accumulate answers, with the idea being close-edit-reopen-answer. As we go for a slower closing approach we do leave ourselves with this issue.
    – Joseph Wright Mod
    Commented Jan 13, 2013 at 10:26
  • Putting it a little drastically: Then you’re basically ignoring how Stack Exchange is supposed to work, and treat it like any old “normal” forum out there.
    – doncherry
    Commented Jan 14, 2013 at 9:16
  • 1
    @doncherry I'm just reporting to be what I think is the normal human behaviour of people asking questions on any forum. The vast majority of questions have been asked before, or could be answered by reading a manual. People ask anyway because it is easier, the social interaction is fun and they get an answer tailored to their particular usage. Making the site easier to search helps people making answers find good older answers to cross link to but I don't think it makes an appreciable difference to most people coming to the site with a problem for which they want an answer. Commented Jan 14, 2013 at 10:06
5

There are some questions closed as "too localized" that should be deleted:

This doesn't work,

Something \emph{is wrong with the font
or \mu makes all text after it in italics

where it's obvious that the errors are a missing closing brace or proper confinement of a math symbol.

Conversely there are some "too localized" questions that might help somebody who's in a similar situation. Where to draw the line? Difficult to say. Some criterion for candidates to deletion:

  • only obvious and easily fixable errors
  • errors which depend on clear misunderstanding of the documentation
  • questions that depend on private packages or sets of macros that aren't made available in due time
  • ...

It's more difficult for me to tell something about the "not real questions". We don't see many of them, actually. Probably most of them should be deleted; I'm unsure about those where somebody attempted to answer.

3
  • 2
    If someone is confused enough by an error message to post a question it can't be completely obvious. We should have one answer for it and close others as duplicate in the normal way, we shouldn't close questions for having easy answers. For the first one you get Runaway argument? {is wrong with the font \par .... ! File ended while scanning use of \emph . <inserted text> \par Which nowhere mentions a closing brace and for people not used to compiler editor messages, such messages are simply scary rather than informative. Commented Jan 13, 2013 at 13:55
  • @DavidCarlisle Some errors are less trivial than others, so the criterion is not "simple error=delete". However, one has to consider also that these questions are difficult to find with a search, unless their title is clear and tags are appropriate.
    – egreg
    Commented Jan 13, 2013 at 15:57
  • As I commented to Joseph in chat I'm unconvinced that users search before asking questions so I'm unconvinced that it is that important whether the answers are findable. Commented Jan 13, 2013 at 16:04
4

Although i agree with upholding the site quality, I disagree with deleting any question as Tex.SX knowledge and experience should reach any dummy guys like me.

Today most of people are lazy googlers including myself even though we know the website we try to google and go.

But I am not lazy to post a good/well framed question at TeX.SX as i wish to be well-behaved civic duty following guy.

In my view as new user, everything on TeX.SX is useful real questions, answers,comments,closed questions,not a real question, too localised, too long to read etc... because it is someone's experience to learn from (for real questions, answers,comments) or stop committing mistakes (closed questions,not a real question, too localised, too long to read).

Deleting question altogether should not be option. May be the OP has every right to do it. Deleting to be executed with great caution on very very exceptional case like spam/flame war etc.

PS: Its just my point of view. Sorry if i would have hurt any one's feeling.

You must log in to answer this question.

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