11

I've been meaning to post this question for a while, but only now got around to it. Over the last year or two, it has been my observation that a) the rate of (good) new questions has decreased noticeably and b) questions have been closed much more quickly as off-topic. (The first is borne out by the site stats on https://area51.stackexchange.com/proposals/28815/computational-science -- although the mods might have access to more detailed site analytics than I as a 10k+ user -- while the second is personal observation, including my own behavior.)

I suspect a link between the two. Assuming we wish to grow our community (which is not a foregone conclusion, and may be argued here), do we wish to revisit our off-topic policy to make it more inclusive?


My own personal opinion -- which I'll detail in an answer below to be voted on -- is that for a site of our size, we don't have much to gain by being exclusive (and much more to gain by gaining a reputation for being welcoming like https://tex.stackexchange.com, which I hold as a model community on the internet), and restricting our scope to have no overlap with other sites would not leave a viable community. I would therefore appeal to voters to spend some time considering whether the question is really irredeemably off topic before casting their vote. (And I do not exempt myself here -- I noticed that I have gotten much more trigger-happy over time; in fact, this observation is what prompted this post.)

  • @ChrisRackauckas do you have an opinion on this? – nicoguaro Jun 14 '18 at 16:19
  • 1
    @nicoguaro I am not sure if Chris can receive the notification, because he in not "present" in the page, I didn't receive the first notification (I saw the question because periodically check meta) – Mauro Vanzetto Jun 20 '18 at 15:24
  • 1
    @nicoguaro I think the moderator "superping" only works in chatrooms, not in comment threads? So you'd have to ping people from chat with a message containing a link to this post. – Christian Clason Jun 20 '18 at 16:11
  • @ChristianClason, the question is: who should I superping? – nicoguaro Jun 20 '18 at 16:28
  • 3
    @nicoguaro Looking at the history of the off-topic review queue, I don't see any names that haven't commented on this topic yet -- so I'd say there's no further need for a superping. The question is where to go from here... My personal resolution on this is that I'm not going to let the review queue indicator pressure me into casting a vote unless I'm 100% sure about a question. – Christian Clason Jun 20 '18 at 21:01
  • 1
    Another thing we should think about is to more clearly write On-Topic Section. Right now it's a bit confusing, including references to initiatives that are kind of inactive, as I know. – Anton Menshov Jun 21 '18 at 16:15
  • @nicoguaro, I looked a couple of months through the list of people actively using the Close Votes queue. I would say most of them (if not all) already expressed their opinion here. I would say, only another moderator Paul did not explicitly express his views here, probably should be superpinged. – Anton Menshov Jun 21 '18 at 16:22
12

Yes, and we should specify our policies to be more inclusive.

(In fact, I believe the original consensus was much more inclusive, but community overturn meant that this more nuanced consensus was likely reduced to the simple close reasons.)

So that this can be linked to as a reference answer should this become our consensus (after possibly some editing), here's my view (based on earlier discussions on Meta) of our site scope:

Software questions

As long as the software is clearly targeting scientific computing (among possibly others), the question should be assumed to be on-topic unless one of the following applies:

  • Questions on installing and running the software are off-topic.
  • Questions that are clearly answered in the documentation are off-topic (with clearly being the operative word here; if it's that obvious to you, courtesy would call for giving a pointer in a comment when voting to close).
  • Question on software that have their own Stack Exchange-style Q&A should be asked there. (The rationale for this distinction is that the Q&A format is in fact superior to the old-style forums or mailing lists, so here would in principle be better, but if they have their own, they should try there first. If the did and couldn't get an answer, I'd welcome the question here.) Again, it'd be polite to accompany a vote to close with a link; I've tried to collect some in Text blocks for frequent comments, but this list is due for some curating.

Of course, one should still make some effort to ask a good software question; I've given my thoughts in this answer.

Math questions

These are much less contentious, but for the sake of completeness: I'd only consider undergraduate linear algebra and calculus/analysis questions as off-topic and to be asked on https://math.stackexchange.com.

All others are on-topic here as long as they are conceivably relevant to scientific computing; in particular questions about numerical analysis and optimization. I think our core audience right now is application-oriented people requiring help with medium to advanced mathematical questions, and these are (very) poorly served by either math.stackexchange.com or mathoverflow.com.

(I'm not sufficiently familiar with computer science to comment on the corresponding situation, but I assume it is similar; I'd be interested in hearing the view of CS experts on this.)

Programming questions

Here I think our current policy of considering pure programming questions in any language to be off-topic and punting them to https://stackoverflow.com is basically correct, but we still should take some time to make sure before voting to close that there isn't a conceptual error behind this, which would make the answer (and by extension, the question) on topic.

  • 1
    I want to support this move with more than one vote (not possible, unfortunately). I have at least 4 questions (LAPACK, GMSH, etc.) I managed to answer just in time before they got closed. (I would say some have been closed already after my answer was given and sometimes accepted) and all of them would fit without a doubt into the guideline proposed in this answer. – Anton Menshov Jun 3 '18 at 13:43
  • 1
    Something they have on MathOverflow is a Meta question where everyone can post an answer to propose a closed or deleted question for reopening (arguing why it should be) -- this gives more visibility than simply casting a reopen vote and trusting the queue. Maybe after the discussion settles here, you want to open such a post and propose your questions there! – Christian Clason Jun 3 '18 at 14:17
  • 1
    I also think that we need to revisit the policy. And reading the Area 51 site, even the definition of the community. But I have two question: how many people should agree to make the changes?, how do we invite people to the discussion? – nicoguaro Jun 3 '18 at 16:10
  • Do you as a moderator have an option to mention all the top users in this post? (I don't think this will be abusing mod privileges) As most people do not check the Meta very often, and this is a crucial question in my mind. Another option is to have a notification in a yellow bubble, like how SO does for important decisions. – Anton Menshov Jun 3 '18 at 17:20
  • @nicoguaro First, I don't see this as an actual change in the scope of the site, just one particular interpretation of it (and I wouldn't worry about the text on Area 51). So I don't think any moderator-level powers are called for (with the possible exception of adapting the help-center text, but a) I don't think that's far from what I wrote and b) nobody seems to read those, anyway...) In the end, people with the privileges will behave as they see fit; but they might be swayed by linking to a highly-voted meta post. – Christian Clason Jun 3 '18 at 18:12
  • 2
    Regarding visibility: There's the featured tag a moderator can apply to a question to make it show up on main. (I think that's what Anton meant in his comment.) – Christian Clason Jun 3 '18 at 18:13
  • @nicoguaro Mentioning all top users may be overkill (and apart from your fellow moderators and Wolfgang Bangerth, they seem no longer as active), but it might be good to involve the two other people who seem to be actively working on the Close Votes queue (Kirill and Mauro Vanzetto). – Christian Clason Jun 4 '18 at 15:34
  • I agree with the ideas here, but one comment: except for the bullet "what is clearly answered in the documentation is off-topic", I kind of thought these are the ideas I was following anyway. I remember voting to close a lot of really lazy content-free questions, but now that you mention it I can't tell if I was being trigger-happy or if the questions would have been off-topic even taking your ideas into account. In any case, I'll try to follow other people and community standards on this. – Kirill Jun 4 '18 at 19:52
  • Another thought: maybe one reason off-topic standards can ossify is that there isn't enough visible disagreement about them? What if there was an informal rule that if I press the "leave open" button on a question with some close votes already, I should seriously consider leaving a comment disagreeing with the close votes? Would that be helpful? E.g., in the questions Anton mentions answering above, he clearly thought them to be worth answering, but looking at one (that I could find) there is no discussion, not one comment. – Kirill Jun 4 '18 at 20:03
  • 4
    I always try to leave comments and vote for closing as the last vote. I also normally wait two days after I leave a comment. – nicoguaro Jun 4 '18 at 23:57
  • 3
    @Kirill Another thing you (or anybody else) can do is to upvote the question -- I'm much more likely to vote "Leave Open" if I see that other people think this question is useful to have around even if it might be somewhat off-topic. (Case in point: the recent MPI question scicomp.stackexchange.com/questions/29671/…) – Christian Clason Jun 6 '18 at 7:22
  • 3
    I also agree with the idea that if a question is quite upvoted also some off topic is goo to left open. It is useful also because we can have the idea about the user orientation -- @nicoguaro I tag you here only to notify the answers :-) – Mauro Vanzetto Jun 13 '18 at 15:43
  • 1
    @MauroVanzetto, I think that the question serves well the purpose for the discussion... since it already has two close votes. – nicoguaro Jun 13 '18 at 15:52
  • 3
    @MauroVanzetto It's especially useful since it highlights a problem I've often noticed: off-topic flags are used to close a poorly formulated question (presumably because they seem more objective). In this case, I'd say the question is in fact on topic but much too broad (or unclear) -- the answer will be completely different in different situations. – Christian Clason Jun 13 '18 at 16:19
  • 1
    @BillGreene It wasn't quite "rejecting", but I agree that we could/should have handled this differently. I guess the concern with install-type questions is that we could end up with a stream of endless variations of the same question -- we'd have to be much more proactive in turning these into authoritative FAQ-type questions. Would you mind writing up an answer based on your view so that people can vote on it? – Christian Clason Jul 2 '18 at 18:57
6

I strongly support the premise of this question and have some suggestions relative to questions about so-called "niche" libraries and applications.

Many times these niche libraries and applications can help someone avoid a huge amount of re-inventing of the wheel. But often they are poorly-supported by their developers. A broader forum, like CSE, gives the user a better chance of finding someone who can help or suggest an alternative package. But the CSE help document discourages users from asking questions about these if another application-specific help site exists or the question is not considered too mundane, e.g., I don't think questions on installing and running these niche libraries/applications should automatically be off-topic. Also, I think the existence of alternate support forums should be completely irrelevant to CSE.

I propose the following specific changes to the What topics can I ask about here? page:

The sentence "Package developers interested in using this site as a resource should look at this meta question." should be deleted. The linked meta question points to a discussion mainly about why FEniCs users were discouraged from using the CSE site. Instead we should be encouraging users of FEniCs and similar computational tools to ask questions on CSE.

Perhaps a sentence like this should be added to this section:

"Questions that ask "How can I do X with software package Y?" should begin with a brief description of the problem being solved and why you chose package Y to solve it."

In the next section, the bullet starting with "Questions about niche software packages and bugs" should be deleted.

I think these changes are consistent with the statement above, "we don't have much to gain by being exclusive "(and much more to gain by gaining a reputation for being welcoming". These small changes may do some good and are unlikely to do much harm.

  • I up-vote the post. Personally I still have got some doubt about the black hole represented by the installation questions, because these kind of questions need, many times, of a support step by step more suitable for a forum. However, if this will became a problem, in future we can restrict the policy in easy way. Good the idea to be more permissive with questions "How can I do X with software package Y?", it can help to expand community. (this comment is a resume of my previous comments writed before this question appears, in this way it is more well-ordered) – Mauro Vanzetto Jul 3 '18 at 10:11

You must log in to answer this question.

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