What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 134 Stack Exchange communities.

As described here (and elaborated on here), we're preparing to roll out some fairly major changes to how closing works that will allow for pre-defined, site-specific off-topic reasons.

On a site with as much history as Stack Overflow, defining the exact boundaries of what is on- and off-topic can be a real challenge. Rather than trusting my gut on this, I sat down and analyzed a random sampling of recently-closed questions, wrote up a handful of reasons that covered the bulk of them, and then threw them at the moderators for feedback. After some discussion and several revisions, this is what I'm suggesting for the initial rollout:

Custom off-topic close reasons

  1. Questions about general computing hardware and software are off-topic for Stack Overflow unless they directly involve tools used primarily for programming. You may be able to get help on Super User.

    this was the most commonly asked off-topic topic, even excluding questions that were eventually migrated

  2. Questions concerning problems with code you've written must describe the specific problem and include valid code to reproduce it. See http://sscce.org/ for guidance.

    Not every question involves a specific chunk of code, but if it does that code must be included, as short as possible, and understood well enough by the asker to allow for a descriptive title; "here's my code, find the bug" questions should be closed ASAP - this covers the bulk of reasonable Too Localized closures.

  3. Questions must demonstrate a minimal understanding of the problem being solved. Tell us what you've tried to do, why it didn't work, and how it should work. See also: Stack Overflow question checklist

    this complements reason #2 for questions that don't involve already-written code. If you're asking others to solve a problem for you, you must be able to understand and communicate it well enough to allow useful answers to be written and found by others.*

  4. Questions on professional server, networking, or related infrastructure administration are off-topic for Stack Overflow unless they directly involve programming or programming tools. You may be able to get help on Server Fault.

    There's no longer a migration path from SO to SF, but that doesn't stop folks from asking system administration questions on SO. Disturbingly, a lot of questions on programming tools get closed for this reason because certain folks think development servers are at all like the real deal - this might help to discourage that.

  5. Questions asking for tool or library recommendations are off-topic for Stack Overflow as they tend to attract opinionated answers and spam. Describe the problem and what has been done so far to solve it.

    This one is long overdue


I should also note that once the new system is rolled out, we'll be able to query and analyze the reasons provided via the "other" off-topic option in a fairly simple fashion - I fully expect to be tweaking and changing these reasons fairly quickly based on observed usage. But that doesn't mean we shouldn't try to hit as many common mistakes as possible right out of the gate, so...

Thoughts? Improvements? Concerns?

share|improve this question

closed as off-topic by Anna Lear Jun 2 at 19:42

This question appears to be off-topic. The users who voted to close gave this specific reason:

  • "This question pertains only to a specific site in the Stack Exchange Network. Questions on Meta Stack Exchange should pertain to our network or software that drives it as a whole, within the guidelines defined in the help center. You should ask this question on the meta site where your concern originated." – Anna Lear
If this question can be reworded to fit the rules in the help center, please edit the question.

8  
It's still unclear to me why reason #3 is specific to questions regarding assignments. Not that I disagree or agree, since I can't really do that until I understand its rationale. –  BoltClock's a Unicorn Jun 20 '13 at 7:57
5  
I agree with BoltClock... that close reason might be more widely applicable as "SO is not a plz send teh codez website. We expect users to have a basic understanding of the problem, and to have attempted to solve the problem at hand themselves first" –  Matt Jun 20 '13 at 8:00
    
@Bolt: it's simple: those were the sorts of questions I observed being closed for this reason. –  Shog9 Jun 20 '13 at 8:00
2  
Ah, as opposed to others that were getting downvoted only but left open to answers. Makes sense now. –  BoltClock's a Unicorn Jun 20 '13 at 8:01
5  
Number 3 is really a good idea, I'd even venture that the original premise for SO was that you'd have minimal understanding of the problem area before asking a question. The examples I have seen from Joel when he is asked what SO is about are all like this: "I have this regex, but it doesn't cover this case", or "how do I do a 64-bit integer multiplication in assembler?" Nice to see the requirement in writing. Like @BoltClock'saUnicorn I wonder why it should be confined to "assignments", though. Edited: Ah, I see Shog9's explanation now, although I'm not sure it would hurt to make it general. –  Monolo Jun 20 '13 at 8:02
    
We can always broaden these, @Monolo; I think it covers an awful lot of ground as it stands though. Given the whole purpose of the change is to increase the specificity of these reasons, I tend to think it's worth seeing how they play out in practice before generalizing too much. FWIW, I'm kinda trying to eat my own dogfood here... –  Shog9 Jun 20 '13 at 8:08
5  
Looks great. ♥ Numbers 2 and 3. Evil powers will try to cut them down. Please defend at all costs. :) –  Pëkka Jun 20 '13 at 8:22
2  
Point taken about #2, and while I think these close reasons will work really well, just a small observation about #5: The wording "library" may not be part of all of the intended audience's vocabulary - many web platforms don't use that name, but call them packages, frameworks, etc. instead. I wonder if it would be worthwhile adapting to that nomenclature? Web development is after all the gateway drug of programming. –  Monolo Jun 20 '13 at 8:37
6  
Could we stuff books in #5? –  Mat Jun 20 '13 at 8:53
    
@Monolo I think they'd be able to figure out, from context, what's going on. I really don't forsee a whole lot of meta posts along the lines of "I'm not asking for a library recommendation, I'm asking for a framework recommendation. Reopen my question NOW!!!" –  Servy Jun 20 '13 at 14:08
3  
Can you help me to understand this a bit better? If Too Localized was ok to use for syntax errors (misspelled variables, missing semicolons, etc.), and Too Localized has now been subsumed into Off Topic, which of these would I use in that circumstance? Or would I have to use a custom off topic reason...? –  joran Jun 20 '13 at 14:10
4  
@Servy Ok, I'm happy to use that reason, but I think it will be weird, as many of those typo questions are quite well formed, with a concise, reproducible example, and the OP will have described exactly what error they're getting. They just haven't realized the cause. –  joran Jun 20 '13 at 14:14
1  
@Pekka웃 Perhaps I'm part of the OP-loving/evil powers you fear, but I'm not arguing with 2 or 3 (or any of this very-well-thought-out list). They're well phrased too. –  AndrewC Jun 20 '13 at 15:41
1  
These are great! Looking forward to having this specificity and clarity on the closing banners. Very nice work! Now we just need to change the heading for them. –  Josh Caswell Jun 21 '13 at 19:05
1  
3 is a specialization of 2. 4 is a specialization of 1. You can roll the specializations into the general reasons and add the text together. We shouldn't be splitting hairs on close reasons that are essentially the same thing. –  casperOne Jun 24 '13 at 3:07

6 Answers 6

You can drop the 'regarding assignments' text.

Whether it's an assignment or homework doesn't matter; it could even be a professional question (How do I move the turtle in logo?). The point is, we want people to have tackled the problem themselves, first.

Questions must demonstrate a minimal understanding of the problem being solved. Tell us what you've tried and why it didn't work. See also: Stack Overflow question checklist

share|improve this answer
7  
That LOGO question is anything but professional ಠ_ಠ –  BoltClock's a Unicorn Jun 20 '13 at 13:29
    
The turtle in LOGO! Slowly I turned. Inch by inch, step by step... –  Josh Caswell Jun 21 '13 at 19:03
    
After thinking about this for a while longer, I've revised #3 to drop "assignments" and clarify what is required. This broadens the coverage somewhat; I'll be keeping an eye on how it is used in practice. –  Shog9 Jun 24 '13 at 19:57

Please rename "Off Topic" to "Site-Specific Reason", or "Off-limits"

There seems to be some degree of angst on these questions and answers over using "Off Topic" for some of these close reasons, whilst there is much less angst about the reasons themselves.

1 (SuperUser), 4 (ServerFault) and 5 (Recommendations) say "off topic" in the text and this is fine. 2 (code dump) and 3 (assignment dump) don't say off topic in the text, but are nevertheless closable.

If we're having this much argument over the use of the phrase "Off Topic" ahead of time, I think we can expect a similar level of "how come my code is off topic on a programming site?" meta questions as "why is this question not constructive?", and for the same reason:

If you extend the meaning of a phrase in a close reason to cover something else, you end up having a pointless argument about the phrase instead of a helpful discussion of the question or indeed simply a quiet edit.

An accurate name encourages accurate use. The new use of Off Topic is for site-specific reasons. Some of those are because the question is about the wrong thing altogether (usual meaning of off-topic), and some are because we know we don't want that type of question (more likely to be called off limits than off topic). Together these are simply "site-specific reasons", and should be called that.

(I'd accept that "Off-limits" is a replacement which would more readily be substituted because it can be used in similar phrases "your question is off-limits because...", and we still avoid the problem with the word "topic".)

I haven't figured out whether I think the free-text custom reason should be described as "Off Topic", "Site-specific reason" or simply left completely open for the closer to type whatever they like. Given the uses we're planning on putting Off Topic to, I suspect we want to allow custom reasons that are not simply that the question is not about programming. Here again, Off-limits may be easier to use.

share|improve this answer
2  
You may be interested in Can we rename the off topic close reason label on the primary screen? if you haven't already seen it. –  Josh Caswell Jun 26 '13 at 18:50
    
@JoshCaswell Yes; thanks ...... and thanks. :) –  AndrewC Jun 26 '13 at 22:41
1  
Thank you for crystallizing the explanation of the problems with the "Off Topic" heading. –  Josh Caswell Jun 27 '13 at 19:22

There are two (of my) use cases for NARQ which I am not clear how they would be mapped to these reasons.

The first was mentioned by joran in a comment: the typo question. The question itself may be well formed with reproducible code and a stated error message. The answer is that there was a typographical error in the code: a missing semicolon, a misspelled parameter name, an incorrectly placed close of a code block, etc. These are legitimate questions (person asking had a real problem, well stated), with an answer that is helpful to the person asking, but the answers would not be helpful to anyone in the future because no one is likely to have that exact typo, and even if they did it would be impossible to find the appropriate question. Should these be closed under one of the listed reasons, or is it expected that the "other" reason would be used for these?

The second use case is questions where the answer ends up being that the problem is a result of a bug or error in the underlying program/library used, not in the asking person's code. The answer is then to update the underlying software to a newer version (which may not have existed when the question was asked because the question triggered finding and fixing the bug). This question is only marginally useful for future readers because, as time goes on, more and more people will have the updated version of the software which has the bug fixed. As with the previous use case, is this expected to go under the "other", should it have its own reason, or does it fit in which one of the already given reasons?

share|improve this answer
8  
But not all typo questions are TL: imagine someone who muddles up comma and semi colon in C++ for example. The compiler error message may not make sense to the coder, but searching SO could lead to someone's typo question and thus enlightenment. Only when the typo is really not going to help anyone (you declared Count but are using Cont) should it be closed - and offtopic feels a little strange for those since you don't know that until it's been answered. –  Kate Gregory Jun 20 '13 at 14:59
4  
For the second case, I just don't see why they should be closed. Framework bugs are actually rather likely to affect other people, and be interesting or helpful to future readers. Old versions of frameworks stay in (often heavy) use for many years for mainstream languages. –  Servy Jun 20 '13 at 15:02
1  
@KateGregory That is exactly the example I've been trying to articulate. –  Tim Post Jun 20 '13 at 15:13
    
@Servy Not in every domain, they don't. Questions about bugs in versions of R >1 year old are virtually useless. –  joran Jun 20 '13 at 15:13
4  
re bugs: You're suggesting closing a question now because it'll probably be less useful in the future. Using this reasoning, we would close a question because it uses something that someone filed a feature request to be deprecated. Until it's been deprecated for long enough that hardly anyone's using it, the Q/A have value. Similarly, "No, actually, that's a bug, work around it (this way...)" is useful. "Bug fixed in previous release; just update" is still a relevant and helpful answer. New bugs should be left for linking duplicates to, and current non-dupe bugs are also helpful. –  AndrewC Jun 20 '13 at 15:36
4  
@KateGregory I think you make a good distinction between different types of typo questions. Those where the nature of the typo can be determined directly from the error message do serve a purpose (since search on that error message leads to an explanation of how to fix the problem). They should not be closed. Where the error message is not informative of the root problem, or where there is no error but unanticipated/incorrect behavior, they are not useful to anyone else and closure is warranted. –  Brian Diggs Jun 20 '13 at 15:42
    
Obsolete seems a better way of dealing with it's-a-bug, once it is out of date. See this suggestion. –  AndrewC Jun 20 '13 at 16:11
    
@AndrewC The situation he was thinking of re:bugs was something that's relatively common in R, i.e. someone pops up with a well stated question describing some problem, no one can reproduce it, and then it turns out they're using a very old version of R or a package. The usefulness of these sorts of questions will vary between languages (a problem we have to live with on SO) but in R they really are quite useless to future Googlers. –  joran Jun 20 '13 at 16:52
1  
@joran If there are so many questions about old versions of R, how can you conclude they're not useful to other visitors? Would Obsolete not be appropriate? –  AndrewC Jun 21 '13 at 0:28
    
@AndrewC My only concern is that when I vote to close these Q's, that the reason description not be absurdly non-descriptive, that's all. –  joran Jun 21 '13 at 1:29

Here is a very simple scenario to consider. I would like to ask you a question: Build me a paper airplane. Was my question off topic? Did I ask a question?

Apparently there is consensus to use "Off Topic" as an umbrella reason to close questions. I don't agree that is a correct path to take because it will confuse people and cause scope creep for "subreasons".

However, we are there. <insert analogy to joining if not beating>. And at the very least I would like to agree with the Close reason "Off Topic" subreason #3 with regards to "assignments" (Requests for work). I agree with George when he said "we want people to have tackled the problem themselves", and believe that the wording could be improved to aim for that type of scenario.

3. "Questions regarding implementing features must demonstrate a minimal understanding of the technology, and include an attempt to solve the problem."

share|improve this answer
    
FWIW, this isn't something we just pulled out of thin air; folks have already been using OT in this fashion for quite a long while. The name may not be quite right, but naming is hard and this is what folks are already used to, so... –  Shog9 Jun 20 '13 at 19:50
3  
Also, I don't like "include an attempt to solve the problem" or any other euphemism for "post code". Code is important when the problem is in the code; as a proxy for demonstrating a basic understanding of the problem you're solving, it sucks. One of the most frustrating things I've seen with Too Localized was its use on questions that didn't include code ("not enough effort!") ...and then again on similar questions that did ("too specific to your code!") Damned if you do, damned if you don't... –  Shog9 Jun 20 '13 at 19:53
    
@Shog9 - With regards to using "Off Topic" as an umbrella, this is just my opinion. I completely understand the issue of going with what is the best option which this seems to be. –  Travis J Jun 20 '13 at 19:54
    
@Shog9 - As for the "post code" request, where does it say code in the text? I am talking about an attempt, either in psuedo, as a description of what they tried, something. I think that questions are sometimes too aggressively closed and that is unfortunate. That is another issue though. There are a lot of questions which come in that ask the community to do the work with only supplying the requirements. How else can we politely but forcefully close and prevent those questions? –  Travis J Jun 20 '13 at 19:57
    
Here's a question for you: do you feel this author has demonstrated sufficient effort? He's certainly gone a bit further than simply demanding a plane... –  Shog9 Jun 20 '13 at 19:59
1  
@Shog9 - Good example. The OP does not seem to be a help vampire to me. They have invested a lot of time into a feature already which seems pretty neat (By the way, very cool finding a coded airplane example). They have one caveat that they cannot seem to get working and ask for an approach. The answer provides a rather ingenious approach to folding content based off of css margins, but the answers aren't really what is in question. This does not feel like a request for work to me because the OP has done so much already and has a very small specific circumstance outlined. –  Travis J Jun 20 '13 at 20:06
2  
that's my take too (the title could use some love, but it's not a show-stopper). And this is what I'd like to encourage: questions that simply show that the asker has thought about the problem a bit. –  Shog9 Jun 20 '13 at 20:12
    
@Shog9 - I would actually like to see some effort, not just thought. Although the OP in the example did not provide the exact work produced (which isn't always necessary), he provided his process that he was using already. It definitely showed effort. I feel like many times people get an idea and then post "how can I implement this idea". There can be a lot of thought behind it, but it should be up to them to actually implement the idea, not the community. Without any effort, there is nothing we can do. If they get stuck then that is where we can step in and help them out. Is that too strict? –  Travis J Jun 23 '13 at 23:05
    
Define "effort", @Travis. Where do you draw the line between thought and effort when both are expressed via text typed into a computer? –  Shog9 Jun 23 '13 at 23:39
    
@Shog9 - In my opinion: It definitely depends heavily on context. In an algorithm, defining what your train of thought was could be seen as effort. This would satisfy the "attempt to solve" clause listed in my answer. However, when talking specifically about a programming question where code is involved, I think that it is very easy to define effort. Effort: Either code was shown or the approach was outlined. No-Effort: No code was shown, and no approach was outlined - even though a solution was requested. –  Travis J Jun 23 '13 at 23:47
    
@Shog9 - What do you think of this: stackoverflow.com/q/17266288/1026459 ? –  Travis J Jun 23 '13 at 23:57
    
I think it's fine, @Travis. I would prefer the JSFiddle code was embedded, but in this case it's not essential to the question. –  Shog9 Jun 24 '13 at 1:20
    
Oh, and @Travis: regarding showing effort for code you're working on... That's the entire focus of reason #2. In one of the first drafts for this, there was a lot more overlap between #2 and #3, but I backed away from that a bit (and may go further yet) because they're separate (though related) problems and the last thing I want to do is encourage more folks to post code just for the sake of satisfying an arbitrary requirement (I've seen too many questions made worse by the inclusion of a block consisting of nothing but an empty function or program shell with "logic I need goes here" in it). –  Shog9 Jun 24 '13 at 1:23
    
@Shog9 - I don't think code should be a requirement. There should at least be an explanation of the attempt though. I guess that is the "tell us what you tried" part of #3. Out of curiosity, how far away are we from these changes being published to stackoverflow? –  Travis J Jun 24 '13 at 4:39
1  
What do you mean "requests for work"? "Please write me a Facebook?" no, that's not ok. "How can I display a dragging element without resorting to dirty positioning trick?" sure. "Please code me up a complete JavaScript drag'n'drop library that is subtly different from the 400 existing libraries found easily via Google and don't tell me to use jQuery I have religious objections to the letter 'Q' kthnxbye" - not ok. BTW: this is getting long; please either break it out as a separate discussion, or join me in the tavern. –  Shog9 Jun 24 '13 at 19:36

Can something to cover reverse engineering questions be added?

Questions like:

I'm trying to X as seen on site/program/app Y. How do I do it?

These questions tend to get down-voted and closed often enough, see examples here.

I've raised a question about adding these to the list of "What types of questions should I avoid asking?"

The response to that question was:

There is a close reason being refined that deals with this, basically saying that questions that just point to third party code without including any specific bits and end in 'how do I do what they did?' aren't on topic. It's not yet baked, so I'm not yet officially answering - but this is one of the cases we considered when coming up with the initial set of close reasons for Stack Overflow. – Tim Post♦ Jun 13 at 1:04

share|improve this answer
2  
I don't exactly see a whole lot of these. Certainly not more than any of the ones listed above. Note that "Unclear" pretty much applies to most of these types of questions (NARQ did before that). If there isn't enough information in the question itself to answer it, that's the GOTO response. –  Servy Jun 20 '13 at 14:03
2  
@Servy If George Stoker's recommendation is implemented, I would guess these could fall under "Questions must demonstrate a minimal understanding of the problem being solved." –  apaul34208 Jun 20 '13 at 14:06
1  
Sure, that could fit as well. –  Servy Jun 20 '13 at 14:07
1  
@Servy the problem is that many of these questions can reasonably be answered. It just requires the answerer to do all the leg work and they often rely entirely on third party links. –  apaul34208 Jun 20 '13 at 14:09
    
It's the "rely on external links" that would qualify it for closure. If the information needed to answer is isn't in the question then it's lacking, for the same reason that an answer that's just a link isn't really answering the question in the eyes of the site. –  Servy Jun 20 '13 at 14:11
    
I don't see a problem with these questions when they include enough information to be answerable - a screenshot, description of the desired functionality, etc. For instance: stackoverflow.com/questions/8657894/… –  Shog9 Jun 20 '13 at 16:51
    
@Shog9 I was referring more to this type: stackoverflow.com/questions/16914729/…. The question you pointed to is the first time I've seen one done well. –  apaul34208 Jun 20 '13 at 16:57
    
"Unclear" is fine for that, @apaul34208. It is an appallingly bad question regardless of topic. FWIW, I picked that example specifically because folks ask duplicates of it all the time - often very, very badly; therefore, the existence of one good version provides an instant close reason for those that follow. –  Shog9 Jun 20 '13 at 17:00
    
@Shog9 I think even the question you pointed out, could be flagged with #3 as it doesn't really "demonstrate a minimal understanding of the problem being solved". Had it included some code and/or an attempt recreate what they were after, it would probably be fine. Don't these types of questions generally point to "code this for me" or "research this for me". –  apaul34208 Jun 20 '13 at 17:31
    
@apaul34208: I don't think being forced to include code when you aren't asking for code to be written for you makes any sense at all. The author described - using words and pictures and even a video - the behavior he wanted to replicate, indicating he had a basic understanding of what was happening; that should suffice. –  Shog9 Jun 20 '13 at 17:53
1  
@Shog9 after reading the comments on Travis J's answer, I think I've got a better idea of where you're coming from. Its the "build this for me" with just a link, that I guess I'm really tired of. You're right those questions are easily covered by other flag reasons. –  apaul34208 Jun 20 '13 at 21:47
    
"the problem is that many of these questions can reasonably be answered. It just requires the answerer to do all the leg work and they often rely entirely on third party links." ...exactly. That makes it completely on topic, and problematic for other reasons. –  AAA Jun 24 '13 at 1:55
    
@djechlin I've already conceded the point see the comment above yours –  apaul34208 Jun 24 '13 at 2:19
    
@djechlin - "That makes it completely on topic, and problematic for other reasons." There are no longer any reasons other than topicality for this scenario because of the changes to the close mechanism. Can you expand on what you meant by "problematic for other reasons" please? –  Travis J Jun 24 '13 at 5:56

I didn't know about this post, so I wrote a review in a separate post. Summary based on opinions here, my post and the comments to it:

  • Options describe the specific problem and demonstrate a minimal understanding of the problem being solved should be moved as sub-options of unclear what you're asking, as they are not "off-topic". Renaming "off-topic" to "off-limits" or "site-specific reasons" is probably okay, but I'd rather see a proper hierarchy of reasons.

  • An alternative website should be added for tool or library recommendations option. That is, if there is any good website for comparisons like on Wikipedia ("C# vs Java", "Text editors" etc.) It should be a sub-option of primarily opinion-based, because that is exactly what it is. The linked question is not broad enough, Why are “shopping list” questions bad? is more general.

  • A list of another sites in the SE network should include Server Fault, for consistency. Maybe it should also include other websites, but I don't know statistics (maybe put all technology sites into the list?). This is a duplication, so the outcomes should still be the same, regardless of how SF or SU were recommended (I don't know how it works now).

  • Period is missing in two of the options.

Additionally, after reading the post:

  • "Off-topic" list is much more verbose than the list one level higher, which uses short and concise wording (duplicate, off-topic, too broad) with more detailed explanations in grey text. Compare this with:

    Questions concerning problems with code you've written must describe the specific problem and include valid code to reproduce it. See SSCCE.org for guidance.

    It's totally not "describe the specific problem", it's "no code to reproduce" or just "need code".

    Questions must demonstrate a minimal understanding of the problem being solved. Tell us what you've tried to do, why it didn't work, and how it should work. See also: Stack Overflow question checklist

    Some folks hate the wording, but it's essentially "What have you tried?" -- no more, no less. Wether these exact words are displayed when a question is "on hold" is debatable, but I think it's perfectly valid as a short and concise title for the option.

    On the whole, formatting of the list should follow the formatting of the list one level higher.

share|improve this answer
    
Hahaha, downvoting without comments — now on meta too! Great. –  Athari Jun 26 '13 at 12:37

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