What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 127 Stack Exchange communities.

I find that of all the "close" reasons, "too localized" is the one with the highest ratio of "times it applies," to "times questions are closed because of it." Which leads me to believe that the community doesn't actually respect the description of this close-reason, which makes me think it needs an update.

TL;DR

I will vote to close as TO in the cases where the question is too localized based on reasons of time or place, but most other very specific problems are probably useful to someone else somewhere.

Quoth Shog9:

So far as I can tell, "too localized" was intended for questions that quickly become effectively unanswerable anyway. The description doesn't quite reflect this, and we'll be either changing it or removing that close reason entirely in the near future...

Moving on:

The description reads:

This question is unlikely to help any future visitors; it is only relevant to a small geographic area, a specific moment in time, or an extraordinarily narrow situation that is not generally applicable to the worldwide audience of the internet. For help making this question more broadly applicable, see the FAQ.

I find this very obviously applies to many, many questions that are about "the error in 'my' code." To be concrete, this is the question that spurned me to post this, although see my activity > close votes for many examples:

Quicksort Algorithm not assigning pivot correctly

I do see why this is subtle. The question in this post is, arguably, a good question, and the fact that it currently stands as +1/-0 votes means the community does feel it is, arguably, a good question. It's precise, well-researched, specific - but yet in my sound judgment completely impossible to help anybody implement quicksort who hasn't made the exact same mistake in this code. For evidence, let's just do an SO search for implement quick sort:

And so on, and so forth.

On the one hand, I would like to argue that the community should be stricter. I think if a question is too localized, it demonstrates that the OP has not put enough thought or research into the question to formulate it in a way that is general and hence researchable. Once a question is formulated generally enough that it can be of use to others, by the same principle in reverse it's probably already on SO, just wasn't searchable back when it was a problem in "my" code. For a problem in a basic C/Java/C++/etc. algorithm implementation, very often the actual problems will be things like misuse of for or if loop, and once someone kindly instructs the OP on how to use a debugger in a few sentences, the problem unveils into a more specific, researchable one.

On the other hand, it is clear to me that the community's actual behavior is out of sync with the description of this tag, so one of a few things needs to be done:

  • At minimum: Questions that currently stand as too localized need to be permitted, since that's what the community does.
  • Ideally: The community needs better resources with which to close such questions. Debugging a question such as the linked one is a two step process: one, narrow down the error to an irreducible problem such as improper use of if or for, assignment of variables, or even "code is too complicated to understand and should be factored into functions cohesively," which is in a weird way an irreducible problem. I should be able to close as "improperly debugged" - perhaps a subset of too localized - and have an automagic link to the expected steps the OP must go through as part of diligent research effort. Does such a guide exist on SO? I may be ignorant / ironically bad at searching for such things, but even if so it needs to be publicized more appropriately, e.g. in the description of too localized that may appear when a question is closed.

e.g.:

This question is unlikely to help any future visitors; it is only relevant to a small geographic area, a specific moment in time, a specific code fragment that does not serve to illustrate more general issues, or an extraordinarily narrow situation that is not generally applicable to the worldwide audience of the internet. For help making this question more broadly applicable, see the FAQ or .

Obviously the second is ideal in my opinion but the status quo does not seem acceptable to me, as the reasons to close a question as too localized are ignored by the community very often.

share|improve this question
6  
So, basically you're advocating closing about 50% of the questions on SO as too localized? I tend to apply a stricter guideline. If someone's missing a brace/semi-colon/random punctuation mark/misspelling of choice then that's too localized. If someone has larger problems with their code, whether or not someone is going to write the exact same code again, it's not too localized. –  ben is uǝq backwards Feb 23 '13 at 22:16
1  
@benisuǝqbackwards no, please read my post. –  djechlin Feb 23 '13 at 22:19
4  
I have... I agree with most of it but I do not agree that it's going to work the the reason stated in my comment. You'll have to close about 50% of all questions; everything that comes with an error message could be assumed to fall within your definition. If I'm reading your post incorrectly (definitely possible!) maybe you could explain why? –  ben is uǝq backwards Feb 23 '13 at 22:24
    
@benisuǝqbackwards firstly, I am not saying 50% of questions should be closed as too localized; our description of "too localized" says 50% of questions should be closed as too localized. Secondly, I think 50% is a wild overexaggeration. Thirdly, I am saying it is possible to close many of these questions in a way that maintains the current meaning of "too localized" while being constructive for the OP, ideally by linking them the resources to reduce their "too localized" coding problem into an "exact duplicate" question. –  djechlin Feb 23 '13 at 22:30
    
@benisuǝqbackwards specific code that reveals a general situation is important but subtle to formulate well in a policy, I'll try to revise my post to address this in a bit. –  djechlin Feb 23 '13 at 22:35
    
and the fact that it currently stands as +1/-0 votes means the community does feel it is a good question That's hardly conclusive. Almost every question gets at least one pity upvote from the review queuers, no matter how bad it is. And you're looking at just one –  Lightness Races in Orbit Feb 24 '13 at 10:27
    
@LightnessRacesinOrbit dropped the word "arguably" on that. Was more saying it's clearly too localized based on current defn yet wasn't rejected outright. –  djechlin Feb 24 '13 at 11:06
add comment

3 Answers

up vote 18 down vote accepted

You shall not muzzle the ox when it treads out the grain.
                                                                    --Deut. 25:4

For all the high-minded language we throw around here sometimes, the truth is that most of SO is built on some form of enlightened self-interest. People ask questions because they have a real problem they need to solve, often so they can complete a job and get paid for it. People answer questions for all sorts of reasons, many of them somewhat self-serving as well.

And yet somehow this all works to create artifacts that are valuable well beyond the situations that originally motivated them.

Well... Some of the time it does. There are also plenty of questions and associated answers that may never again help anyone. But as Rosinante notes, you can't always know ahead of time what's going to be useful.

If we adopted an interpretation of Too Localized that shut out any question specific to the asker's situation, that would shut out many - probably even most - of the questions asked on SO. It would make writing a good question even harder than it already is, and discourage the people writing answers.

And for what purpose? Search engines have gotten really good at letting us sort through massive amounts of relevant information; a well-written, broadly-applicable answer to even a localized question may well prove useful to someone else even if the actual question was only a rough match for their problem. And the real beauty of SO is that questions aren't set in stone - you can always go back and edit them to make them easier to find, more generally-applicable, less clogged with irrelevant details.

Save "Too Localized" for stuff that cannot possibly benefit another person: "oops, forgot a semicolon" / "oh, I found the error - it was in a library I didn't mention I was using" / "my question is a needle in a haystack on another website", that sort of thing.

We've discussed changing the wording for Too Localized quite a bit over the years, both internally and here on Meta. But the real problem with it is, it doesn't accurately represent a real problem in many cases. The proper solution is to replace it with something much more specific to the actual questions that should be closed: too broad, too narrow, obsolete, etc.

share|improve this answer
    
I disagree that most of the questions that seem too localized, might turn out to not be. A good answer to many of these questions I could copy and paste, and it would consist of 1) run this in gdb (yes, usually C, C++ and sometimes Java questions), include some link to getting started with gdb; 2) put the onus fully on the OP to create an MRC because i. it's relatively rote to do this even for very novice programmers and ii. a great way to teach someone to learn to fish instead of catch a fish. Those should be closed, ideally with resources on how to do those steps. –  djechlin Feb 23 '13 at 22:52
5  
You just described how to answer a question in a way that could easily help others facing the same general problem... And then cite that as a reason for not wanting to answer the question? On forums, "segfault - HALP!" questions are a scourge, because no matter how well you try to teach the asker how to fish, there are a thousand hungry people following in his wake who'll never find or follow your conversation. On SO you can not only find it, you can edit it to reflect the actual problem instead of burying the answer under three pages of back-and-forth. –  Shog9 Feb 23 '13 at 22:53
    
Very fair. By now I might just be asserting that "it would be nice if" SO included a standard resource that would consist of that boilerplate - which I may formulate and propose separately. Perhaps I could just garner a bunch of rep in the meantime by maintaining my own though. Next I thought "but there's too many duplicates," but if none of the duplicates include what, with respect to my opinion, is that good answer, then they're not really good duplicate questions with respect to that same opinion. –  djechlin Feb 23 '13 at 22:59
3  
My top-ranked answer on SO is a response to both a specific problem, and one that could've easily been answered by just reading and understanding the documentation. There are thousands of answers like that on SO... And thousands of people looking for answers like that on SO. It's actually very hard to have a unique problem when you're operating on Stack Overflow's scale. If you see a need for a good answer, write one; that's usually a whole lot more fun than closing anyway. See also: blog.stackoverflow.com/2010/11/… –  Shog9 Feb 23 '13 at 23:01
    
Okay, so it sounds like you're claiming nothing is too localized and we should get rid of that category - which is consistent with my claim that the description is out of whack with actual community standards. –  djechlin Feb 23 '13 at 23:03
    
Re. duplication I think it's good to close but not delete questions as duplicates. And that this feature really needs to be implemented completely consistently, albeit unintuitively so, with the spirit of that blog post - meta.stackexchange.com/q/164620/183887 i.e. duplication sounds pejorative but really should be thought of as doing the good research effort of linking two strongly related questions. –  djechlin Feb 23 '13 at 23:06
1  
Not quite - I'm saying there are tons of Too Localized questions, but identifying any save the most egregious is a real crap-shoot. Of the questions I've closed as Too Localized today, all either referenced problems on an external website (which no longer exist once the problem is solved) or were solved by the asker in some way that invalidated the entire question ("I rebooted and it worked"). –  Shog9 Feb 23 '13 at 23:10
    
Regarding rep for duplicates: savvy folks write good answers and then go out and find duplicates of the question they answered - thus driving more voters (and rep) to their answer. This is a win-win for both the askers and the answerers. –  Shog9 Feb 23 '13 at 23:12
    
let us continue this discussion in chat –  djechlin Feb 23 '13 at 23:13
    
Too localised is just fine. Keep it please. –  Lightness Races in Orbit Feb 24 '13 at 10:28
    
+1 for For all the high-minded language we throw around here sometimes, the truth is that most of SO is built on some form of enlightened self-interest. –  Adam Rackis Apr 5 '13 at 17:40
add comment

Let me give you a counter example. Today I wanted to refresh my knowledge of named pipes. I searched and found C++: Implementing Named Pipes using the Win32 API - this question was highly specific to someone who didn't know that the pipe had to be created with a pipe-specific API, then could be treated like a file. I don't have that misconception. Nonetheless, the question and its answers were useful to me (and if any of the people involved were wondering why they got an upvote almost 3 years later, that's why.)

In fact, I have a stream of mystery upvotes from questions that at first glance couldn't possibly help anyone else: How can I tell if I have Service Pack 1 for Visual Studio 2010 installed? for example. Or Why is my configuration manager disabled in Visual Studio 2010? - lots of people have this problem, I guess.

If you think the question is too localized, edit it to be less so. On The Workplace, "which of these two job offers should I accept?" is too localized, but "how can I decide between two competing job offers?" might be ok. Looking at my two examples, "Do I have SP1 installed?" is too localized, but "How can I tell if I have SP1 installed?" is much better. "What is wrong with this code?" is too localized, but "what causes this error message?" is not. Others may make that mistake and get that error message.

About the only close-and-stay-closed Too Localized I see are on travel where people ask what something will cost - eg a flight from Toronto to Paris on March 10th. Even that you might rescue into "How can I find inexpensive flights?" and then use that trip as an example.

share|improve this answer
    
Not a counter example at all - "specific to someone who didn't know the pipe ..." is general to people who "don't know the pipe..." You just called it specific anyway. Good judgment on the part of the would-be close-voter can reveal this; and I am still claiming that there are many usually basic C/C++/Java questions that come down to a bunch of if block, for loop, bool operator, etc. issues and furthermore that these questions are left open not because they might turn out to be not too localized, but because the community just doesn't like the current definition of too localized. –  djechlin Feb 23 '13 at 22:56
add comment

A middle way is to ask: "Is this person's situation likely to shed light on someone else?" One tricky aspect is that you can't always tell until an answer emerges. If dialog between the OP and others reveals some purely local issue, well, then it's too localized.

However: A bit of editing can sometimes make a big difference. Ironically, some people pose questions that seem to localized but, in fact, illustrate a perfectly reusable situation. Often, the most important edit opportunity in this case is the title.

I agree, though, that many questions that arrive deserve to be closed as too localized. I cast many close votes.

share|improve this answer
1  
I agree - most too localised questions are too localised because the OP didn't abstract enough to take the problem from the "what is wrong" stage to the "how does this language feature work" stage. Ultimately, though, if said OPs did this, then all such questions wouldn't exist to begin with. So, close as "too localised", job done. –  Lightness Races in Orbit Feb 24 '13 at 10:28
add comment

You must log in to answer this question.

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