34
\$\begingroup\$

I have no doubt that this thread will get many down votes, but lately I often find questions that are closed with "Closed as too localized". What is the point of a Q&A site if questions cannot be answered because a group of fortune-tellers decides it will be not relevant for anyone in the future? Guess what, some of them I actually find relevant and would not think that I am alone with this. On a pure wiki styled site this procedure would make sense, but here, where people come to get answers for their problems I do not think so.

While some finds a question unworthy to even read through and votes to cast it to the seventh hell, others write really good answers for them that actually helps others. (eg. https://gamedev.stackexchange.com/a/54497/11909)

What makes it worse is the fact that these questions does not even get deleted, simply just prevent others from answering it. I am not saying that the whole closing thing is just bad, there are some really stupid questions out there, but closing something with the cause of it only helps one person is just bad. I do not really come to this site to ask totally irrelevant questions for me that would help others, but to find help for my problem, and if I help others with it, that is just icing on the cake.

\$\endgroup\$
  • 1
    \$\begingroup\$ I agree, the too localized rule should be loosened. \$\endgroup\$ – user15805 May 9 '13 at 10:01
  • 1
    \$\begingroup\$ +1: Very recently I argued that "Too Localized" was the worst of all reasons to close question and I'd rather we didn't have the option. \$\endgroup\$ – William Mariager May 9 '13 at 13:55
  • 2
    \$\begingroup\$ The linked question doesn't appear to have ever been closed. It doesn't even have any close votes right now. Can you edit the question to include a few examples of the "these questions" which you're saying ought not to have been closed, so that we can have a substantive discussion? \$\endgroup\$ – Trevor Powell May 10 '13 at 3:14
  • 2
    \$\begingroup\$ @Tetrad has clarified that the linked question was previously closed, and he reopened it (apparently in a way which left no traces in the moderator tool). So never mind. :) \$\endgroup\$ – Trevor Powell May 10 '13 at 23:27
  • \$\begingroup\$ It's always tough for me to determine what to do with pending "too localized" close votes in my review queue. Some questions are clearly "debug my code for me" and ask the Internet as if it was a glorified debugger, but the line is blurry and most of the time I just can't make my mind and skip. Maybe "too localized" is too imprecise. \$\endgroup\$ – Laurent Couvidou May 10 '13 at 23:55
  • \$\begingroup\$ I think there is a really worthwhile question implied here, RE: what we want to consider to be "too localised" on gamedev.SE. But I don't think that discussion is going to happen in this meta question, since this question has been written as an angry rant, rather than as a question to be answered. So people are just responding to the rant (or more commonly, not commenting at all), rather than addressing the underlying question. Maybe worth phrasing this as an honest question, either as an edit or as a fresh question. \$\endgroup\$ – Trevor Powell May 12 '13 at 11:14
  • 2
    \$\begingroup\$ @AlexM.: I would go further, even. Too Localized doesn't need to be "loosened"; it needs to be removed entirely. It's the ultimate in arrogance; to say "I don't see how this could apply to other people, therefore there is no way it can apply to other people" is pretty darn close to claiming omniscience. You don't even know how many times I've been Googling for an answer to some obscure programming problem, found a question on SO that fits the issue perfectly, and then seen it closed as "Too Localized" with no answers. \$\endgroup\$ – Mason Wheeler May 16 '13 at 16:00
14
\$\begingroup\$

You didn't list any particular questions as examples (the one question you link to never received a close vote), so I'm going to list some recent "too localized"-closed questions and discuss them. Here is a list of every question closed as "too localized" during the last 48 hours or so (as of when I posted this answer):

That's six questions closed as "too localized" during the last two days (for comparison, there were eleven others closed for entirely other reasons during the same period).

That percentage does seem higher than maybe it should, so let's look at the questions more closely, one at a time, and decide whether that's valid for each:

Rectangle collision is not working correctly

Maths bug in collision response code. The asker provides a big block of code to sift through, including reading keyboard input and etc. On the plus side, the question provides a clear explanation of exactly what is going wrong.

I'm not a huge fan of big code-dumps, personally. But this one at least is reasonably well formatted and it's easy to skip past the obviously irrelevant bits. I guess my concern is that fundamentally, this is a "find the math bug in my code" question, rather than one which addresses a concept which can be explained or a technique which can be learned. See my comments re: discoverability of "solutions to a math bug" questions under the "arcball" question, below.

Has an accepted answer.

creating arcball in cocos 3d

This is another "help me my code isn't working and I don't know why" post. Which probably isn't tremendously surprising as it involves quaternion math, which isn't a trivial thing for anybody.

The question is a huge code dump, and includes a download link to get even more code. If this had focused down on "what is the right way to do quaternion math", then I think it could have been a really good and useful question (if there isn't already such a question), but as is, I have trouble mustering enthusiasm to go delving through somebody else's math code just to go searching for an arithmetic bug.

I'm not certain how finding this guy's maths bug and indexing the particular located bug in Google helps the Internet as a whole; the next guy will presumably have a different maths bug, and we'll just wind up answering this same question again and again, just with different swapped variables or subtracts that should have been adds in each new asker's version of the code. So I'm inclined to agree with the "too localized" evaluation. Again, my view is that this isn't somebody asking for something to be explained, or to be taught; he's just asking us to find his maths error for him.

Maybe that view is a personal failing of mine.

2d array bounds

I have trouble parsing the question, here. It appears to be another plea for someone to debug their code for them. Whatever the question actually is, even if someone was motivated to go delving through source code to find the bug, the question is fundamentally unanswerable, because not enough code is provided. My instinct is that this question, if pursued, would have turned into a "teach me how to program" question.

But I wouldn't have voted "too localized" on this question. I'd have voted "not a real question", myself.

OpenGL noob: Using VBO to draw a colored triangle

This one doesn't look terrible to me. At least the problem is well-defined, and the source code has been trimmed down to a somewhat reasonable amount for people to peruse. Fundamentally, it's still a "I haven't checked for existing learning resources" sort of question, but I think this one could have been salvaged.

The real question from my point of view is whether someone else would ever have found it -- the asker's misunderstanding was regarding the precise usage of glBindBuffer(), which is fair enough. But none of that shows up in the question text -- there's no way for another user to find it. Once the real solution was found, I wouldn't have been against someone editing the question to really tighten it up to fit the actual problem, to aid in discoverability.

I suspect that this one was closed due to its very low prose:code ratio, which is usually a strong indicator of questions which are only of use to one person at one moment in time.

Has an accepted answer.

Random AI movement on tiles

I'm having trouble understanding this question. I'd have asked the poster to rephrase in an edit, and voted "Not a real question" in the meantime.

My logic in collision detection in diagonal code correct?

This one doesn't describe the problem and doesn't explain what needs to be fixed. The question appears to be "Is it necessary to have a diagonal collision for the wall?", which there's no way for us to know without a lot more context (what the game is, how big the wall is, where it is, how thick it is, how far game objects move in a frame, etc). I think I'd have gone with "Not a real question" on this one as well; I don't think the question is defined enough to be "too localized". (Although I can see an argument that if nobody else can understand the question, its answer is probably not going to be of much use to them, either.)

My conclusion

Just from my examination of the last 48 hours, I guess I'm not really seeing a problem. Granted that there are some questions which I wouldn't myself have classed as "Too Localized", but many of them I would have voted to close for other reasons.

I was a little surprised that there were only these six "too localized" questions in the last two days -- I've had the same sense as the OP that there have been lots and lots of extremely localized questions being closed on the site, but that doesn't seem to be borne out by this data. Having only six closed in two days surprises me a lot. And further investigation suggests that this rate doesn't seem to be unusual -- there were only six closed for that reason in the previous two days, as well. And even fewer in the two days before that.

It's also interesting that both of the questions which I considered to be potentially answerable did get answers, and the original poster of each accepted those answers, even though the questions were later closed as "too localized". So the whole "too localized" thing doesn't seem to actually be stopping people from providing personal assistance when the questions asked are truly answerable. (as per my admittedly subjective opinion re: answerability).

Maybe a wider analysis of more data would show a different trend, but just going on the basis of the data from the last 48 hours, I'm not seeing anything particularly alarming in regards to questions closed as "too localized".

\$\endgroup\$
  • \$\begingroup\$ Note that I got my numbers and links from the moderator tools, which probably weren't designed for data mining like this, so it's quite possible that I've misinterpreted the tools. If somebody more familiar with the moderator tools could verify that I haven't missed any "Too Localized"-closed posts over the past 48 hours, I'd be very grateful! \$\endgroup\$ – Trevor Powell May 10 '13 at 11:20
  • \$\begingroup\$ I'm surprised by that data also, especially since there are scenarios like this one. There, the majority close vote was too localized (3 votes) but there was one "not a real question" and my own "not constructive" vote, which killed it. But only the majority reason is displayed. So I'd expect a potentially slightly higher number of "too localized" votes. \$\endgroup\$ – Josh May 10 '13 at 19:14
  • \$\begingroup\$ I reopened the question he linked to so it removed the close votes. \$\endgroup\$ – Tetrad May 10 '13 at 19:50
  • \$\begingroup\$ @Tetrad Okay, that's interesting -- I'd gone into the moderator tools to check that to verify that it hadn't been reopened . The question doesn't show up in the "Recently reopened" list, for me. If not everything that's been reopened is showing up in the moderator tools, then it seems likely (or at least plausible) that there are questions being closed that aren't being reported in the "Recently closed" area, too.. So the whole "six across two days" statistic may therefore be completely wrong, for all we know. \$\endgroup\$ – Trevor Powell May 10 '13 at 23:21
7
\$\begingroup\$

While some finds a question unworthy to even read through and votes to cast it to the seventh hell, others write really good answers for them that actually helps others.

No, that's not how it works.

The purpose of every Stack Exchange site is to help build a useable, searchable knowledge base. The fact that a question has "really good answers" is completely useless if the only people who find those answers are the kind of people who sift through a garbage dump looking for a peach pit.

Questions are how users with a problem find the solution (ie: answers). They do a search based on their problem. The search reveals a question that matches their problem. And therefore, the answers to that question will match their problem.

That's how Stack Exchange is supposed to work.

If a question is too localized, then what it means is that nobody's problem description will ever coincide with that question. So even if an answer to that question happens to be the solution of that person's problems, they will never find it.

That's how Stack Exchange stops working.

Basically, answering too localized questions is a waste of time. Nobody's going to find them except the person asking the question.

Just take a look at your example. If I were having timing problems that the accepted answer would solve... do you honestly think a Google search or a GameDev.SE search would ever lead me to this question? Well, it might. It will probably also lead me to about 50 other timing questions that won't help me.

Buried under the avalanche of choices, I don't get helped.

The OP lacks any real detail. He basically said, "I'm having a timing problem. Please debug my code." That's not something Google is going to be able to find.

And that's bad.

What makes it worse is the fact that these questions does not even get deleted, simply just prevent others from answering it.

Closed questions can be automatically deleted if they don't have answers.

Remember: the point of closing a question is to prevent people from wasting their time by answering a question that is currently inappropriate. A question so narrowly stated that it will never be useful for anyone else is inappropriate. To answer it helps only one person, and therefore is a waste of the person's time.

We don't need to delete a question that was answered before it was closed. It can still be here in the off chance that someone decided to look at all 50 of their search results and happened upon useful information by sheer dumb luck. But we don't want to encourage that.

\$\endgroup\$
  • \$\begingroup\$ In defense of that OPs example question, it might be better suited to edit the title to make the source of the problem more easily searchable. A lot of people don't get how to write good question titles and simply improving them helps the question dramatically. \$\endgroup\$ – Tetrad May 15 '13 at 14:34
  • \$\begingroup\$ I feel that a great counter example is the following question - stackoverflow.com/questions/8176493/… I encountered this exact same issue and it has been viewed 1600 times. Why is this closed for being to localized? 'To Localized' always feels like moderator abuse. At the very least closing threads as to localized should only be allowed for threads in the negative vote range. \$\endgroup\$ – NtscCobalt May 16 '13 at 20:04
  • \$\begingroup\$ @NtscCobalt: Two things: 1) You might notice that this question was closed by a single moderator, not 5 people, and thus is more likely to be erronous. 2) Hurray, in all of the five million questions on Stack Overflow, you found one that may have been closed as Too Localized incorrectly. The claim here is that Too Localized is never a good idea. The fact that you found one question that helped you is not proof of anything. 3) It's closed, not deleted. \$\endgroup\$ – Nicol Bolas May 17 '13 at 2:08
5
\$\begingroup\$

I agree, people are sometimes a little too gung-ho with that close vote.

Some questions I don't understand why people vote to close. Especially if there isn't really any explanation given or even down votes (question down votes are free these days so it's not like you'd lose rep).

Yes, at the end of the day, most questions are going to involve someone's "code being fixed", but there's a big difference between someone who doesn't understand simple logic or the fundamentals of the programming language they are using and someone showing us their code and describing a "long tail" problem. Having a block of code that needs some changes does not mean the question is too localized.

My guess is that people are using the review queue, see that one person cast a close vote, and see that close vote as legitimate and pile on it so they can get their review badges or something.

Feel free to cast reopen votes or even flag questions you think were unnecessarily closed.

\$\endgroup\$
  • 4
    \$\begingroup\$ You don't need to actually vote to get the counter to increase for the badge, though -- "Leave Open" still increases your review counter. I do think there's a psychological issue -- I find myself more likely to vote in agreement with questions with 2 or 3 votes, but less likely to vote if I'd be the deciding vote, for example, which probably indicates a bias. It's also really hard to judge what "too localized" means. Perhaps the description in the reason could be improved. \$\endgroup\$ – Josh May 10 '13 at 19:18
  • \$\begingroup\$ I like the point made in the "long tail" post : "That’s why we actually don’t mind having several versions of every question, where there are variations in wording or circumstances. The more chance that someone types a question into Google and finds their exact question already answered, the better a job we’ve done." We should almost promote a certain level of localisation when it comes to questions \$\endgroup\$ – Tom 'Blue' Piddock May 14 '13 at 14:05
2
\$\begingroup\$

I think this happens to mostly 2 classes of questions.

1) The longwinded "do it for me" type questions, where they post [pages] of code and kinda shrug their shoulders and ask "So what's wrong with it?" Or, they post just a smidgen of code, then ask something which nobody could know without browsing the whole codebase. (The latter question type comes from noobs who think because "You know C++" you have a mind-body connection with every CPU register on the planet).

These types of postings are from possible help vampires, anyway, so methinks we do well to close them.

2) The other type of "closed as too localized" happens I think somewhat wrongly. It happens to these "wrongheaded" questions that seem to promote some kind of agenda/have an element of show-offness ("Imma program in C!") and possibly unsettles the moderator folks. I don't think we need to close those, as an answer responding to the underlying bad assumption or misunderstanding the OP has, is a lot more productive there.

Closed as vampiric

enter image description here

\$\endgroup\$
1
\$\begingroup\$

Generally when parsing a question I rely heavily on first impressions - I'm a busy person and don't have time to disentangle what looks on the surface like an awful question to see if there is anything of value in it. The responsibility is on the person asking the question to make it a good one.

Like the others answering here, I tend towards "too localized" for "please debug my code for me" questions. These questions generally have many (but not necessarily all) of the following features:

  • A very poor text-to-code ratio, often with a huge code dump that one must scroll through to find the the relevant parts.
  • Little in the way of indication that the person asking has tried anything themselves to resolve the problem; they may have hacked and slashed at the code a bit, but that would be the limit.
  • Little in the way of any coherent description of what the problem actually is - often not much more than "this doesn't work". No indication of how it doesn't work, what the OP expects to happen, what actually happens.
  • Code that's poorly formatted, frequently inconsistently formatted, and appears to have been copy/pasted from several different sources (it's occasionally possible to actually track down some of those sources).
  • Some fundamental misunderstandings about how what they're trying to do actually works.
  • A "just give me the code!" attitude; I recall one from some years ago where the person asking literally said "don't give me concepts, give me code!"
  • Sometimes an insistence on use of an inappropriate solution for or approach to the problem they're trying to solve (if that problem can be even determined from the question).

On they're own any one of these is probably salvagable from the "too localized" bin, and may even become a good question with heavy reworking, but when they come together it becomes difficult to see much hope. The ultimate arbiter of whether or not such a question goes into "too localized" is therefore: "is any possible answer going to be of any possible use to any possible person besides the person asking the question and in the very specific circumstances they are in now?"

If the only answer to that is "no" then it's "too localized" for me.

\$\endgroup\$

You must log in to answer this question.

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