What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 128 Stack Exchange communities.

I often find myself asking a Stack Overflow question about how to perform a very specific operation X.

X attempts to accomplish a single, specific task, in a specific way, using a specific tool.

X is part of one method of accomplishing larger task Y, which in turn is one way of achieving larger, more nebulous goal Z.

I may already know of two or three ways of accomplishing Y that do not involve X, and two or three ways of achieving Z that don't involve Y, and may not even use the same tool(s) as X. I would prefer X over those options if possible, because it requires less work, simplifies the code base, etc.

If I mention Y or Z, people tend to post answers suggesting other ways of accomplishing them that go off on wild tangents ("Use this other tool instead, it's better!") and/or re-hash things that I already know ("If you completely change your software's architecture, then X is no longer an issue."). I feel obligated to accept answers if they do answer the question, even if the information provided isn't what I wanted.

On the other hand, I'm wary of making questions too specific, lest I make it impossible for people to figure out what I'm trying to do.

Just how much context about the relationships between X, Y, and Z should be included in the question? Is it better to be really specific, and not include any context at all, or should I try to describe the complete background? Or somewhere in between?

share|improve this question
2  
Do you have a specific example? –  Some Helpful Commenter Apr 17 '12 at 15:43

1 Answer 1

Your question needs to:

  1. Be written clearly,
  2. Have enough information to be answerable, and
  3. Be of sufficiently narrow scope that it doesn't require a treatise to answer it.

If there are certain solutions you don't want, state that in your question.

Good questions attract experts. If you want to see examples of good questions, look at some of the questions answered by Jon Skeet or Eric Lippert (or any high-rep user, really), where the answers have two or more upvotes.

Accepting answers is optional, but certainly a polite thing to do if someone really did answer your question satisfactorily. To a certain extent, your accept rate is a measure of your question quality, since good questions have answers.

share|improve this answer
1  
Though I take your point, it is perhaps worth pointing out that I also seek out "bad" questions. First, because by encouraging badly-written but valid questions to be clarified, both the person asking the question and the person answering it benefit. And second, because badly-written questions are often correlated to the people who have the most misconceptions about the tool that need correcting. I learn what people are confused about by reading "bad" questions. –  Eric Lippert Apr 18 '12 at 15:20
    
I suspect that the kind of "bad" questions you refer to are actually written clearly and suitably scoped, but contain some sort of faulty premise. I have no quarrel with those. –  Robert Harvey Apr 18 '12 at 15:23
    
Indeed, there are plenty of those. But there are also genuinely badly-written, confusing, vague questions that are still useful to me. –  Eric Lippert Apr 18 '12 at 15:26

You must log in to answer this question.