This question has been asked elsewhere on Stack Exchange (for instance in Stack Overflow meta ).
Basically, there is a type of question that exists, that follows the format "I need x, it should be y and/or z". It's not so much a question as it is a work request. I call these 'homework questions' because on most sites, it's often the case that they've been copy/pasted from a homework assignment.
The problem with homework questions, and the answers we have been providing to such questions, is that they ask for and receive specific solutions rather than general ones. It's like you ask "what is an even number?" and top-voted answers are "2" and "4". The 'right' answer should be "2x where x is any integer", because it allows the OP to better understand the process behind what makes a number even and how he can find one on his own.
I believe we should focus more on providing general solutions to problems, instead of being a place where people can come to have others build their worlds for them. The problem is, there is no close reason for well-constrained questions asking for a specific world to be built, and answers that provide such worlds can't really be blamed for doing so instead of providing a more general solution.
For example, see this question that made me start wondering about this:
This question is asking for a world with specific criteria. The asker suggested that they followed all the rules of the site, and there was no good reason for their question to be closed. Plus, it got 21 answers, each providing a different option. I don't really think the question was too story-based.
Also, this question I think gives another good example:
Kingdom with helpful bands of mercenaries; avoiding devolving into bandits
I believed the answer to be 'give them other jobs to do', but all the answers I saw last time I checked was providing different ideas for jobs that could be given. Again, there are too many answers, when a single answer should be able to provide a framework that would lead to all the others.
Quoting from the Stack Overflow answer linked above:
I wholeheartedly agree. The existing close reasons are not always sufficient.
Consider the following question:
Write a method in Java that takes an int[] arr and an int k, and returns the kth largest element of arr. The method should run in O(n) time, where n is the length of arr.
This cannot be closed as too broad: it's quite specific. It cannot be closed as unclear what you're asking: it's perfectly clear. It cannot be closed with Questions requiring debugging help... because it doesn't ask for any debugging help.