As a forward, this is not about whether or not code golf is valid on Stack Overflow. If you argue against this, you're arguing against a mound of evidence, and you should be arguing in the other questions about this topic.
Most people (myself included) seem to be of the opinion that code golf questions, in and of themselves, are acceptable. Most people (again, myself included) also seem to be of the opinion that there is a certain frequency at which there are too many code golf questions being asked too quickly, or of "too low quality." Also brought up is whether or not low-reputation users should ask code golf questions, as they are not as well established in the community (which is the entire point of code golf).
So, in light of this, I suggest we discuss (and hopefully establish) a rule or guideline concerning the asking of code golf questions. Post answers suggesting metrics you think should be used to govern code golf, and upvote answers that you think are best. That way, we'll get a community consensus on what is "too much" or "too low quality" code golf.
Points to address (reiterated to put them all in one easy-to-grok place):
- How far apart, time wise, should code golf questions be? What frequency of code golf questions per your favorite period of time do you think is an acceptable maximum?
- How should the quality of a code golf question be evaluated? Should quality be an important factor in determining whether or not a code golf question is allowed? (Note: I know this can easily be subjective, but all the rules about what is "appropriate" for code golf are subjective. Just because it's subjective doesn't mean we can't get good guidelines or ideas on how to answer it.)
- Should the asker be taken into account when evaluating a code golf question for appropriateness? What guidelines should be used for that?
- What actions, if any, should (or can) be taken to inform new users about code golf rules?
EDIT: Add this:
- What are the minimum specification requirements for a code golf question? I don't think this is subjective, because it's impossible to answer a code golf question without knowing certain things, so this one shouldn't be easy. However, it is a separate issue from the quality discussed in point 2. That point is intended to deal with the quality of the challenge, not the quality of the challenge's specifications. I think should separate "well-asked code golf" from "good code golf" and evaluate (and treat) each separately. If you disagree, of course, say so.