Asking good self-answered questions.
Since I've earned lots of reputation from self-answered questions since the earliest days of Stack Overflow, I figured this is as good a place as any to share my hard-won insights into the process. Asking good questions is hard and asking good questions you already know the answer to is exponentially harder. In terms of reputation and learning a subject, it's also exponentially more rewarding.
Self-answered questions come in two (sometimes overlapping) forms:
- You have a question and later you discover the answer, and
- You have an answer and want to post it to a question that doesn't exist.
The overlap occurs for me when I have a question, do some research in order to ask it intelligently, and discover a satisfactory answer before the question is completely written. If you have a passing knowledge of the topic at hand, it's difficult to ask a question without having some idea about how it might be answered.
The closer you get to form #2 (having an answer, but not a clear question), the harder it is to ask. Basically, a good question must be answerable in a variety of ways and there's a huge temptation to funnel the question into one or another answer. The results can be ugly. In essence, you have to suspend your confidence in order to ask what appears to be a genuine question. It requires a bit of self-deception to ask as if you really didn't know. (This is not, by the way, dishonest. Rather it's a vital exercise in self-evaluation. I believe this is why self-answered questions can be so enlightening.)
In no particular order, here are some tips that might help in a quest to ask questions of oneself:
Pretending that you are a writer for Jeopardy works well for me. Yes, there might only be one right answer, but you need to preserve some mystery and suspense for the audience. Try not to make your argument in the question itself. (It's a bad sign if your question is longer than your answer.)
Sometimes it's helpful to list two (or more) positive alternatives. (E.g., "I've been thinking about buying a kayak: Should I buy a plastic, fiberglass or inflatable?") This keeps the question from being too broad and shows that you've put in at least some research (though your answer will show much more). It also keeps some options open for other people to provide answers even after you've picked one choice and defended it.
Quoting others can really help. If you are able to find somewhere in the literature a question that your answer fits with, you can just quote the question. (Often I find these sorts of questions in surveys of the relevant academic literature, so they often include a good list of options as well.)
Before you submit your question, take a moment to re-read it (a good idea in any case) to see if you could imagine writing more than one answer to it if forced. (For instance, if you were on a debate competition and the positions were assigned by coin flip.) Consider moving some parts of the question into a potential answer if it seems like you've tilted the contest too far in one direction or another.
Should we encourage self-answered questions at this stage in the site?
To get at the question you actually asked, I think we should encourage these types of questions as long as they are good fits for the site. The fact is we are in great need of good questions and we really can't be picky about why they were asked.
We don't need (now or ever) bad questions, however. Since self-answered questions are difficult to ask well, we need to be vigilant to be sure they fit the mission of our site. We should not be hesitant to downvote, vote-to-close, edit or otherwise correct problem questions.
The only other caveat I have is that all questions should encourage answers. Sometimes a question that is answered by the asker has the appearance of not wanting (or needing) additional answers. That's not quite as productive as questions that are left unanswered for a few hours or days. For that reason, I have been trying to limit my answers to my own questions until at least a few other people have had a chance to read them. (If that means that a question with an obvious answer is left unanswered for a while, so be it.)