People upvote both the question and the answer, so this question must is useful -- I quick glance confirms that the answers are useful. Does the one-line question itself add more value then an average voted-10 questions? Perhaps yes.

On the other hand, sure, I also wanted to learn about categorification but I took time and read the Baez text some time ago, long before Math Overflow existed. Most regulars here likely did something similar at some point of their lives as the resources are plentiful.

Does it make sense that a person who doesn't provide any indication that s/he actively thought about the question (there's no indication to the contrary, either, so this should not be interpreted as picking on the particular question) is now directing the discussion about it, and, yes, reaping the reputation benefit? If it does, tomorrow a nontrivial number of people will be posting a list of questions "What is XXX" using the dictionary. Heck, it's not hard to write a bot to do that and the competition in bot writing doesn't seem like a good plan for the site.

It's ironic that initial part of this thread was devoted to discussing essentially one person (me) using as the main example the sl_2(R) question -- that one actually was a continuation of an old conversation, both spoken and email, with David Vogan, where I wanted to learn a bit of "real" representation theory and he should get the credit for answering the "stupid" parts of my question (which were therefore not posted). And no, even though I accepted the answer with the reference to park city/ias proceedings, that answer is useless to me about 300 miles away from any half-decent public math library - writing an overview article would be a much better answer for me. One is welcome to say, of course, that having a smart person who could be doing something else instead wasting time on writing overview articles for me (or someone else, for that matter) would lower the overall society benefit -- I won't argue with this point of view, it may very well be so.

Also, some of my other questions were worse, and given that I know about my conversation problems, including "annoying people who have better things to do rather then explain me obvious things", I plead guilty as charged (sometimes better to try to work it out with the prosecution :) )

This makes me go back to the original question: what is the strategy to apply to what-is questions. The only suggestion I came up with so far is thus: [turn the craziness filtering off] one could write a bot that posts all possible what-is questions using Wikipedia + the dictionary, as well as all open problems. Then if somebody wants to post another one, it can be immediately closed as a duplicate. Meanwhile, there will be nothing preventing people who want an answer to particular what-is question from upvoting it and nothing stopping people who like it from writing overview articles, with everyone clearly understanding how the process works. [turn the craziness filtering on].

Now the above strategy might sound impractical, but perhaps the policy exists with practical results that would be indistinguishable from the results above for a typical observer. That is, one could try to act as if such a bot existed and had already posted all what-if questions and open problems.

The wikipedia article is pretty terse, and categorification seems more like a yoga than a precise thing, so I feel like it might be okay, but it feels borderline.

The reason I brought this up here on meta is that I can't quite put my finger on what it is that I find objectionable about these questions. But it looks like other people agree with me that MO is best suited for "questions that have an answer". Obviously, "what is X?" is a question that has an answer, so let me try to elaborate what I mean by this. Everybody who visits MO knows basically how to learn and do mathematics. But we all get stuck sometimes, and this is where I imagine MO saving the day. When you're stuck, you can go to MO and say "I'm trying to do X. How can I do that? Does this work? Does anybody have a reference?" The idea being that for an expert, it should take very little effort to understand your confusion and set you on the right path. Or maybe a non-expert has come across the same sticking point and can explain how she resolved it.

Of course, I do think there is a place for big picture questions on MO, but even these philosophical questions should "have answers." An excellent example of a big picture question (and answer) is Limit Linear Series. Notice that even though this is a big picture philosophical question, it is definitely very to the point. In this case, an expert was able to see the question, understand very quickly exactly what problem the asker was having, and provide a succinct to-the-point answer.

Even opinion questions have some place at MO. Though the MO framework is terrible for discussions (by design!), the voting mechanism makes it useful for making sorted lists of useful resources. Obviously, these kinds of questions should be community wiki.

Obviously, I don't actually have any more experience with MO than anybody else, but I did spend several months thinking about it. I'm glad that there are smart people here on meta to discuss exactly what niche(s) MO should fill. Here are some of my opinions about what MO is not (subject to change/addendum, of course). I'm happy to hear other people's thoughts.

• "MO is not the place for help with your homework." Hopefully, this one is crystal clear to everybody by now. There are lots of other sites where students can ask for math homework help. I don't think MO should take over that niche.
• "MO is not a discussion forum." Though there's place for opinions, MO is absolutely the wrong framework for having a discussion. Discussions are more appropriate on blogs or threaded forums. I don't think MO should take over that niche.
• "MO is not an encyclopedia." As I discussed above, MO should be a place for "questions with answers." My understanding is that nLab is a good place to ask for expository articles about a topic. I don't think MO should take over that niche.

@Scott: I really like the "MO is not an encyclopedia" slogan.

Am I over-reacting? Do other people want to see these kinds of questions? Certainly there are some examples where the question is appropriate (e.g. What is the field with one element?) which don't fall into the offensive category I'm trying to describe.