I would like to know of types of bad examples of MathOverflow questions that are tantamount to reference requests. I can think of two:
a) A reference request for something that is properly outside mathematical literature. For example, biological examples of Lindenmayer systems are likely to be found in the biology section of the library, while a formal mathematical treatment might be in the computer science or mathematics sections. Karnaugh maps for logic reduction and their history are usually relegated to texts treating digital logic design, and I can't think of any pure mathematical treatments or generalizations, although my experience may just prevent me from seeing such; I would expect most questions on Karnaugh maps that are reference requests to be off topic for this forum. (On the other hand, "Are there any mathematical analogues of Karnaugh maps for n-valued logic systems, and is there such analysis in the mathematical literature?" would be a generalization I would be interested in seeing, even if the answer turned out to be "look in the biology section of the library".)
b) A reference request for a "thing" that is not concisely stated. One of my favorite questions Erik Westzynthius's cool upper bound argument: update? is a borderline example, primarily because of my desire to share the sieve argument; with a briefer summary of the argument I think it would serve as a good example. A (more) bad example would be if the "thing" were a sequence of words that is supposed to express an idea or collection of ideas, but are not well parsed by MathOverflow community members.
I have read other questions touching on this topic, especially Appropriate Reaction to a Failed Reference Request on MO of Manfred Weis, so I am aware that some might consider this a way to "self-promote". However, I am willing to tolerate some self-promotion if at the same time it is also promotion of an interesting and/or useful idea. This would encourage MathOverflow to become a repository of reference links to interesting and/or useful ideas, which I think is one of the goals of this forum.
Note that I am not classifying poor attempts as bad. A person may have trouble expressing their idea, in which case (some helpful portion of) the community should say "This is not well phrased for this community to help: see the FAQ for how to write a successful reference request". Or the person asks a "newbie" question, to which the answer often is "Do a web search; you might find XXX among the first hits". Or the person might say "Here is a link to my preprint; anything like this in the literature?" to which a proper answer should be "We consider focused questions containing brief and well-stated specific ideas here, not preprints. Ask a focused question as indicated in the FAQ".
Although I am focusing on bad ways, it is my hope to generate a list of good and bad ways of making a reference request, as well as ways which suffer from defects to which certain standard responses and fixes apply. Also, I make the assumption that the community wants MathOverflow to serve as such a repository. I ask that challenges to this assumption occur in a different question.
To spur discussion here is an example which I think might be OK, but could in fact be something to discourage. Some humorous math department anecdote is related (perhaps with the tagline "It could be infinite!"), and the request is if this is published. It is OK if the anecdote is accompanied by appropriate interesting and/or useful ideas, while it might be bad if it is just something to add to a math jokes compilation.
Gerhard "Will Tolerate Some Math Jokes" Paseman, 2014.01.08