This question of mine was made community wiki by someone other than me. I'm aware of discussion (e.g. here and here) arguing that big-list questions should be CW so that the voting system can be used to rank the answers without affecting the reputation of the posters. I suppose someone might have believed my question was a big-list, but I would argue that it is not: I am not looking for a list of all mistaken uses of notation, but a single example of a notation that turned out to be incorrect and led to mathematical mistakes, and if I get such an example I will accept it. The question is certainly soft, but I haven't seen anyone arguing that all soft questions should be CW (and offhand I would probably disagree with that). Is there another convention that applies to make this question mandatory-CW?
After consulting with the other site moderators, the decision for now was to let the CW status stand for the question at Main. But this may be a good time to revisit the discussion of some of the ways MathOverflow has been known to use CW, which differ from how many other StackExchange sites use it, and which are consistent with certain traditions since before we were part of the SE network.
Currently it's only the site moderators who can make questions CW, or who can undo that action. If a question is CW, then so automatically are the answers. On the other hand, if a question is not CW, a user who has answered can make his/her post CW independently, without affecting other posts within the thread.
Speaking as an active moderator, most times I make a question CW are after it was flagged as a possible candidate for such, although sometimes I'll do it without prior prompting if it seems to me well within the wheelhouse of questions we've traditionally made CW.
Questions that are typically or at least often made CW include
Career advice questions,
Questions regarding style (e.g., in writing papers),
Big-list questions, especially if they quantify over all mathematical fields.
A justification is that, especially in such cases, there may be strong differences of opinion, each with weight and argumentation behind it, regarding which answers are better than others. [For example, whether to punctuate at the end of a display line, or whether in a paper one should refer to an earlier Proposition using a capital P.] Here it seems sensible to put it to the community and sort it out with large numbers of votes, and the community generally feels more invited to vote up or down when reputation points are not at stake.
(By the way, when I asked, site moderators felt that Mike's question came under the umbrella of the third bulleted point.)
An answer is often made CW in order to avoid an unseemly appearance of existing for the sake of gaining reputation. For example, a question may have answered in comments, and someone culls the comments together to form an answer, but makes the answer CW in order not to take credit (in the form of reputation points) for answers by others. As a community service. This can also be done in cases where different people answered different parts of a multi-part series of questions within a single post, whereupon arises the nettlesome difficulty of which answer to officially accept. The difficulty can be neatly finessed by assembling all the partial answers into a single master answer and then marking that one CW, courteously assigning all credit where credit is due. Marking that as the official answer then seems a satisfactory solution.
Another example: if someone asks a question but already knows an answer, and posts the answer concurrently with the question, then she might make the answer CW, again to signal that it's not meant to step in front of other answers (after all, often the early bird gets the worm where points are concerned), but more along the lines of selflessly adding to the public commons.
None of this is actually written down as MathOverflow official policy. That in itself might make some people uncomfortable; my hope is that this answer might help lessen the discomfort a little. My understanding as expressed above is based mostly on informal traditions and things that have frequently been said over the years, much of it in comments at Main, or in flags. So some of it is a bit hard for me to quantify. One early discussion of such matters may be found at 'tea': http://mathoverflow.tqft.net/discussion/6/when-should-questions-be-community-wiki/, and here's another: http://mathoverflow.tqft.net/discussion/247/question-mysteriously-becoming-community-wiki/?Focus=4242 More recently there was this: Why was this question made community wiki?