I have seen several cases on several different sites on the network where I have been tempted to add a tag to a question before realizing that the subject of the tag appears only in the answer, often an accepted or highly-upvoted one. I've seen two major types of questions where this tends to happen:
- "Identify this" questions - Once an answer is accepted and/or a consensus is reached (depending on site standards), it could make sense to add a tag associated with the answer. For example, SFF.SE permits "Identify this book" questions, and many of the books people ask for actually already have tags.
- Problem-solving or debugging questions where the type of problem is not evident from the problem description. For example, a user might post a debugging question on Stack Overflow where it turns out that the problem was due to a race condition, but that fact only becomes evident in the answer (i.e. the problem statement doesn't have any classic telltale signs that a race condition is likely behind the problem). Would it make sense to tag the question with [race-condition] once that does become evident?