One of the most important questions to be settled is: How are we supposed to tag questions on this site? I happen to have some opinions on this matter, and I will use this meta post to lay out my 'vision'. Please provide alternatives by means of an answer to this question, or express your opinion by up- or downvoting this question. Most importantly, please suggest ways to improve this post! This could be done by improving my classification scheme, pointing out mistakes or anything else, really. Just leave a comment, or talk to me in chat.
Important things to keep in mind:
Tagging is important. It is crucial in keeping the site organized, allowing one to efficiently search the site, and helps those who feel the need to filter out content that they are not interested in.
The tags should be of use, and not just for this particular question. This means that a very specific tag (e.g. a specific date, name of a book) that is not likely to be used again (note that I am, here, talking about used several times again) is a bad idea. A 'pitfall' that I have seen several times already (although there are >20 questions on the site at the moment of writing) is tagging a question with the name of a scientist. I think this is generally a terrible idea, since one can hardly expect more than, say, 5 questions about a single person to come up. This means that this tag is not broad enough to be useful. Note that there may be a few exceptions to this general rule (Newton?, Einstein?), but this is something that should be discussed further at a later time, when we all have some more experience on the site.
The tags should be sufficiently descriptive that someone seriously involved in the general discipline which it applies to (e.g. mathematics, biology) will be able to tell what this question is about, and whether there is a reasonable chance that he/she will be able to understand the question (or answers) completely. This implies that just a single tag "mathematics" will not suffice: Tags need to specify different fields of research in order to be accurate enough to be useful.
Keeping all of this in mind, I propose the following rough classification scheme for good tags:
Discipline (e.g. mathematics, chemistry, computer-science, experimental-physics). These kind of speak for themselves. I think these are 100% necessary, so people can filter out questions on fields that don't interest them, if not for a 100 other reasons.
Subject matter (e.g. number-theory, quantum-mechanics, photosynthesis, machine-learning, organic-chemistry). The distinction between this category and the one above it is somewhat artificial and arbitrary, but certainly not harmful (in my opinion). Often, a clear-cut case can be made whether a tag should be a 'discipline-type tag' or a 'subject matter-type tag', but it is never essential that this should be possible, so there is no need to worry about possible inconsistencies.
Historical (and/or cultural) context (e.g. ancient-greece, renaissance, industrial-revolution). I just think it's a good idea to be able to separate questions based this criterion, both for searching and filtering purposes. Note, however, that the tag should be meaningful in the context of science. That is, a tag for a time and place should only exist if that time and place has characteristics that are (i) meaningful to the history of science and math, (ii) don't extend beyond that time and place, and (iii) apply to the entire time and place.
Type of question (e.g. reference-request, biographical-details, academia, examples). These will mostly be necessary to classify questions that are not of the standard format: "What was the historical development of idea X?" (very broadly speaking).
Note that this will, of course, not cover everything but hopefully it'll be sufficient to help along with the majority of the questions