I am not sure to which extent I can answer. I'd guess I am probably just going to say a few self-evident things.
First of all, we should not retag too many questions at once - but this comes under "not too much bumping" and every experienced user knows this.
Some question will come organically - people will notice a new tag, somebody might retag an old question here and there, somebody might add new question with this tag.
I do think it would be over-ambitious to find and retag all old question related to this tag. But perhaps the most important ones should get this tag.
So what would I do is that if I can think of a result in this topic which is quite important and I can find questions about this, I'll add tag there. (Or maybe just mark the question to add the tag later if I have already bump too muched question.)
And I would also search for the relevant keywords and then sort the results by score - this might also help to find some relevant question.
And from there questions will be added just by "random browsing". At least the users who were involved in the creation of the new tag are aware of it. So it's safe to assume that they might occasionally notice a questions from that topic. (For example, if you or me are just randomly browsing the site and we stumble upon question related to semicontinuity, we might add the tag.)
@MichaelGreinecker This was discussed some time ago, but the site has probably grown too much since then to be still considered relevant.
I have to admit I am not always too careful about this. But some stuff I try to follow:
- I do not bump too many questions at once (certainly not more than 10, I try to stay under 5 at the same time; probably it's better to do more smaller batches).
- I have a few lists on my computer of questions where I should get back and edit something - if I have already bumped some questions I simply add a question to that list instead of editing.
@MartinSleziak Okay, I think we can go on and at least create the semicontinuous-functions tag. The case of multifunctions needs probably more effort in identifying the relevant questions.
- At least sometimes I also check whether there were many old posts bumped recently by checking something like this: math.stackexchange.com/… (If I see only a few bumped questions in the last 30 minutes, I usually consider this as the green light.)
I have mentioned 30 minutes exactly for the reason that it's about the size of the first page at the present rate (if you are displaying 50 questions per page).
- Another thing to do is to look whether some other improvements can be done to the question you bumped (or answers posted there). If you bump questions in this way, sometimes one question will keep you occupied for a few minutes.
- And I should add that I do not count questions which has already been bumped for some reason. (If a question was bumped anyway during the last day or two and I see some further improvements to the tagging, title, formatting, ... then this is certainly a good time to edit - this is certainly better than two separate bumps long time apart.)
