I've noticed that when it comes to straight-up programming concepts, questions are generally either tagged pretty well in the first place or retagged fairly quickly by someone with the privilege. We've got dedicated tags for repository, dependency-injection, exception-handling, and so on.
I've also noticed that in other areas - most notably databases - the tags are almost always generic, and that's something I'd like to help rectify. I got the idea from another SO user who'd retagged some 30-odd database questions with greatest-n-per-group, which isn't the most concise wording, but I thought it was a good thing because it was true, the questions were all very similar and referred to the exact same concept. Every so often I see some of the same patterns and think about doing the same retagging.
But at the same time, I don't want to step on any toes, as it appears that part of the community frowns upon the practice of "mass retagging". Although that post was about redundant, "stupid" tags, it will still bump a bunch of old posts, and I think in order for the tag to "stick" it has to have at least 10 entries, maybe more.
So - what is the recommended practice for when you see a good candidate for a new tag? Should I:
Just start retagging the old questions, and let the volume of new questions on SO bump the older questions back down if they aren't interesting?
Retag only new similar questions, without touching the old ones?
Retag a small handful (5-10?) of the old questions, just to get the tag started, and then stick to new questions?
Just leave the whole issue alone and contend with the generic tags? Stick with the philosophy the tags should only represent the platform and not the pattern/concept?
Just to be clear: I really don't care about a badge, for all I know some of the tags already exist and just aren't being used properly.
