Qmechanic just raised on chat the need to take some more action on the mathematical-physics tag, and I think it's about right: the tag, despite some loud all-caps in the tag wiki excerpt, gets misused all the time, by people who simply have no idea that mathematical physics is a discipline unto its own and that the term should not be used to describe posts just because they have a bit of highschool algebra in them.
For the more data-minded, this query finds all the posts which were originally posted with the tag, at a current count of 1175 over the site's history and there's plenty of clonkers in there.
Now, Qmechanic proposed some manipulations with the tags that I rather strongly disagree with, but I agree that there's a need and I think there's a better solution to the problem.
Let's have a review queue for this.
Here's how I think this could work:
If a new user posts a question with the tagged mathematical-physics (or some pre-defined set of tags, such as e.g. mathematics), then it gets sent to the First Posts review queue.
There the review, instead of being described with the usual tagline
This is the first question asked by a new user. Help them learn to use the site by reviewing their post.
it gets the header
This question by a new user has the frequently-misused tag mathematical physics. Please check that the tag has been used correctly and remove it if it hasn't.
or some similar wording.
After enough time (say, the user has three positive-score posts on the tag, or 1500 rep, or some linear combination of the two), users' posts no longer get sent to this queue.
We could additionally impose a similar experience requirement on reviewers, so that users with 500 rep will only be shown these reviews if they've got three positive-score posts with the tag, knocking off one of those per extra 500 rep.
(All numerical thresholds obviously subject to adjustment, particularly if we find data sources about the existing misuses that are more than me pulling numbers out of thin air.)
Now, once you divide those 1200 posts over the six years of the site's existence you get something like one post every few days, which is a bit sparse but non-negligible, but mostly, Qmechanic does the bulk of the re-tagging around here, and if s/he says it's getting out of hand, then I believe it, period. I think this is enough to show that the problem is real.
It does involve some nontrivial coding and changes to the UI, but I don't think they are too radical, and they can essentially ride piggyback on the existing queues. And, if the queue system does get an overhaul at some point, this could be brought out into its own branch if so required.
Moreover, I think this kind of mechanism would be useful in the wider SE arena: there are tag warnings for misused tags (which is something we should also implement) and those have been seen with interest by other sites (example, example), so presumably the sites that implement tag warnings might also be interested in a review queue along these lines.
I'm asking it here first, though, because it's probably easier to test the mechanism in a site where it's less likely to break things before scaling it up to full-SE-network availability.
Also, for clarity, there's several other tags that probably need this kind of additional scrutiny, including mathematics, numerics, laws-of-physics, computational-physics, and equations-of-motion.
For the moment, I've set up a dedicated chatroom: Tag review for misused tags, with feeds from those tags, as an ersatz review queue, and as an experiment to see how well that works. (If you do edit tags from feeds in that room, leave a message saying that you've done so, to help track how useful it is.) Hopefully this will help get a sense of the scale of the problem and its available solutions.