Well, theoretical physics has failed (the site, not the discipline). We should make some effort to recapture whatever interest and audience it had. The easiest way to do that would be to introduce a "research-grade" tag. This idea has been raised many times before. It allows people who do not want to see pop-physics or physics 101 questions to subscribe only to this tag.
I would suggest we go forward with it.
The exact idea I would suggest is a tag for any question whose answer is not blindingly obvious to someone who has completed a graduate core curriculum. I think that this is basically the right criteria, rather than trying to say something about "current research" or the like. Remember, the point of this tag is to filter out questions that would be pure chaff to an active researcher.
The tag would be policed aggressively, but erring on the side of being over-inclusive, since there is little harm in that. It would also have to be added aggressively to appropriate looking questions.
Here is a suggested tag description
"Any question requiring beyond an undergraduate level of physics understanding to answer. Questions involving summaries of current research for the lay audience are not appropriate. Please attempt to find an answer to your question in textbooks, wikipedia and arxiv before using this tag."
I'm honestly not very happy with that description, but the main idea is just to scare people away from using tags for questions like "What if there is no Higgs boson?", "What if dark matter is (crazy idea)". These questions involve current research, so the asker might be tempted to use a "research" tag, but are almost exactly what people would want filtered out. I would be glad for other suggestions.
To demonstrate, the only questions on the front page that would obviously qualify for this on the front page at the moment are:
Note these are not very hard questions. The idea is not to collect the world's hardest physics questions, which would be useless since no one could answer them. The idea is only filter out the obvious chaff from the perspective of an active researcher, for those who would like such a filter.
This tag could also be reasonably applied to high level "engineering-type" questions such as Why are turbines built differently when a change of entropy or temperature is involved? . (I actually don't understand what that question is asking, but you get the idea). They are probably not what a lot of the theoreticalphysics crowd want, but it would be good for them to remember there is a real world. So I'm fine with that.
It's genuinely unclear to me why theoreticalphysics failed, when mathoverflow and cstheory have succeeded so well. Therefore I can't say that this tag will definitely be productive. It may simply be that no physicists want a Q&A site. However, the effort required to put such a tagging system in place is small, so it should be tried.
I am very interested in hearing other's opinions. Hopefully we can develop a consensus quickly and put something in place soon.
As dmckee points out this would be a meta-tag, which is frowned upon by the stackexchange system. But I believe this situation is exceptional as evidenced by the fact that there were two separate physics sites in the stackexchange system. If you read the comments to that post it is clear that they are not against meta-tags if they are performing a valid function. After all, if the meta-tags are performing a valid function, why shouldn't they be used? Further we actually have the meta-tag "homework", which is an extremely important and useful tag.
If the concern is that stackexchange would actively prevent the creation of such a tag, I find that unlikely, especially if there is some consensus about the issue. But that is a separate issue.
I have added this as a response to David's comment and just to clarify. Although the tag is being called "research-level" or "research-grade" that is not is a good description of its function. The function of the tag is to specify questions that might be of interest to researchers. To that end, I have suggested a very specific criterion for the use of tag
a tag for any question whose answer is not blindingly obvious to someone who has completed a graduate core curriculum.
This definition is based on the one fact unifying the community of researchers - mastery of the core of known physics. The criterion is sufficiently precise, and definitely excludes the vast majority of questions on the site. It definitely includes the two questions I have linked, even though the answer to first one is contained in textbooks and is not a question under active research, and even though the second one is a purely pedagogical question. They are both not obvious to all researchers. That is the only criterion. I agree there is some ambiguity in the criterion with the definitions of "blindingly obvious" and "graduate core", but I would simply suggest that we be err on the side of inclusiveness since there is no cost to a small amount of overtagging.
I also agree that the tag description and tag name do not suitably capture what I have suggested. I had a lot of trouble coming up with something. I would be glad for improved suggestions, but keep in mind the main function of the tag name and description would be to discourage low level questions from being tagged by the asker, simply because they thought it involved some aspect of current research.
While I'm here, let me take a step back and try to argue why we would want such a tag, and justify the criterion I suggested. I realize that in the initial post I simply jumped straight into the specifics of the suggestion without explanation. I should put it writing.
The main reason is simply pragmatic. There is a group of very smart people who at least say they would like to contribute to a stackexhange site, but who do not like the crosstalk from the lay audience. Their participation in the site would be very good. Therefore the creation of a tag tailored to the "researcher audience" might bring them into the site at very low cost, and therefore such a tag should be tried. Moreover, even for those who don't mind the jumble of questions of all levels such a tag would be very useful. So why not introduce it?
Underlying the pragmatic reason are a set of issues that were discussed heavily when theoretical physics was proposed. The physics site has multiple audiences crammed together (lay people interested in physics, undergraduate homework solvers, and researchers). Using tags to help separate these audiences is entirely consistent with the philosophy of stackexchange.
With all that said, this post has received 11 upvotes as of 5/01, which indicates at least some support. I have not seen any fatal objections raised to the idea. I'll give it another 24 hours to stew. If nothing has changed by that point I will simply change the description of the existing research-level tag and start retagging old questions as appropriate. I believe that those changes will need to be approved because of my low reputation. Hopefully the mods will respect my initiative and let the idea play out.