With regard to “what’s a low number”, I just created https://wiki.openstreetmap.org/wiki/Tag:club%3Dmodel_railway after mapping one, nit being able to find a page for them, but finding a German discussion that said that would be the best option.
Currently 69 uses, but I left it at “in use”. Is 69 “good enough” to say yes it is in use? 
Greater than 0, so yes.
But taginfo data and infobox status is the same; in fact taginfo fetches its data from the infobox of a wiki page. So taginfo will and can only show the status set on the wiki page.
On status:
I see it largely the same way like Mateusz.
I think that there might be some confusion here? If I look at taginfo for a tag, one of the tabs is for a wiki page (if one exists), but the others are nothing to do with it.
You got me excited for a moment. 
I think the proper term for this classification would be unclassified.
As if that word hasn’t caused enough trouble in OSM already 
proposed - if one likes to use and suggest that tag, not only if there is a proposal process, but if usage is very low (less than 25), or only from one or a hand full of mappers
Proposed always was defined as the status according to the proposal process after “draft”, you submitted an rfc and the status became proposed. Any tag page documentation without following the rfc step should not be “proposed”, rather “draft”.
True, I saw it from the wiki perspective and refered to the tag status and other information (description, image, applicability, required tag, etc.) that is only present if the wiki page provides it for taginfo to fetch.
The main information of taginfo, the statistical data on usage, spread, combination, etc. is independent from the wiki.
The actual usage of a tag is an important thing, for sure, when you’re looking for a way to tag a feature. But in cases with low usage you may not get a reliable or satisfying answer and a tag status may help.
Errr, but “draft” has been deprecated lately (among with others, like “rejected” and “abandoned”) in https://wiki.openstreetmap.org/wiki/Tag_status#Former_status_values, no? Talk page gives some rationale (the gist being “These are statuses of proposals, not tags. Proposal statuses are different”)
I’d hazard to say many do. Try to propose removing status field from infobox at the wiki, and I’d guess you’d get a lots of flame. Why they care, is another matter altogether, and quite likely multifaceted.
At best, all it tells you is if some aspect of the tagging has been through the proposal process.
I think it conveys more than just that.
Yes, being “proposed” through proposal process means that it goes at least some peer review (the infobox would then link to proposal page for more information too). Those in turn can point to known unsolved issues with the tag (regardless if proposal passed or not) and maybe intended solutions or possible alternatives for those. ATYL on the other hand might be more misguided if they went even without that minimal scrutiny.
Some statuses like “de facto” convey meaning that that tag is in so wide use, that you should likely use it too (it is also much more telling then just “taginfo count > 100k” or similar alone - some tags have less use due to their occurrences in the world, but are still de facto usage in OSM as everybody uses them for that purpose).
So does “in use” signalling that while that tag has been used, it is not widely popular, but requires more careful reading and thinking before applying (but still better to use than inventing another ATYL alternative with the same meaning!)
“imported” would signal to take care when looking just at row usage numbers, as it is likely to be quite high, even if being used only by one particular project somewhere, so there might be better alternatives
“discardable” and “deprecated” might indicate that (even if the tag may be used many times in current database!), that it should not be used for new/updated objects (and if editing the object, if “discardable”, that it is preferred to be removed completely)
“obsolete” would indicate while it no longer exists in the database, that it was some time in the past (useful when tracking back / looking at the planet history database and seeing usage of some tag and disappearance - possibly replaced by other tag?)
etc.
What is more important is the taginfo data
Absolutely, the number of uses on taginfo (as well as spatial and temporal distribution and other info there) is very important information, but it is not be-all-end-all.
and how honest the page author has been in indicating if there are more commonly used tags for the same concept.
Well, attempts at unhonest things are in my experience most often detected and edited on the wiki, correcting them with additional data and disclaimers.
How about something like routing:entrance, which was coined, documented, and almost exclusively tagged by a single mapper, but tagged enough to outnumber many other documented features? I just changed its status from proposed to in use since it had no actual proposal, but I wonder if at least informal discussion or scrutiny somewhere would be necessary before we could accurately call it “in use”. Obviously it’s in use in a literal sense, but it seems like mappers are interpreting that key to mean it has enjoyed some organic uptake.
I agree, it is “in use” – that does not mean it is “useful” (to general population, or to anyone in particular).
Well, I don’t really see why would they infer such meaning? I do agree that “Organic uptake” is quite important, but the best indicator we have towards that requires users to go to taginfo and look at Chronology, Map and “Objects with this key were last edited by 34 different users.” – although I’m not quite what that phrasing actually means… Does it mean that:
What else could someone infer from a status of “in use” above a nontrivial element count in the taginfo box? That it was the result of a mechanical edit or import, perhaps – but the implication is the same, that ostensibly the key was adopted or scrutinized by multiple members of the community.
This is the case with routing:entrance. One mapper added it to many elements that others had reason to edit further. In general, this statistic isn’t as meaningful for secondary keys as it is for primary feature keys.
I read it as, if you look at all objects with that key, and then you look at who last edited that object, you’ll find 34 different usernames, which doesn’t mean it was that user who added the object. They could have modified it after someone else added it.
It would be interesting to know how many users are actively adding the tag and also which editors they are using. Then you could really see if a tag has “grassroots” support or if it’s just being added by a single user or a preset.
How about a status of “undiscussed” / “unproposed” to show that this tag was just created without going through the RFC - Proposal procedure?
Yes, that would distinguish ATYL tags from proposed ones, but I think it would be helpful only if we can establish in advance some clear and unambiguous way to determine when such unproposed status becomes in use.
Without that, I’m afraid such status would introduce much more confusion than it would solve.