First of all, sincere thanks for @supsup for bearing with me! Sorry for all the drama. I seem to suck at explaining my stance.
To me, mountainbiking firstly seems just a much too specific mode of transport (or sport) to warrant a top-level =mtb tag to replace =path in use. Secondly, it seems far too vague to work, because people can an do MTB on a very wide variety of paths.
I mean, just take a look at the mtb:scale wiki page (the existence of which is to me an independent argument against a top-level =mtb tag). Even if we exclude the 0 category, there are still six types of paths that people actually MTB on. Those categories are based on (hint, hint) complex physical criteria and cover quite a big range. My real worry is that because of this, a huge chunk of =paths would just be turned to =mtbs. Then most of the point of having such a new sub-category would be lost.
Again, we don’t have highway=tractor, =suv, =atv, … , but rather highway=track that implies that you might need an SUV, ATV, tractor, etc. to traverse it. Depending on the physically delineated secondary tractype=-tag a normal car might really get stuck in the mud.
I get the idea, and kinda support it (though, again, picking out MTBing sounds to me awfully specific). A sincere further question on this: if the new =mtb tag would not be based on an mtb:scale tag (as you indicate above) and would not be based on the fact that such paths are used mostly or indeed exclusively by MTBs (as you intimated earlier), what would the criteria be for applying it? Doesn’t the fact that someone actually assigned an mtb:scale-value to a path tell you that it is being fairly regularly used by MTBs?