The “permission or reachability” definition dates back to the original access tagging proposal and has been with us ever since. It was only later that mappers insisted on purifying the keys to only pertain to legal access restrictions, but in practice, everyone draws the line somewhere.
The reason this issue keeps coming up is that, in many cases, to indicate permission is to answer a question that nobody asked, yet a purist interpretation of access keys leaves the question that matters – reachability – either unspoken or relegated to an obscure key that routers and renderers don’t recognize.
Most of the time we tag access restrictions, we’re doing it to influence data consumers to inform users about where they’re supposed to go. Few if any routers are intended to send users on the most extreme, barely legal route. Rather, the job of a router is to answer the question, “What’s the best way to go?”, for some definition of “best”. It’s our job to make sure that “best” route is neither illegal nor a legal fiction.
It would be (and has been) quite difficult to define a bright line rule that no edge case would confound. For every case where an access key has been “misused” to make a subjective judgment, there are countless cases where no means impassable based on common sense. Mappers didn’t need to come up with a whole new scheme or scale just to record that fact.
I’d favor establishing subkeys to clarify access restrictions where necessary. However, there should be subkeys pertaining to both permission and reachability, with the main access keys being a summary of these aspects when no distinction is necessary or relevant.