You are right in what you say, as far as it goes, but I think you are overlooking the other fundamental piece of "the web" -- the existence (or at least the pot

You are right in what you say, as far as it goes, but I think you are overlooking the other fundamental piece of "the web" -- the existence (or at least the potential existence) of multiple, competing client implementations. This is as important as all that other stuff. If only one proprietary implementation/app can access a set of links, it doesn't matter that they are standard HTTP URIs that serve standard JSON or whatever. They might as well be links to CompuServe or eWorld resources, at that point. That isn't always the case, but it is sometimes the case (and is sometimes enforced with some kind of token or other mechanism to prevent access by "unauthorized" client apps). I would argue that those cases are fundamentally different from "the web".