> If the client knows nothing about the meaning of the responses, it cannot do anything with them but show them to a human for interpretation. This is the best

> If the client knows nothing about the meaning of the responses, it cannot do anything with them but show them to a human for interpretation. This is the best counterpoint in this discussion and it deserves a lot of reflection. But that reflection should include the realization that this is what the browser does all the time. Browsers don't have any particular semantic model in their head of what any particular form action or hyperlink "means" in terms of domain logic... but they do have some semantics for broad classes of verbs and other interactions associated with markup-represented potential actions, and so they serve as an agent presenting these to the user who does whatever wetware does in processing domain semantics and logic and makes a choice. This has worked so well over the last 30 years it's been the biggest platform ever. We're now in territory where we're creating software that's almost alarmingly good at doing some semantic processing out of no particular structure. The idea that we're facing a dead end in terms of potential for clients to automatically get anything out of a hypertext API seems pretty tenuous to me.