> Very little HTTP clients used by API are sophisticated and intelligent enough to behave like an internet user browsing hyperlinks.
But they easily could be and should be. Obviously if only HTML is getting transferred, that's a difficult representation for a machine-based API to reason about. But XML or JSON is easy to teach a client to reason about.
This is kinda the whole point of the Semantic Web and just about everybody missed it. So now everybody is finally actually doing Semantic Web [ish] stuff; they just call it GraphQL.
Actually no that's incredibly hard, because you have to specify somehow every possible relationship within your domain and write an engine which can use them appropriately.
But they easily could be and should be. Obviously if only HTML is getting transferred, that's a difficult representation for a machine-based API to reason about. But XML or JSON is easy to teach a client to reason about.
This is kinda the whole point of the Semantic Web and just about everybody missed it. So now everybody is finally actually doing Semantic Web [ish] stuff; they just call it GraphQL.