A preliminary outline of the place of XML in a world wide web of data is documented in the Cambridge Communiqué (Swick & Thompson 1999). We observe that XML initially shared a design philosophy with SGML, focussed on a simple data model for a tree-structured view of documents and a flexible linearisation or transfer syntax for documents represented using that model. The majority of energy going in to XML has, however, been sustained by the empirical observation that tree-structured documents are a pretty good transfer syntax for just about anything, and that by converting from application data models to the document data model, the utility and ubiquity of XML become available for moving application data around and making it available outside its original producer/consumer community. The Cambridge Communiqué says it this way:
It goes on to recommend that XML Schema make provision for facilitating this layered approach.
The layered approach can be seen as complementing a more constrained approach to data interchange based on abstract APIs and Remote Procedural Call ( RPC), e.g. COM, CORBA, IDL.