It seems to me that the SGML content model system is primarily
concerned with describing the future of a document (what can
go where). The XML-data schema system seems more
concerned with the 'present' of a document.
So SGML is concerned with information management, while
XML-data is concerned with information retrieval. Does that
seem right?
AFs seem to sit in between these two extremes, being
able to constrain futures and give more sophisticated
labelling for data retrieval.
-ricko