Dear Matt,
There are some tricky issues dealing with granularity of the identifier for digital data (does the identifier point at a tuple in an entity, or at a whole entity, or at multiple entities). In addition you still have the very thorny issue of what is data and what is metadata. I'll write another note regarding this issue.
the Object Model formulated by the Object Data Management Group (ODMG; /www.odmg.org) which is the "de facto" standard for object oriented design has formalized this issue with the separation between literals (entities carrying no identifier) and first class objects (entities carrying identity). A first class object is composed of other first class objects or literals, while literals themselves are atomic. How this granularity should be used for a particular application is completely application-dependent, eg. if a certain application requires that the instances of a given object (that were considered as literals before) should be individually identified, then they must be regarded as first class objects on their own. It is worth to delve into this formal standard, before making attempts to reinvent the wheel.
Kind regards,
Peter
Further reading:
Catell R. & Barry D. (1997). The Object Database Stanard: ODMG 2.0. Morgan Kaufmann Publishers