David, an amazing parallel of thoughts. Just read your posting, and fifteen minutes before (on my way to the office) I had the following thoughts.
(a) It is questionable to pretend that any model involving people and telephones is a faithful model of reality. From the point of view of physics and chemistry, telephones are just a chunk of material space, not more important than the lower left corner of your office. Modelling is inherently done with purposes in mind, a perspective and an intuition of usefulness.
(b) It is the usefulness what is crucial, and I reject the "political correctness" of rejecting hierarchic relationships on the grounds of the possibility to define them otherwise. A conceptual model must satisfy intellectual needs, and should be judged on the grounds of how well it does that. Tree structure responds to fundamental intellectual needs.
(c) It is crucial to be aware of the possibility to transform "the model" into other structures at any time - this awareness frees us to select an intuitive structure to be "the model". Think of Margritte's famous painting of the painter painting a bird, seeing an egg.

---

Von: David Lee <dlee@c...>
An: Hans-Juergen Rennau <hrennau@y...>; William Velasquez <wvelasquez@v...>; "xml-dev@l..." <xml-dev@l...>
Gesendet: 1:33 Dienstag, 1.Oktober 2013
Betreff: RE: Re: XML Schema as a data modeling tool

Thanks everyone for this thread, its very interesting.

What is a projection of what, and What is modeling what is fascinating - not only in the markup and computer modeling world,

but in physics as well.   Richard Feynman had a great chapter in one of his books about the physical universe seemingly designed specifically to maximize the number of equally valid but distinct models - So when I hear people claim we are not modeling "the real world correctly" it does make me chuckle a bit inside ... not in pejorative way, just that the whole concept of modeling "the real world" is a funny thing as it is.  

 

But back to computers ....

 

First trees vs Graphs.   I can see arguments both ways ... and other ways.  Trees and Graphs arnt the only models. No model is "right" ... but is it "useful" ? Thats a useful question (self-referential reference intended).

 

Projections of a core data model is intersting.   For example I am currently trying to imagine whether it makes sense to take the UBL messaging model and use the actual messages as the  data store and model business transactions directly as a revision chain of message documents.   Or does it make more sense to model the underlying business model different and "map" the messaging model at the fringes ...   Neither of these are "perfectly right" ... but there is a compelling elegance to imagine the messages as the actual model itself.  

 

With the right views, the right (or "good enough") technology, is it possible to not care which is the model and which is the projection?

The same as with physical systems where you can model the system as the set of interactions instead of the set of states ...

can you model a business process solely as the set of messages and the interactions they have with each other ?

Is a particle just a transient stable state between two different energy states ? Or are the energy states the projection of a change of state of particles ?  Which is real and which is a projection? Both ? Neither ? does it matter ?  Is a "Shipment" really the stuff in a truck ? or is it the set of messages and contracts which constrain the business ramifications of items leaving one organization's ownership and entering another's ?

 

I suggest what's "real" is not so important as whats useful.

Trees are definitely useful.

Now if only money would grow on them :)

 

 

 

 

 

 

----------------------------------------

David A. Lee

dlee@c...

http://www.xmlsh.org