RE: Arch Doc: 26 September 2003 Editor's Draft (review of some t erms)

The problem of going deeper will be increased abstraction.

URI persistence is an issue of the energy budget of the system. 
In other words, it is an entropy quality and one could describe 
it in terms of entropy measures.   How does one measure the 'persistence 
of a URI' unless it is done operationally?  Do you really mean 
'the persistence of the resource identified by the URI' or 
'the persistence of the association of the name and the resource'? 

I note the request for sources on the terms 'selectors' 
and 'choices'.  I am debating with myself on that request 
because I fear driving this thread to more abstract terms 
than the ones I had hoped to see removed.  When I made the 
reference, it was to evoke Shannon's abstraction of 
the network architecture away from the semantics or meaning. 
It is possible to reproduce the 'choices' at the receiver 
without knowledge of the 'meaning' or semantics' of making 
the choice (a selection).  The only way I know to describe 
this better will be as others have suggested, a layered 
model where it is clear that each system layer builds on 
the one below it.  Now, what about the World Wide Web 
as a system distinguishes it from the network as a layer? 

Information space feels flimsy, but longer explanations 
are worse.  Should the abstract of the document paint 
an understandable if undefined sense, or should it provide 
the initial set of formal terms upon which all the rest 
of the content relies?  

How formal should this document be?

len

-----Original Message-----
From: Dan Connolly [mailto:connolly@w3.org]

 * URI persistence

This bit about "strong social expectations" and "always a matter of
policy" seems more awkward and arbitrary than it should be. It
seems like there should be a simple, compelling argument
from information theory and economics about URI persistence
and ambiguity.

Received on Monday, 29 September 2003 15:30:31 UTC