- From: Bullard, Claude L (Len) <len.bullard@intergraph.com>
- Date: Mon, 18 Oct 2004 09:10:28 -0500
- To: "'Patrick.Stickler@nokia.com'" <Patrick.Stickler@nokia.com>, noah_mendelsohn@us.ibm.com, chris@w3.org
- Cc: skw@hp.com, www-tag@w3.org
No. Every time you enter a change, it is a discrete change. There may be an infinite number of discrete entries, but they are discrete. This is the heart of the range problem: cantor sets. Entropy is a problem of addressability. What is clear is the architecture is much simpler if we accept the web and the semantic web are distinct systems. Shannon and Weaver make it clear why that is useful in the opening of their work. Stay away from 'meaningful' exchange until you want to layer a semantic system over the addressing system. At the heart of this issue is overloading URIs. It makes the web work but it requires two distinct systems to make it work. Identification is one system; semantic loading is another. len From: Patrick.Stickler@nokia.com [mailto:Patrick.Stickler@nokia.com] > The dog is continuous (potentially infinite set of > states) and a representation is discrete > so even if you can name the dog with a resource name, you can't > retrieve all possible states of the dog with it. You can > only name the dog. > > Does HTTP range map to infinities? This question seems to apply also to information resources. The dog's veterinary record, which I think we all agree constitutes an information resource, is also continuous, as it reflects the medical history of the dog, which varies over time.
Received on Monday, 18 October 2004 14:11:00 UTC