RE: Use of "assign" for URI -> resource

> Suppose I mint the URI http://www.example.org/ij# and
> establish that the resource it identifies is Dan's
> car. 

But you can't really do that. You, as the owner (presumably)
of http://www.example.org/ij have the authority over the
'ij' resource, and can (if you so wish) establish an RDF
document there that describes enough of Dan's car for you
to believe (yourself) that http://www.example.org/ij# identifies
Dan's car.

If you hand that URI to someone else, the only way they
can know that you really mean "Dan's car" when you
say http://www.example.org/ij# is to fetch a representation
of http://www.example.org/ij, see that it is RDF, understand
what the fragment identifier means, and then guess
about what is inferred.

For URIs to be useful for communication, the receiver
has to know what the terms mean. The value of "URI" over
any other kind of "RI" (resource identifier) is the "U":
no out-of-band communication or understanding is needed
for the receiver of the URI to know what the sender of
the URI meant.

Otherwise, you get into a humpty dumpty world, where
terms (URIs) mean whatever the sender wants them to
mean, and the reciever just has to know.


> Dan is clearly the owner of Dan's car. It was
> confusing to state that I was somehow the owner of the 
> resource identified by the URI.

But (presumably) you are. Otherwise, I could say
  http://www.w3.org/Larry# 

really means "Green Cheese".

> Because of that unclear relationship, we moved from "resource owner" 
> to "URI owner" prior to the (initial) Last Call draft.
> URIs can be allocated and thus "owned" (or rented). 
> It seems more difficult to explain how resources can be 
> allocated.

Resources can be allocated and rented, and URIs mean
whatever they mean. To make a URI mean something other
than what it meant before, you actually have to arrange
the resources.

IMHO

Larry
-- 
http://larry.masinter.net

Received on Thursday, 16 September 2004 05:01:02 UTC