RE: argument from authority considered pointless

From: Tim Bray []

Bullard, Claude L (Len) wrote:

>> In short, it appears that URIs *as formally defined* 
>> are not appropriate for RDF *as formally specified*. 

>This may be a fair statement of where the angst is coming from.  Let's 
>assume, as a thought experiment, that it is.  So, do we fix the 
>definition of URIs or do we fix the specification of RDF?

I get the advantage of posting, going home, then coming back the 
next day to see the responses.  This one that you made to David 
Booth is striking to me:

"It is quite possible that the Web Architecture works 
*because* it works around the intractable problems of meaning and only 
deals with comparing identifiers and shuffling representations around; 
avoiding a lot of problems that historically have been intractable."

Very Shannonesque.  Only define the systems required for a selector 
given equally probable choices.  Leave "meaning" out of it.  
As I read the responses and the RFC, that is the role of the authority.  
Also Shannonesque.

So, the web works because given a URI, the authority over the 
resource selects/identifies the representation. 

In effect, saying the URI identifies the resource doesn't buy one 
much but the outer braces for the set, so to speak.  Quite useful 
because it definitionally allows for a set of representations and 
that is semiotically, a sound approach.

It doesn't work for RDF because, as I understand it, RDF requires 
the URI to unambiguously map to the object and doesn't have the 
concept of the authoritative selector.  It presumes, in contrast 
to Shannonesque systems, there exists widely shared selections, 
an authoritative semantic, a pre-existing map.  It seems the 
URI should identify the owner of the map.  [SIDEBAR:  The Golem 
problem still exists in such KR systems and always will.  Choosing 
who chooses choices is a problem but cannot be solved architecturally. 
It is political so I won't debate it here, but given the DARPA 
systems such as TIA, it can't be ignored either.]

I agree with you.  The problem here is that RDF attempts to 
use the URI for something which it formally doesn't do.  The 
URI names the resource, an abstract concept which includes 
an authoritative selector for the set of representations 
which as Roy says, may vary in time.   I think as I said 
in the response to Roy and Sandro, RDF needs the extra 
concept of a selector/mapping/identifier.

Oddly, RDF always argues from authority.


Received on Thursday, 23 January 2003 11:41:49 UTC