RE: //example.org/car (was lack of consensus on httpRange-14)

> -----Original Message-----
> From: Mark Baker [mailto:distobj@acm.org]
> Sent: Saturday, October 05, 2002 8:27 PM
> To: David Orchard
> Cc: www-tag@w3.org
> Subject: Re: //example.org/car (was lack of consensus on httpRange-14)


> 
> I think that's a bad idea.  I've never met an identifier that couldn't
> benefit from being dereferenceable.  As the first TAG finding said;
> 
>   "o URI's for important resources should be dereferencable. 
>    o Dereferencing URI's for important abstract concepts (for example,
>      Internet protocol parameters) should return human and/or machine
>      readable representations that describe the nature and purpose of
>      those resources."
>    -- http://www.w3.org/2001/tag/2002/01-uriMediaType-9
> 

I'm quite sure (from previous correspondence) that a new URI scheme for
non-dereferenceable resources is exactly what Micah Dubinko had in mind.
I've had the same thought throughout this debate, but wondered if the TAG
hadn't already rejected the idea.  Thanks Mark for the reference.  I think
there's also another TAG finding to the effect that new URI schemes should
be discouraged.

But its increasingly clear to me that the ideal of not deprecating prior
findings (and W3C Recommendations, for that matter) and the ideal of having
a consistent set of "axioms" for the Web are at odds.  Finding a consistent
set of axioms may not be possible, but finding a consistent set of axioms
about ANYTHING under the constraint of not ever changing your mind seems
almost certainly impossible.  (

Thomas Kuhn's less-famous book THE COPERNICAN REVOLUTION gives a quite clear
explanation of how the transition from Ptolemy's view of the solar system to
Newton's occurred by successively giving up dearly held "axioms"  -- that
the earth didn't move, that celestial bodies moved in circular orbits, that
geometry was the ideal language for mathematical modeling ...).  History is
fully of examples of brilliant people who are more or less forgotten (Tycho
Brahe comes to mind) now because they just couldn't give up that one little
"axiom" that they held dear.  The consistency of a SYSTEM of axioms is
what's important, and those that cause inconsistencies have gotta go.  I
think that's what Micah is saying about the "axiom" that Mark points to.

All findings and Recommendations need to be reconsiderable if the TAG is to
succeed in its architectural mission.  Micah's proposal (apparently
referencing some F2F discussion) to resolve the abstract URI dilemma seems
very sensible to me:  There is a new scheme (or perhaps two) for resources
that are either abstract and don't exist anywhere such as namespace names,
and for the proverbial "car" that physically exists but can't be accessed in
a meaningful way via the Web.  

Thus, if the "now" scheme (for "nowhere" I guess) were adopted, then:

now://www.w3.org  is the W3C itself for purposes of identifying and
reasoning about the abstraction "World Wide Web Consortium"

http://www.w3.org is the W3C website

now://me.example.org/MyCar is for abstractly identifying the thing that sits
in the driveway. It may have some type property with the value "Honda".

http://me.example.org/MyCar identifies one or more representations of the
thing.  It may be have some type value with the property "JPEG".


So, what breaks (besides some previous axioms, which clearly have not led to
consistent "theorems" about the Web) if this approach is taken.  I might add
that it resolved the persistent debate about namespace URIs at one fell
swoop ... 

now://www.example.com/Widget666 is the abstract name for my Widget namespace
http://www.example.com/Widget666 is the URL of a machine-processable
description of that namespace

Maybe a "now://" is functionally the same as a urn://  and all the problems
with URNs re-emerge (I dunno, because I never understood the argument
against using URNs to name abstractions ...) Anyway, at least a couple of us
would appreciate a clear explanation of why this "now://"" scheme to refer
to URIs that are "nowhere" on the physical web doesn't work.  It sure seems
a lot more intuitive than overloading "#" to sometimes refer to locations in
representations and sometimes referring to abstractions that aren't located
anywhere except in URI-space.

Received on Sunday, 6 October 2002 11:10:46 UTC