W3C home > Mailing lists > Public > www-tag@w3.org > February 2011

RE: "tdb" and "duri" URI schemes...

From: Larry Masinter <masinter@adobe.com>
Date: Mon, 21 Feb 2011 08:15:56 -0800
To: Jonathan Rees <jar@creativecommons.org>
CC: David Booth <david@dbooth.org>, "www-tag@w3.org" <www-tag@w3.org>
Message-ID: <C68CB012D9182D408CED7B884F441D4D05A00B29B2@nambxv01a.corp.adobe.com>
I had A -> B and you've added another link  A -> B -> C, but it's a graph (a DAG)
of communication, which includes   A -> B, C, D, E, ... (broadcast),

A-> B -> C -> D -> E 

but I convinced myself that those more complicated stories are built up
one link at a time.

If  A -> B and A -> C, then A, B and C should share some agreement that
B and C understood the "same" thing, and would have some common idea
of what "same" meant, and that A's choice of URI (scheme, path, longevity
of services involved in maintaining the mapping between URI and resource)
would be essential to the reliability of the communication.

In this case, no message passes between B and C, yet they share
a common understanding. 

How reliable would corporate financial data be, if everyone going to
get the annual report got a different one? If the auditors get a different
set of "books" than the stock analysts, for example? If politicians
could publish different statements about their platform depending on
the politics of who was reading?

Part of the trust relationship is not just the provenance and reliability
of the communication between the two parties involved in an instance
but also the uniformity.

> In your account you need a disclaimer that by 'resource' you mean
> 'network resource' or 'resolvable resource'.

I don't think I need such a disclaimer or intended that there be one.
"tdb" is clearly an example of method for identifying other kinds
of resources, so 

>  I'm confident that sense of the word is different from what
> you mean here by 'resource' (although 'determine' makes me wonder).

I don't know why you're confident I meant the narrower interpretation.

> I think it's important to recognize that in David's account 'URI
> owner' and 'resource owner' are orthogonal

I recognize that he meant that; in fact, it is the entire concept
of "owner" that bothers me and I was working hard to avoid.

"Brought to you by the letter A and the number 3" is funny because
there is no "owner" for the letter A or the number 3 (unless
you think the Unicode consortium is the 'owner' of letters?)

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



-----Original Message-----
From: Jonathan Rees [mailto:jar@creativecommons.org] 
Sent: Monday, February 21, 2011 6:45 AM
To: Larry Masinter
Cc: David Booth; www-tag@w3.org
Subject: Re: "tdb" and "duri" URI schemes...

On Fri, Feb 18, 2011 at 6:55 PM, Larry Masinter <masinter@adobe.com> wrote:

> I would write the lifecycle as:
>
> 0. Resource comes into being
> 1. Someone (Party A) makes up ("mints") a URI to identify the resource
>   (This 'someone' doesn't have to be an 'owner' or have any relationship to the resource)
> 2. Party A then communicates this URI to party B ('uses the URI in a statement')
> 3. Party B  reads the statement and attempts to 'resolve' the URI [determine/
>             contact/interact with] the resource

This is good, but I think a better version of the lifecycle (one that
reflects better the reality) has an additional step:

0. Resource comes into being
1. Someone (Party A) makes up ("mints") a URI to identify the resource
  (This 'someone' doesn't have to be an 'owner' [of the resource] or
have any relationship to the resource)
2. Party A then communicates this URI to party B ('uses the URI in a statement')
3. Party B communicates this URI (in a second statement) to party C
4. Party C  reads the 2nd statement and attempts to 'resolve' the URI
[determine/
            contact/interact with the resource]

That is, a 'definition' of a URI can be used by two non-defining
agents in meaningful communication. The dictionary writer enables two
agents in possession of the dictionary to communicate.

For example, A might be an agent 'minting' an http: URI for a
document, who communicates the URI to B by transmitting email to B
containing the URI. Then B puts the URI in a document s/he controls,
which C reads, and C tries to resolve it.

The point being that once B knows the URI, B can 'introduce' the
resource to C using that URI. (The 'Granovetter diagram.') C and A
needn't precoordinate any more than randomly selected pairs of agents.
If there are only two agents, there is only bilateral communication,
and much of the richness of the Internet/Web - standards, registries,
DNS, etc. - isn't needed.

In your account you need a disclaimer that by 'resource' you mean
'network resource' or 'resolvable resource'. RFC 3986 and RDF both
permit 'resources' that you can't interact with, such as the earth's
core, and I'm confident that sense of the word is different from what
you mean here by 'resource' (although 'determine' makes me wonder).
This is a simple terminological spat that can be cleared up with a
definition or citation to a definition.

I think it's important to recognize that in David's account 'URI
owner' and 'resource owner' are orthogonal. *If* I were to convince
you that http://example/magnacarta is a URI 'identifying' the Magna
Carta, this may be because I exercised 'ownership' over that URI
(perhaps I own the domain). That doesn't mean that I own the Magna
Carta. Whether we want to talk in this way is a separate matter; but
it is a self-consistent way to talk.

Jonathan
Received on Monday, 21 February 2011 16:16:33 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:48:30 GMT