Re: ACTION-682 suggest to TAG sections of HTTPbis specification that TAG should review

On Tue, 2012-05-22 at 17:48 -0400, Jonathan A Rees wrote:
> ACTION-682 suggest to TAG sections of HTTPbis specification that TAG
> should review
> 
> TAG members who care about this kind of thing should probably check
> the following sections to see how they like them.
> 
> Part 1:
> 
> http://tools.ietf.org/html/draft-ietf-httpbis-p1-messaging-19#section-2.7.1
> 2.7.1.  http URI scheme
> 
> esp. paragraph beginning "Although HTTP is independent of the
> transport protocol"
> 
> I find it peculiar that there is no discussion of what http: URIs
> identify, or how they come to identify anything at all.

This is a protocol specification -- not a logic specification.  It
*shouldn't* be discussing issues that are irrelevant to the protocol.

The fact that you are attempting to read HTTPbis and other related
documents (AWWW, RFC3896, etc.) as though they are a unified set of
logic specifications for indicating the global meaning of a URI and
finding that they fall short is *your* problem -- not the fault of those
specifications.

[ . . . ]
> FWIW I complained about a detail in section 5 of part 2  here:
> http://lists.w3.org/Archives/Public/ietf-http-wg/2012JanMar/0412.html
> and elaborated a bit here:
> http://odontomachus.wordpress.com/2012/02/09/when-identification-and-representation-fight-who-wins/
> Basically I'm saying that the words "identify" and "representation of"
> are being used in the spec as if they are meaningful, but it is very
> difficult to see what, if anything, they mean.
> I have not received any response.

Your blog entry says:
[[
1. Web architecture suggests that a URI owner is an authority for what
is identified by its URIs (AWWW 2.2.2.1 bullet #2).

2. The HTTP protocol suggests the URI owner is an authority for what is
a representation of what is identified (HTTPbis v.18 part 2 section 5.1
bullet 1 taken together with part 1 section 2.7.1).

If both kinds of authority hold, then Jabberwocky is a representation of
the Magna Carta, since a URI owner can say both that the URI identifies
the Magna Carta and that Jabberwocky is a representation of what is
identified. But this is not true. How to resolve this paradox?

There are (at least) three solutions, based on modifying either of the
two authority axioms.
]]

Those are not axioms in a logic system, they are explanations of how the
web architecture works and how HTTP works.  This "paradox" is solved by
reading these specifications the way they were intended -- not as logic
specifications.  

Your analysis of this "paradox" places great emphasis on distinguishing
between *claim* (which might be false) and universal *truth*, such as
distinguishing between "what the URI owner *said* was a representation"
of a resource and the truth of whether that thing really *is* a
representation of that resource.  This is fundamentally misguided.

You seem to think that the purpose of the semantic web is to enable
people to make statements that express universal truths in formal logic.
It is not.  The purpose is to facilitate useful machine processing of
web information.  Whether users happen to interpret that information as
expressing universal truths is their own business.  The fact that you
may interpret a URI owner's statements as contradictory is irrelevant.
The web was designed to allow "anyone to say anything about anything" --
including things that are contradictory.  That's a feature, not a bug.

The notion of universal truth is completely *irrelevant* to the
architecture of the semantic web, just as the meaning of an HTML page is
completely irrelevant to the HTTP protocol.

HTTP just shovels the bits around; the semantic web just shovels the RDF
around.  Certainly we users may care about the meaning of those bits or
that RDF, but the HTTP protocol and the semantic web architecture do not
and should not.

It would be extremely difficult to understand the HTTP protocol in terms
of the meanings of the messages transferred.  Similarly, your quest to
interpret web architecture as a formal logic makes your task of
understanding it extremely difficult.  

Try forgetting about universal truths, and looking at this as an
*engineering* problem of designing the architecture to best enable
machines to usefully process web information.  The problem gets a whole
lot simpler.

>From an engineering perspective one can begin to understand the purpose
of the role of "URI owner" in semantic web architecture by asking: Why
was the notion of URI owner included in the architecture at all?  What
purpose does it serve?  The purpose is *not* to provide an oracle for
determining the truth about what resource really *is* identified by a
URI.  The purpose is to provide a simple, efficient *mechanism* for
independent parties, acting without knowledge of each other, to obtain
the same (or similar enough) definitions of that URI, regardless of
whether a definition is implicit in a protocol response code (such as
HTTP 200 OK) or explicit in a document.  In this sense, the URI owner
simply acts as an arbiter, selecting one particular URI definition over
all others that might otherwise be proffered.  The reason such a
mechanism is important to semantic web architecture is that it enables
RDF statements that are authored independently to be more sensibly
merged if they are written based on the same URI definitions.  And
merging independently authored RDF is the holy grail of the semantic
web.

The whole question of whether the Jabberwocky really *is* a
representation of the Magna Carta is completely irrelevant.  All that
matters is that it was served with a 200 OK status in response to an
HTTP GET request on a particular URI.

The text of HTTPbis does not need to be "fixed" to conform to your
personal attempt to view it as stating axioms in a formal logic for the
web.  



-- 
David Booth, Ph.D.
http://dbooth.org/

Opinions expressed herein are those of the author and do not necessarily
reflect those of his employer.

Received on Wednesday, 23 May 2012 16:43:46 UTC