[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

RE: Mailmen, POST, Intent, and Duck Typing

• To: "Andrew S. Townley" <andrew.townley@b...>
• Subject: RE: Mailmen, POST, Intent, and Duck Typing
• From: "Bullard, Claude L \(Len\)" <len.bullard@i...>
• Date: Fri, 24 Feb 2006 11:25:52 -0600
• Cc: <xml-dev@l...>
• Thread-index: AcY5YnheSIv5ire4RBmAbMcKJC8NoAAAyBEg
• Thread-topic: Mailmen, POST, Intent, and Duck Typing

Thanks, Andrew.  I understand that you understand. :-)  
I agree with what you say.  Control is never absolute. 
In olden days, I called that the triple-omni problem: 
perfect acts require perfect beings and systems, so 
you have to be God. Omniscient, omnipotent, and omnipresent.

Short of that, it's all about shared risks.  Shared risks 
are why open source works so well.

Security is an explicit act to restrict implications.  See 
last email to Chris Burdess.  The failure to secure data 
implies it is public given REST and http and URIs.

Yes.  Ontological domains aggregate and bifurcate.  They 
are the quintessential living document.  Yet as John Sowa 
points out on the CG list, you can scrape knowledge from 
extremely old artifacts where all of the living witnesses 
are dead, so intent based on querying the author isn't 
required. (It isn't always possible to ask the cat so 
uncertainty is a fact of dynamic process with multiple 
controls).  

That means (D'oh) that our environment and our natures change slowly 
enough that we can do a good job or discovering implications. 
What one hopes is that one doesn't stumble into the cave 
where we stored the nuclear wastes in 10k years unable to 
read the symbols.  That would be bad.   Or be too anxious 
like the guide in the first Indiana Jones movie and find 
yourself eyes wide open but pinned to the wall.

To stop that sort of thing, the REST design tries to make 
the act of using the URI/HTTP technology as free of 
assumptions and implications as possible.   What do the 
SOAP advocates answer in reply?  Essentially, it is a 
dangerous world and you might want to share more assumptions. 
The ontologists say, 'we'll make it possible to share 
assumptions and implications as widely as you can afford 
to share them'.  Again, pragmatically, there is no free 
semantic lunch, so the pragmatists try to figure out 
the cleanest way to negotiate agreements by formally 
declaring implications and recognizing that these include 
first and second order implications.

len


From: Andrew S. Townley [mailto:andrew.townley@b...]

On Fri, 2006-02-24 at 16:08, Bullard, Claude L (Len) wrote:
> As I recall, the reason for UDDI and other discovery processes is 
> to provide a facade of what is public knowledge so that one can 
> then begin a negotiation to get more private knowledge, therefore, 
> establish a relationship.  Typical contracting works like that 
> and I won't get into details here.

While it may not have been clear from what I said earlier, I understand
what you're saying here.  I was just trying to make the point that there
are times when a discovery process of this nature won't fly because of
the type of your application or your security requirements (not very
well, apparently ;).

> Pragmatics takes up that aspect of communication.  Ontologies are 
> a way to get that done but there is no free semantic lunch.  Standard 
> ontologies are one answer, but you'll find the debate about the 
> uppermost ontology to be fatiguing.   When you look into business 
> intelligence systems (eg, applications of OLAP cubes vs massive 
> indexing over distributed resources), you will find the ontology 
> challenge yet again.   What I am still sorting out is if there 
> is a clean separation of the pragmatic and the semantic layer. 
> I sort of doubt it but I'm still learning.

Yes, but you're working from the premise that the Semantic Web of
everything on the planet is a fundamentally good idea.  I can see some
of the merits, but I'm not 100% convinced that there shouldn't be
different strata, perhaps orthogonal, and that this isn't ok.  Society,
and even information/representations, need to have some clear
delineations at times.

Isn't it being pragmatic to realize that you can exert some control and
structure over a particular corner, and then define how that does (or
doesn't) relate to the rest of the universe?  Or, maybe if the semantics
can be identified, then an agreement on the invocation mechanism can be
achieved?  Maybe not.  If not, then don't you risk diluting the
semantics so much that they become meaningless and you're searching for
something else.  You allude to this type of problem with your document
example below.

• Prev by Date: RE: Mailmen, POST, Intent, and Duck Typing
• Next by Date: Re: xml xsl web architecture
• Previous by thread: RE: Mailmen, POST, Intent, and Duck Typing
• Next by thread: RE: Mailmen, Post, intent, and Duck Typing
• Index(es):
  • Date
  • Thread