Re: Comments on Section 3

(as you can see, I'm a bit behind on email)

   Date: Tue, 15 Apr 2003 07:54:09 -0700
   From: Kevin Smathers <ks@micky.hpl.hp.com>
   Cc: matsakis@MIT.EDU, www-rdf-dspace@w3.org
   Content-Disposition: inline
   X-SBClass: Nonlocal Origin [156.153.255.237]
   X-Spam-Status: No, hits=-4.4 required=5.0 tests=IN_REP_TO version=2.20
   X-Spam-Level: 
   X-SpamBouncer: 1.5 (2/23/03)
   X-SBPass: NoBounce
   X-SBClass: OK
   X-Folder: Bulk

   Hi David,

   On Mon, Apr 14, 2003 at 06:17:09PM -0400, David R. Karger wrote:
   >    The main problem that I have with this approach is that I don't consider
   >    two files to be the same just because they happen to have the same 
   >    content.  Likewise two facts.
   > 
   > If by file you mean a specific arrangement of magnetic particles on a
   > specific disk drive, I agree with you.  But I also think it valuable
   > to envision a unique, platonic ideal of a certain bit sequence (and it
   > is that I want to associate with an MD5).  A lot of the assertions
   > that people make about a file are, I think, actually about those
   > platonic bits.  "These bits were written by David" (and ditto for much
   > else of the dublin core), "These bits are 100 bytes long" and "these
   > bits are currently stored at http://foo".
   > 

   Actually by file I mean an entry in a directory structure, on an 
   operating system which maintains directory structures.  

and there are indeed some assertions to be made about that directory
entry, but many others that are independent of it (and that the user
would love to see persist even if the file is "moved" to a different
directory)

   > I feel the same way about statements (I'm not going to try to define
   > facts).  An "a R b" statement is unique.  If two people make the same
   > "a R b" statement then that is exactly what happened: they asserted
   > the SAME statement.

   Just because the content of the statement is identical doesn't mean that
   you can validly collapse all of the statements with that content to a 
   single instance.  The content will remain identical even if there are
   multiple instances, but the address distinction between the statements
   will be lost if the instances are collapsed.  Neccessarily therefor, 
   collapsed statements represent a loss of information.

   The only time that you can validly collapse statements without a loss
   of information is when the statements are both inaddressable and 
   immutable.

Interesting issue here.  Why should anyone care about the "location"
of a statement in the sense of which computer is storing it?  I can
certainly imagine wanting to record where (in the physical world) the
statement was originally asserted, but this isn't really what the URL
is being used for---the URL just tells us where the statment can be
"fetched".  One might care about the location as a kind of proxy for
the creator that can be used to decide trustworthiness (ie statements
with URL cnn.com are presumably "backed" by CNN), but I suspect that
the ability to use location as a proxy in this way is going to decline
rapidly as RDF gets more popular.  But why use a creator when
one can instead just add creator metadata?   In any case, a
just-as-safe proxy for creator would be "who served the statement" as
opposed to "who is in the statement URL"

   Your argument that the users intent all along should have been to 
   assert the same instance as had been asserted previously is presuming 
   to know the intent of the user.   If the user had that intent, then 
   there is no reason for them not to use the preexisting statement 
   directly.

Received on Thursday, 22 January 2004 16:49:10 UTC