Inge Henriksen made me realize that URIs are being used as a shorthand for the {set of properties} that identify subjects.
A user recognizes a subject by observing/recognizing some {set of properties}.
They then choose a URI as the shorthand for the {set of properties} they recognized.
To interchange a URI with others, the other users need to know what {set of properties} map to the URI.
Corollary: If no {set of properties} maps to a URI, there is no interchange.
Well, no reliable interchange.
Inge could use http://psi.ontopedia.net/Inge_Henriksen to identify himself. I could use http://psi.ontopedia.net/Inge_Henriksen to identify Gjetost cheese. If the URI did not map to a set of properties, how would you choose between them?
(Detail: Less than all the properties in the {set of properties} may identify a subject. I’ll talk about that at a later point.)
Coded messages are fine things, as long as the recipient knows what they mean, and nobody else can figure it out with any certainty. A URI can, for example, have the same meaning as one of those codephrases that the BBC broadcast during WWII.
The BBC codephrases were designed to be ambiguous for anyone who wasn’t meant to understand them. URIs, too, make excellent communications tokens between those who are “in the know”. What astonishes me is the abundance of nonsensical claims that URIs, by virtue of their URI-ness, can reliably bear semantic loads from broadcaster/publishers to all of their potential customers. The set of all URIs is not a universe of discourse, even if all computers know how to use them to establish data delivery channels. It’s commonplace for two human beings, both of whom use the Web, to have no universe of discourse in common whereby certain information can flow from one to the other.
Comment by Steve Newcomb — April 28, 2010 @ 4:33 pm