RE: fragment identifiers

> > The *web* is about accessing resources.  The *semantic* web is about
> > *describing* resources, even those which cannot be accessed.
Describing
> > does not and should not require accessing.
> 
> EARL and Annotea are used to describe resources that must be accessed
> for the metadata to be meaningful.  That is clearly not obvious to

Wrong.  The metadata can be created and queried without accessing the
resource.  The resources themselves are resources which are useful only
when they are accessed via HTTP GET.  But that is a limitation of the
resource and the REST model, not a limitation of the metadata.

> > If you are making RDF depend on GET at all, I perceive a serious
> > misunderstanding.

Let me explain.  The *web* (not the semantic web), allows anyone to
hyperlink to anyone else.  For me to hyperlink to your page, your site
does not have to be available, and I do not need to do a GET on your
resource.  Someone reading my page can bookmark your page without ever
doing a GET on it.  

This is a fundamental characteristic of the web, which falls into the
category of "don't even try to mess with it".  Many previous efforts at
hypertext systems tried to eliminate link rot by enforcing a
relationship between linker and linkee, and they all failed.

The concept is even more fundamental to the *semantic* web.  If you must
do GET on a resource to be able to make assertions about it, you have a
brittle system that will never scale.

Received on Friday, 26 July 2002 14:42:37 UTC