RE: Documents, Cars, Hills, and Valleys

> Joshua Allen wrote,
> > Interoperability means that if I aggregate 500 different assertions
> > from different sources, all about http://www.microsoft.com, I know
> > that they all are talking about the same "thing".
> 
> Yes, but _what_ thing is that? I guess it's possible that there are
> 500 different different assertion about that document. I'm sure that
> there are many more which use the very same identifier to refer to
> Microsoft Corporation.
> 

That is exactly the point.  There *will* be thousands of assertions
about both the home page and the company, but only if people can trust
that their assertion will apply to the right thing.  If I publish
assertions about http://www.microsoft.com and there are some other fools
using the exact same URI to publish assertions about the *organization*
that maintains www.microsoft.com, then my metadata is absolutely useless
(except in my own closed system).  We might a well go back to the days
of hypercard, since metadata will be useless.  Unless people can know
that a particular URI identifies one particular "thing", there will be
no point in publishing metadata, and there will be no semantic web.