"Floating" vs "rooted" URIs

Hello all,

While looking through the Geonames project's ontology
(http://www.geonames.org/ontology/ontology_v2.0_Lite.rdf), I noticed
that they define things in the following fashion:

    <owl:Class rdf:about="#Class">

In ontologies I've put together, I've tended to use formulations along
the lines of:

    <owl:Class rdf:ID="Class">

I can see the difference as being that, when written in an ontology
available as http://example.com/ont, both say "here we assign
properties to a URI", but only the latter version says "and this is
it", providing a sort of physical presence as an identified fragment
within the document.

Is the former approach a better route to take? It feels less
restrictive in some fashion. Or is there not really an appreciable
difference? I suppose that with the open world assumption in effect,
it's not as if either approach has any bearing on external resources
making statements about my URIs, either.

Cheers,

Earle.

-- 
قبائلَ صوتي – على صمتها
Earle Martin | http://downlode.org/