[rdfweb-dev] Tags, not mboxes?

[Adam Atlas]

If draft-kindberg-tag-uri-06 [1], which registers the Tag URI scheme 
[2] is approved as an RFC, what would be the feeling of this group on 
encouraging the use of tags as unique identifiers for FOAF resources 
instead of relying on InverseFunctionalProperties such as email 
addresses and chat IDs? This is unreliable, because on some occasions, 
people share these, and more importantly, people change these. The main 
advantage of using a tag is that it itself becomes the URI of the 
resource (rdf:about="tag:..."), instead of RDF document URLs with 
fragment identifiers (i.e. http://example.org/myfoaf.rdf#me -- what if 
the original document moves?), misleading HTTP URLs serving only as 
identifiers (i.e. http://example.org/rdf/me -- probably 404! And if 
not, then it's a web document, not a person!) or worse, bnodes that are 
impossible to reference, all of which I've seen used. In my opinion, 
tags improve readability, as well as making it easier for RDF 
processors to connect the dots.

Of course, people can already do this (<> dc:creator 
<tag:adamatlas.org,2005:/me>.), but if the internet-draft for tags is 
approved, how would people feel about having the specification 
recommend identifying agents by unique tags rather than associating 
them with InverseFunctionalProperties or HTTP URLs?

-- Adam

[1] http://www.ietf.org/internet-drafts/draft-kindberg-tag-uri-06.txt
[2] http://www.taguri.org/