>Back in January on NGC4LIB I proposed doing this, a universal ID system
>use when browsing, using the FOAF structure.  I got back answers that
>me they were not getting the concept.  This discussion on OpenID is very
>interesting and I hope this can be made to work.

Hi Steven,

Tim Berners-Lee [1], among others, have brought up FOAF
(Friend-Of-A-Friend) in connection to OpenID as a way to establish trust
networks. You could argue that the attributes support in shibboleth could
accomplish the same thing, but the difference might be that people like
Sir Tim are seeing some synergy in OpenID and FOAF whereas shibboleth
doesn't seem to capture the attention of the mainstream web folks. Not
that FOAF is without its own detractors, but here's a variation of a
syntax I have seen for indicating trust, in this case, in Ed Summers' Ruby
knowledge (the syntax is a little dated but you get the idea). Extend this
to indicate, for example, a fondness for the music of Howlin' Wolf, or
some other kind of preference information, and creating systems that
respond dynamically to user background and preferences might be possible :

<foaf:Person rdf:about="#edsu">
<rdfs:subject rdf:resource="" />

OpenID implementations may already have a little plumbing for this kind of
thing with "personas" but it still comes back to how much a service is
willing to accept from a particular OpenID provider.  I would be curious
whether the above kind of syntax could fit into the Yadis system used by
OpenID because I am very unclear how FOAF and OpenID could/should
intersect. There is also talk about OpenID support being built into
browsers, things would get really interesting if the web browser started
to broadcast an OpenID to web services.