What follows are my lunatic rantings in response to Plaxo and FOAF: What's the right model? and the posts / discussions Danny Ayers links to in his A trail of FOAF. My ranting is not meant as a comment on those posts, which I think are quite sensible.

FOAF data appears in files. Some sites are now trying to support FOAF, but they are sites that report from things like databases, which resports are information culled from data not particularly locked into files. Sites' information, and data files are very different, architecturally.

So, sites are getting stuck matching their information with FOAF files, and FOAF files are getting stuck inside sites. I think this issue is being approached (generally) by trying to the make the sites' information more matchable to the data files and otherwise to make the files' data more dependent on the sites' information.

One example of this approach is RSS and blogs, where the sites and the RSS files, in some sense at least, have been made to look like each other, architecturally. However, the scope of this matching is limited to blogs that are restricted in their information architectures, and RSS is used in a way that is dependent on blogs or the parts of sites that look like blogs.

I think there is a scope of match-up between sites that have information about people and their relationships, and FOAF data files. It is good to match those up, but I don't think the architectures of the information on sites (e.g., with all of its context, including things like privacy and access control) and the data in the FOAF files can be made to fully match up.

I also think another approach is to create a different kind of mechanism that isn't a file and isn't a site, but express the same kind of information and the same kind of underlying data.