In the linked description there appears this bit “… server then returns to the client’s browser the: … the user’s identity server URL …
Perhaps a FOAF URL …” That’s a red flag for me; since FOAF tends to encourage users into revealing more information than is actually necessary.

The second reason was that I didn’t see a clear policy to keep what’s revealed extremely minimal; and it’s necessary complement a means to extend that. Any such mechinism will have to interact with the user to gain his permission, and that notches up the complexity of getting it all right.

Now having been forced to think about this aspect some more I recall that it’s key to get right early that the sources for revealed information not get tied to the vouching entity too tightly. The vouching entity has a advantagous position, as the first stop for info about the user, and a good design strives to keep the vouching agent with as little information about the user as is technically possible.

OpenID explicitly avoids profile exchange because that’s someone else’s problem. I believe the current thinking is that the identity URL can also have a FOAF auto-discovery URL which sites could potentially use, though that’s not part of OpenID and something like this will likely just become a de-facto standard as people start to use OpenID for different things.