[{"author": "Nat Sakimura", "text": "

The draft I talked about is https://datatracker.ietf.org/doc/html/draft-sakimura-oauth-meta-08

", "time": "2023-03-31T01:58:41Z"}, {"author": "Michael Sweet", "text": "

FWIW, with the Internet Printing Protocol we are trying to have a printer report which auth server to use - on the client side we are requiring an allow list, i.e., the client doesn't just blindly trust the URL the printer reports. Instead, the client has a configured list of allowed and denied auth servers (which can be configured by the admin and/or user depending on the environment) and if the printer reports one that is not allowed then the user gets a Big Scary Error and you don't go any farther. I don't know whether a similar approach could be used for the more generic Resource Server providing metadata.

", "time": "2023-03-31T02:02:25Z"}, {"author": "Michael Sweet", "text": "

For functionality, the WWW-Authenticate approach is useful when you have multiple auth servers in use, while .well-known force one auth server to rule them all...

", "time": "2023-03-31T02:05:38Z"}, {"author": "Roman Danyliw", "text": "

CIBA = https://openid.net/2021/09/01/openid-connect-client-initiated-backchannel-authentication-ciba-core-is-now-a-final-specification/

", "time": "2023-03-31T02:25:32Z"}]