U-Prove seems to me to be pretty close to a perfect auth system. It is possible to disclose only specific attributes and every prove is unlinkable (given no unique attribute is disclosed). Also it supports generating an unique, identity-linked ID per domain.
So I wonder why this technology is not used anywhere I know of?
I didn’t read to much of the FIDO2 spec, so I can’t really compare.
But U-Prove can be used for state-issued E-IDs. Is this also possible with FIDO (including dynamically issuing attributes)?