Confidence level is not sufficient on its own

[David-Chadwick]

Having confidence that the entity X who is presenting the VC to the verifier, really is the holder of the VC; this is what the confidence method provides. However on its own it is not sufficient. Another layer is needed on top of this.

The verifier typically needs to know what the relationship is between the holder and the subject, and if the holder is authorised to present the subject's VC. For example, if the subject is a child, or a pet, and the holder is the parent, or pet owner, then the verifier needs to have confidence that

a) the presenter is the holder of the VC and

b) the holder is the parent, or pet owner of the subject.

Whether this second layer should be a separate specification, or a second section of this specification, is TBD.

[denkeni]

+1 we should clearly address the use case when the holder ≠ the subject of the VC. Probably as a delegation mechanism.

One way of doing it: #10

[TallTed]

confidenceLevel is entirely subjective, and must be considered part of business logic, no matter what role the evaluator is playing.

confidenceMethod could reasonably be defined as the means by which a confidence evaluator arrived at their confidenceLevel, and the two together might be used as part or all of the inputs another party uses to arrive at their own confidenceLevel.

As I recall how we arrived at specifying these two properties, they were (and are) multi-purpose, i.e., they are not only about "confidence that the entity X who is presenting the VC to the verifier, really is the holder of the VC". Especially, please note that a holder of a VC is simply an entity who has a copy of that VC. In other words, any entity that presents a VC is really a holder of that VC!

There are some use cases and some deployment scenarios which include some restriction on what entities are nominally permitted to present a given VC. Evaluating that restriction, and evaluating whether the presenting entity is so permitted, and any other evaluating that goes on, is business logic and outside the remit of this group.

[dlongley]

In the three party model, authorization and acceptance are determined by the verifier, not the issuer.

A confidence method is just something an issuer can express (in a verifiable credential) as an optional mechanism the verifier might use to establish confidence, for example, that the same party about which they (the issuer) have made claims is the party presenting (e.g., a biometric).

Another example would be that the presenter is able to generate a cryptographic proof that is verifiable by an expressed confidence method. Notably, this alone does not necessarily say anything more about the presenter, i.e., the subject of this confidence method is anyone or anything capable of generating such a proof.