Add did-auth as initial confidence method

[jandrieu]

In the spec today, we have discussion of different kinds of confidence methods, but without discussion of how those are processed.

https://w3c.github.io/vc-confidence-method/#example-usage-of-the-confirmationmethod-property-of-type-verificationkeyconfirmation

Possibly use material from the VCALM spec, e.g., https://w3c-ccg.github.io/vcalm/#the-did-authentication-response-format

Also, OID4VCI and OID4VP there are ways to provide proof over a DID.

[jandrieu]

Discussed today. "did-auth" is maybe not the right map to confidenceMethod.

"DID Auth" was meant as a short-hand for public key proof-of-use.

Also, it's one thing to say there's a person associated with a DID versus, the subject of this VC has a suitable confidence method usable to improve the verifier's confidence that the presenter is the subject.

Another way to establish confidence is BBS psuedonyms. That might be interesting to standardize.

@PatStLouis wanted to discuss further and we all shared the sentiment.

[jandrieu]

Ok. I think the right way here is a verification method not "did-auth"

[denkeni]

At TPAC, the editors met and think we should clarify the scope of confidenceMethod and the existing verificationMethod. There’s no specific terminology definition for verificationMethod but there’s one for verification material. I think it’s reasonable for verification material to stay in verificationMethod, and put anything else in confidenceMethod.

[iherman]

This was discussed during the vcwg meeting on 14 November 2025.

View the transcript

Confidence methods

subissue: w3c/vc-confidence-method#20

JoeAndrieu: I think did-auth is a good candidate as a first authentication method

manu_: +1 to some variation of did-auth
… If I understand what you are saying: presentations today amount to do a signature
… you propose to add a public key that people should use when presenting
… to denkeni: there are other methods, e.g. challenge via email address, MFA
… biometrics base
… we want to get this first one done, but I would like to see a list of other methods that people are interested in

JoeAndrieu: +1; let's finish one and discuss the others
… I think what you described 2 types of methods
… one is embedding something in the VP establishing confidence
… another is going through another channel; I want to enable that
… what we have is not how issuer establish confidence, but hints to help the verifier establish confidence

phila: I'm wondering what the crossover is between this idea of establishing confidence in a VP, and what we discussed yesterday and briefly this morning?
… Am I wrong?

JoeAndrieu: that's brilliant. That's a good case of that attribute.

denkeni: we have something similar in Taiwan.
… We require 2 credentials (e.g. banking and telecom). That's one way to improve confidence.
… Not sure it is a good place to talk about that, it is verifier's logic.

Kazue: one part is confidence in the holder, the other part is confidence in the issuer.

phila: yes, but both are about what you need beyond/around the main credential to establish confidence

JoeAndrieu: do we still have the "evidence" property? it was at risk at some time...

manu_: yes, we do

phila: in our own case, I have been talking to our legal team, asking how we find out if someone is authorized to talk in the name of the organization.
… I would use the "evidence" property to point such a document. Not exactly the same thing, but related.

Wip: I think it is good to separate the confidence in the presenter and the confidence in the issuer.

manu_: the confidence method can be attached to any subject. You can have 20 subjects, each with their own confidence method.
… In a parent-child relationship, you may have one confidence method about the parent, and another one about the child.
… It is not just about the presenter.

---