Re: [Privacy-pass] context unlinkability

[Christopher Wood <[e-mail]>]

Thanks, Nikita! This is a nice, succinct change that I think we should include. I merged the PR into the architecture document.

Chairs, we can rev the document again if needed. Please let us know how you’d like to deal with this.

Best,
Chris

> On May 9, 2023, at 9:47 AM, Nikita Borisov <nikita.borisov=[e-mail]> wrote:
> 
> The current architecture document defines three types of unlinkability (Origin-Client, Attester-Origin, and Issuer-Client). These maintain client privacy within any single context. However, I think an important property of privacy pass protocols is that the issuance (and attestation!) contexts should be unlinkable to the redemption context. This is actually mentioned in the Overview under point 5, but it isn't listed among privacy goals. I added a pull request (361) to add this property to the architecture document.
> 
> I think this is important because without it, an issuance protocol that adds, e.g., a serial number to each issued token which MUST be shown at redemption time would still satisfy the three original unlinkability properties. 
> 
> Thanks,
> - Nikita
> -- 
> Privacy-pass mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/privacy-pass