Open apreiml opened 3 years ago
If i understand correctly (and correct me if im wrong), these papers propose alternate ways of discovering other users based on the Third-party identifiers (3PID), while still keeping the privacy of these users and their 3PIDs.
If that is true, then I think this paper assumes this on a centralized model? I.e. one server where all data is collected and cross-referenced, but in a "blind" way. How could this (potentially) be expanded in a model like matrix? Could/would servers have "trust" in other servers (either gained implicitly or explicitly, per-server and/or per-user), or would this be in an open "broadcast" mode?
How do you envision this working for an open federated model like matrix? (Where you cannot trust each and every server implicitly)
I just wanted to share an interesting paper about contact discovery and how to do this privately:
https://eprint.iacr.org/2019/517
Also related:
https://eprint.iacr.org/2017/670
https://eprint.iacr.org/2020/1119