At EthCC and Nebular (2023), I talked to several people about privacy. I realized that one important aspect of privacy is often overlooked: meta-data privacy. Even if you use ZKPs or content-hiding tooling, exposing meta-data on the network level can enable various attacks and violates privacy according to many definitions in [1]. Therefore, my main question for now is: Is network layer privacy a priority for Anoma? If yes, I want to spend some time on this and talk to people from other projects for potential collaboration (e.g., NYM [2]).
[1] Pfitzmann, A., & Hansen, M. (2010). A terminology for talking about privacy by data minimization: Anonymity, unlinkability, undetectability, unobservability, pseudonymity, and identity management.
[2] Diaz, C., Halpin, H., & Kiayias, A. (2021). The Nym Network.
Network-layer privacy is definitely something we need long-term. In particular, I think we want to integrate techniques such as mixnets in a way which preserves the user’s ability to trade between trust, efficiency, and privacy in a runtime-determined manner - for example, sending messages on a LAN between cryptographic identities of my friends, I probably don’t need to add mixnet-style packet routing and delays, but when using the public internet, I do. It would be interesting to figure out if this is possible with existing designs & software such as Nym - I’m not sure.
I agree we need network-layer privacy, and should also support user-specified tradeoffs. We would most likely need to support multiple transports for this, and the P2P engine should choose an appropriate transport based on these preferences.
Nym/Tor could be one of these transports but definitely not the only option.
Another project that looks interesting and we should look into is Earendil (blog, git, docs). The author gave a talk about it at AE2, which will be online later.
In the Circles Entropy blackpaper we mentioned both NYM and our friends at https://katzenpost.network/ as potential ways of doing network level privacy. Afaik, the issue with the current NYM mixnet is that for a payment app you would basically need 60 gb which is not really practical at the moment, although they’ve been very open for collaboration and making their stuff adaptable to different apps.
@Ajmaq True. Additionally, mixnets are not very practical in terms of bandwidth and latency costs. I chatted with the Nym people and they are currently making an effort to optimize the protocol. That being said, we should also be open to other solutions than mixnets.