@gpestana @anorth, ... . We started by discussing the privacy attack surface of P2P protocols such as DHT routing, peer assisted CDNs and decentralized discovery mechanisms in general. We focused primarily on network level metadata leaks that happen within collaborative networks of untrusted peers. We then discussed some of the privacy enhancing protocols that can be used to solve or at least attenuate the metadata leaks mentioned before. We covered the following ideas/protocols:
-
p3lib-fullrt which consists of requesting a full or partial routing table in the DHT routing protocol in order to not disclose to network peers the node interest.
-
p3lib-sphinx how to delegate DHT lookups with onion routing using libp2p and ipfs. We also saw a couple of demos about this protocol, one of them can be checked here https://www.youtube.com/watch?v=j64C5CTb8J8
-
OctopusDHT for DHT lookup initiator privacy [1]
-
We also bridfly discussed about using Private Information Retrieval [2] databases as a complement to DHT lookups and other decentralized discovery mechanisms in which we can achieve provable privacy and security, even in an adversarial context (ie. with untrusted PIR providers). Early implementation and specs can be now found at p3lib-sinkhole
We wrapped up discussing about the importance of research and development of privacy preserving protocols for P2P networks.
Actionable steps and ideas:
- To benchmark privacy preserving protocols (e.g. onion routing) with libp2p and p3lib;
- Important to lay out clearly what are the privacy concerns in the dweb context;
- More research efforts and work is needed in this area!
[0] p3lib on github
[1] Wang, Qiyan & Borisov, Nikita. (2012). Octopus: A secure and anonymous DHT lookup. Proceedings - International Conference on Distributed Computing Systems. 10.1109/ICDCS.2012.78.
[2] https://en.wikipedia.org/wiki/Private_information_retrieval