gpestana
commented
6 years ago Metadata resistance
3 defines metadata resistance as a property and threat model that can be used here.
Open questions:
Q1: what are the transport and routing protocols that IPFS nodes can chose to enhance privacy (over, say, performance). implementations?
Traffic analysis:
IPFS uses content addressing of location addressing, which means that the nodes where the user downloads files is not directly linked to what the used is downloading. Q: how to perform attack to link accessed content with address itself?
Bootstrapping nodes:
When using bootstrapping nodes, usually a lot of user metadata can be inferred from the process.
DHT:
A lot of metadata is being leaked in the DHT. examples how (see priv. trello)
Private DHT:
There are research efforts to build a private DHT which aims at hiding as much DHT metadata as possible.
Content/DAG encryption: (bonus)
Metadata resistance != end to end encryption, but IPFS does not encrypt in-wire by default. For more on this, check private-content label on ipfs/notes.
P2P network are potentially more secure than centralised services in terms of metadata leakage and censorship resistance, since they don't rely on centralised authorities which make it easier to tear down and/or manipulate critical nodes. However, P2P networks are not censorship resistant nor metadata resistant by default. Those properties have to be accounted for when designing and implementing the network protocol.
The goal of this thread is to gather information about how the IPFS protocol and network and the different transport and routing schemes used by IPFS nodes perform in terms of censorship and metadata resistance. It also explores how these properties can be improved by enhancing and/or adding more privacy and censorship resistant transport and routing schemes.