Why Nostr? What is Njump?
2023-03-28 13:09:51
in reply to

Cyph3rp9nk on Nostr: In keet it is very easy to find out the user's ip since it is p2p along with his ...

In keet it is very easy to find out the user's ip since it is p2p along with his fixed id with which you could create a communication graph, keet is very bad in terms of privacy.

On the other hand I attach the official documentation of SimpleX comparing it with p2p clients.

Comparison with P2P messaging protocols

There are several P2P chat/messaging protocols and implementations that aim to solve privacy and centralisation problem, but they have their own set of problems that makes them less reliable than the proposed design, more complex to implement and analyse and more vulnerable to attacks.

P2P networks use some variant of DHT to route messages/requests through the network. DHT implementations have complex designs that have to balance reliability, delivery guarantee and latency. The proposed design has both better delivery guarantees and lower latency (the message is passed multiple times in parallel, through one node each time, using servers chosen by the recipient, while in P2P networks the message is passed through O(log N) nodes sequentially, using nodes chosen by the algorithm).

The proposed design, unlike most P2P networks, has no global user identifiers of any kind, even temporary.

P2P itself does not solve MITM attack problem, and most existing solutions do not use out-of-band messages for the initial key exchange. The proposed design uses out-of-band messages or, in some cases, pre-existing secure and trusted connections for the initial key exchange.

P2P implementations can be blocked by some Internet providers (like BitTorrent). The proposed design is transport agnostic - it can work over standard web protocols, and the servers can be deployed on the same domains as the websites.

All known P2P networks are likely to be vulnerable to Sybil attack, because each node is discoverable, and the network operates as a whole. Known measures to reduce the probability of the Sybil attack either require a centralized component or expensive proof of work. The proposed design, on the opposite, has no server discoverability - servers are not connected, not known to each other and to all clients. The SimpleX network is fragmented and operates as multiple isolated connections. It makes network-wide attacks on SimpleX network impossible - even if some servers are compromised, other parts of the network can operate normally, and affected clients can switch to using other servers without losing contacts or messages.

P2P networks are likely to be vulnerable to DRDoS attack. In the proposed design clients only relay traffic from known trusted connection and cannot be used to reflect and amplify the traffic in the whole network.
Author Public Key
npub1lnms53w04qt742qnhxag5d6awy7nz6055flnmjkr6jg39hm86dlq7arrnt