Skip to content
This repository has been archived by the owner on Feb 13, 2019. It is now read-only.

Latest commit

 

History

History
112 lines (93 loc) · 5.94 KB

README.md

File metadata and controls

112 lines (93 loc) · 5.94 KB

Mercury

Mercury is a distributed network built on a revolutionary person-to-person protocol and identity model aiming maximal privacy. Our goal is to make the Internet ours again by true decentralised secure communication, social networking and enabling peer-to-peer business and apps with no middlemen, even on your phone.

Why?

The internet was designed to provide open distributed peer to peer communication, but your phone and PC don't have that anymore, only servers in data centers. You're closed behind ISPs and home routers (e.g. NAT) so you need intermediaries to communicate. Those intermediaries tie you by heavy vendor lockin: you can't change service provider (consider email, storage, etc) without sacrificing your old identity and data. Even worse, the biggest of them make a living from taxing your payments, constantly spying on you for selling your data and usually hinder or censor you for their advantage or political orders, already much further than just targeted ads. Just as an appetizer, some Asian countries require a mandatory digital identifier for every payment you make and they plan to punish or reward their citizens based on their collected digital data, essentially building the Thought Police.

Mercury aims to brings balance by protecting you from all of this. Your identity is a cryptographic key owned by you and noone else. Data storage and communication is organized around this identity which you can keep even changing service provider or application. Your data is encrypted until you decide to publicly share. Furthermore, you can have several unconnected identities, e.g. for work, family and hobby. The network is truly distributed and built on encrypted peer to peer communication so you're safe. You can add your own node to the network under your control and trust.

In the end you can get rid of intermediaries or middleman and directly connect persons, business with clients or even machines, returning from oligopoly to true competition.

Comparison

Mercury is somewhat similar to a cellular mobile network, it provides features similar to SMS, calls, data connections, push notifications, etc, but

  • built as an overlay network on top of any transport layer (currently Tcp but could use Tor, I2P, mesh, etc)
  • your own "cell tower" can join or leave the network any time
  • selecting a "provider" is the only trustful part of the system
  • uses cryptographic keys instead of phone numbers for p2p encrypted communication, cell/provider and applications use the same kind of identity
  • user data and calls are encrypted, you cannot be spied on or be cheated with contact identity
  • you're free to keep your identity and contacts moving to another provider or application
  • supports you having different unrelated identities (family, professional, dating, etc) within the system
  • you can restore your identities from a "cold wallet" after lost/broken device
  • the network is extremely resilient, dies only with the last cell
  • built to support any kind of decentralized/distributed application on top

Project status

Mercury is a redesigned and advanced version of the IoP Profile Server and IoP Connect which were a step in the right direction but lacked several features from our vision, were created by developers who left the community and were hard to fix and maintain.

Please be aware that this project is still in a very early and experimental phase. We opened up the source code to give a sneak peek to developers interested in either developing Mercury itself or building distributed applications on top of it. We'd like to have feedback to learn problems in the earliest phases, priorities of missing features and your requirements we haven't thought of yet.

We think to have an initial functional implementation of the architecture. We're drafting our SDK API for distributed applications. There are still a lot of important components to be added, existing ones might be changed or redesigned, documentation is still lacking nearly everywhere.

Rough edges of the existing server and client are

  • documentation
  • profile metadata structure and protection levels

Missing important parts are

  • dApp SDK, including
    • native GUI plugins interfaces for profile, home and contact management
    • HD Secret key generation with seeds (Bip32, Bip39)
    • hiding all possible tech details to be convenient
  • Diffie-Hellman key exchange and encryption, maybe with TLS
  • hole punching support (Stun, Upnp, NatPmp, etc)
  • DHT integration (IPFS with custom IPNS, Kademlia or others)
  • profile search on distributed storage
  • language bindings
  • undelivered message persistance (e.g. missed calls)

Code structure

Directories/crates of the project are

  • home-protocol is where you should start looking to understand the architecture. It contains the basic definition of basic data structures, interfaces and network protocols plus some common utility code. Your starting point is maybe protocol/mercury.capnp describing a simple network protocol with Cap'n'Proto while lib.rs translates it to business logic in Rust. Note that the handshake modul is only for testing, we'll drop it when a Diffie-Hellman key exchange is properly implemented.
  • home-node implements a server for the protocol.
  • connect implements a client. You can use HomeClientCapnProto from protocol_capnp.rs as a proxy object to a remote home server. File lib.rs aims to provide more convenience on top but we're planning to redesign it.
  • test contain integration tests to check if our server and client implementation work together as expected.
  • storage contains experimentation on a generic storage layer using hash-based "indexing" that could use IPFS, BitTorrent, StoreJ, etc as a simple plugin. We currently use minimal code like a KeyValueStore interface and AsyncFileHandler implementation from this crate, you should ignore it for now.

Copyright © 2017-2018
Libertaria Ventures LLP, UK
Decentralized Society Foundation, PA