Challenges

Technicals, humans and more

Goals of Berty

  • Provide a secure chat application over P2P network on desktop and mobile platforms
  • Provide an easy to use Protocol, so that developers can create secure applications over P2P networks on desktop and mobile
  • Improve the strenghts and entropy of P2P networks by adding a lot of new peers
  • Be able to do off-grid communication (offline-first)
  • Be resilient against censorship
  • Be resilient against mass surveillance by minimizing metadata leakage
  • Being an alternative to the existing models powered by GAFAM or controlled/controllable by the government

In working towards those goals, Berty is facing certain challenges.

Technical challenges

Implementing P2P networking on mobile devices

Constraints on mobile platforms:

  • Cellular connections are more problematic:
    • NAT traversal is more difficult to achieve
    • Connections are less reliable
    • Connections are generally slower with a higher ping
    • Data plans have limited daily/monthly quotas
  • Less CPU power
  • Smaller maximum amount of concurrent networking connections
  • Limited battery life
  • Smaller amount of CPU time allocated to the app (app in the background)

Some projects that implement P2P on mobile

  • Textile.io is probably the most well-known project that builds mobile applications over IPFS by running a full node on mobile devices using gomobile. They already have several apps released (including pure-demo applications) and a SDK.
  • Jami.net is doing P2P on mobile, they are focused on delivering a direct Skype competitor, including the features like audio call.
  • status.im: Status brings the power of Ethereum into the user’s pocket by combining a messenger, a crypto-wallet, and a Web3 browser.
  • matrix.org: Matrix owes its name to its ability to bridge existing platforms into a global open matrix of communication. Matrix is more of a decentralised conversation store rather than a messaging protocol.

How Berty tackles it

  • Gomobile-IPFS: a repo created by Berty Technologies that aims to provide packages for Android, iOS and React Native that allow one to run and use an IPFS node on mobile devices.
  • Berty-Protocol: we are currently designing a protocol that can be used by everyone in their projects to create secure P2P applications that run on both desktop and mobile.
  • Berty-Chat: a concrete example of a simple app that is built on top of Berty Protocol, designed with non-tech people in mind, trying to get P2P on mobile to the mass user.

Preserving Privacy and Securing Communication in a Distributed Environment

Constraints

In a centralized environment, there’s an authority that can validate the identities, signatures, encryption keys and so on. In a distributed environment, we need to find ways to achieve trust between peers without a central authority.

In a distributed system, we cannot ensure the message arrival order, especially in off-grid communication context, which can cause problems, mainly when encrypting messages with Perfect Forward Secrecy algorithms that use rotating encryption keys.

Metadata are required to operate a messaging system. However, they could be an important source of information if they were collected and analyzed.

  • Signal.org: the gold standard for any centralized environment. Berty uses many of its cryptographic algorithms/protcols, having them adapted for distributed environment. They work hard to limit their metadata footprint as much as possible, but their servers are still hosted on Google Cloud and are automatically subject to the Cloud Act.
  • Messaging Layer Security: MLS offers a more scalable alternative to Signal Protocol. This solution still requires a central server in order to work.
  • Olvid.io: Olvid claims to be the only messaging service that also encrypts metadata. They are centralized and host their servers on AWS, which also belongs to the Cloud Act.
  • status.im: Status designed Whisper to overcome some of the above shortcomings AND be configurable to the end user, who ought to be able to specify what trade-offs they are willing to make between battery life and mobile data charges, privacy and possible metadata leakage.

How Berty tackles it

  • go-ipfs-log: ipfs-log - the base layer of orbit-db - is a distributed peer-to-peer message log on IPFS using CRDT for message order consistency. This open-source library intends to provide a fully compatible port of the JavaScript version in Go.
  • go-orbit-db: orbit-db is a distributed peer-to-peer database on IPFS built on top of go-ipfs-log. The majority of its code has derived from the JavaScript’s orbit-db project. This project intends to provide a fully compatible port of the JavaScript version in Go.
  • Berty-Protocol: the protocol is using, among other things, NaCl and Signal’s Symmetric-ratchet to ensure the reliability of identities on the network and encrypt messages with Perfect Forward Secrecy. In addition to searching for a way to hide public IPs on IPFS, we are trying to reduce metadata leakage as much as possible:
    • No phone/email required for registration
    • Optional zero-knowledge replication servers
    • Rotating rendezvous points and peerIDs
    • Metadata encryption (as much as possible)
    • Noise generation

Organizational challenges

Funding as an NGO

Constraints:

  • No selling user data (we don’t want to and we can’t).
  • Free & Open Source application - no revenue.

How Berty tackles it:

  • We’re getting donations from sponsors.
  • We plan to crowdfund some tasks, like security audits.

Being incorruptible / unkillable

Constraints:

  • Government can force you to add a backdoor in your software.
  • Government can force you to give them metadata and messages. This is strictly framed by SBDU in France (anti-terrorism section) and Cloud Act in USA.
  • Government can force you to kill the project.
  • You must comply with the regulations so you can’t be sued.

How Berty plans to tackle it:

  • Being open-source, so community can freely audit the code to check if any backdoor was added to it.
  • Being transparent, not having any secrets that the community doesn’t know, so even if Berty Technologies is killed, the community can continue the project.
  • Losing control of the release process: we’re still thinking about it but the idea is to delegate the release process and validation to the community.
  • Allowing other projects to use our protocol: it will add more noise over the network, which is good to make the metadata less useful.

Didn't find what you were looking for?