Comment by walterbell | Hacker News Reader

Comment by walterbell | Hacker News Reader

Parent Full thread

> I wish we had a messaging app with a market as large as WhatsApp's, UX as good as Telegram, security as good as Signal, run by an organization like Mozilla.

An open protocol is a prerequisite. There's a 2018 IETF proposal for interoperable E2E messaging, initiated by Cisco, Google, Facebook and Wire:

Architecture: https://datatracker.ietf.org/doc/draft-omara-mls-architectur...

Protocol: https://datatracker.ietf.org/doc/draft-barnes-mls-protocol/?...

"Messaging Layer Security (MLS) ... is not intended as a full instant messaging protocol but rather is intended to be embedded in a concrete protocol such as XMPP [RFC3920]. In addition, it does not specify a complete wire encoding, but rather a set of abstract data structures which can then be mapped onto a variety of concrete encodings, such as TLS [I-D.ietf-tls-tls13], CBOR [RFC7049], and JSON [RFC7159]. Implementations which adopt compatible encodings should be able to have some degree of interoperability at the message level, though they may have incompatible identity/authentication infrastructures."

It's not a good plan. They propose to design, de novo, in an open standards group, a cryptographic secure group messaging protocol based on a design that has never been deployed at any scale. The underlying design is cool --- it's basically Asynchronous Ratcheting Trees, which was presented in a paper at RWC just a few weeks ago. That's how de novo this particular "standards" effort is: the underlying theory is just 28 days old.

Already, in the starting point draft, it's been crudded up: it has "ciphersuites", and comes with support for the NIST P-256 curve --- despite the fact that the underlying design wants to take byte strings to curve points, which is tricky to do on the P-curve. It will only get worse from here. They'll figure out some reason to bolt a PAKE onto it soon enough.

Signal Protocol is exceedingly well-documented (and even before those documents were written, it was open enough for Wire to lift the protocol wholesale).

The IETF is bad at cryptography. Your default position should be distrust of IETF crypto standards.

Can Signal contribute to the IETF design, discussion and documentation?

Since the underlying standard is really a political maneuver meant to thwart Signal, this seems unlikely.

(I like my source on this, but can't share it; I expect to be able to this year, though. In the meantime:

0e49002152a374d9c11251cf856a7ccf25ef9bd0db54c3e97bef2a4109dad4f0)

Isn't Facebook (Messenger and Whatsapp) already using Signal Protocol at scale?

Yes.

Why would Facebook by trying to thwart the open Signal protocol that they are already using? In a standards discussion like IETF, wouldn't they be more likely to specify Signal interoperability (i.e. Facebook Messenger and WhatsApp interoperability) as a requirement on any proposed standard? That would be good for Signal.

1. Facebook is a big place.

2. Facebook's interests in the standard are not necessarily what drives the standard; Facebook can just be along for the ride.

I think what's going to happen with this is what happens to all de novo IETF designs other than TLS, the one that the market requires actually work: it's going to fail. That's also the outcome that I'm hoping for.

In the meantime, if Millican wants to get together with Katriel Cohn-Gordon and do a Facebook-only ART protocol design for Facebook Messenger, with an eye towards replacing Signal Protocol in WhatsApp, that would be a great development. So would a Cisco-only ART messenger, or maybe even a Mozilla ART messenger (though Mozilla's motives are the ones I trust least here).

What does not make sense is for a protocol whose service model we barely understand even in theory to be designed from scratch in an open standards group. The IETF motto used to be "loose consensus and working code". Now it's "take an RWC paper, add the P-curves to it, and use it to fuck over the most successful secure messaging protocol".

We shouldn't be cheerleading this. It's capture, not progress.

Reply on news.ycombinator.com