Encryption and Globalization 15 Years Later: E2EE and the "Going Dark" Debate | Dark Hacker News

Encryption and Globalization 15 Years Later: E2EE and the "Going Dark" Debate | Dark Hacker News

aliasxneo 3 hours ago |

I've been working on a project for the last six months that brings mTLS to the masses. Meaning, post-quantum end-to-end encryption with human readable identities. It uses a decentralized root of trust for, basically, a lot of the history this article covers.

iamnothere 3 hours ago | |

Human readable identities are the hardest part (unless you just mean BIP-style word chains). What’s your approach look like? How do you prevent collisions?

aliasxneo 3 hours ago | | |

You register a DNS-style name. Registrations lasts a year with a 3 month post-expiration window to reregister, otherwise it gets released back. The system goes out of its way to make name trading uneconomical - specifically to work against speculators and name-sitters from coming in and ruining things.

mspecter 2 hours ago |

Coauthor here, somewhat surprised to see this on HN. Any thoughts, questions, or feedback welcome.

iamnothere 1 hour ago | |

Thanks for putting this together. Found via Schneier’s blog; I thought it was a good summary, and perhaps needed information in the face of attacks like Chat Control.

mspecter 45 minutes ago | | |

Oh, I didn't realize it'd made it to Schneier's blog, thanks for letting me know!

A goal of the work was to try to explain to a new audience (specifically nontechnical legal folks) the ongoing challenges with Going Dark. Hopefully it helps!

iririririr 24 minutes ago | |

The paper says ghost protocol has never been implemented, but wasn't it widely used on trump administration version of the signal-custom-client, which sent messages to a ghost account responsible for indexing and backup of plain text copies?

mvdwoord 3 hours ago |

“Everything is end-to-end encrypted… depending on how you define the ends.”

g-b-r 3 hours ago | |

It is not great that they use the E2EE term to describe TLS, although it's for a good cause

iamnothere 3 hours ago | | |

Ah, ok. Is there a fallback for using bare keys instead of human readable names? If not, you have just recreated a central gatekeeper problem; you could be ordered to terminate name service for someone. Unless it’s using some decentralized crypto system like eth.

Optional human readable names could be a good system combining convenience with resilience.

aliasxneo 3 hours ago | | |

It's using a decentralized backend. No, it's not using any existing blockchain, nor does it have any tradable token. It does have some blockchain properties, but the main layer is built on top of libp2p.

Agreed it's a very hard problem. We're in the process of forming a syndicate of like-minded parties who are interested in seeing this come to life because the model itself has no exit plan and is basically anathema to traditional capital funding.

Also, to be clear, names are hashed/salted in the backend with Argon2ID and a rotating salt. So the name index itself isn't enumerable, but it doesn't necessarily guarantee names are private once they are shared publicly. Theoretically you could lookup someone based on their public key without knowing their name - so in that sense it's somewhat close to what you're describing.

iamnothere 2 hours ago | | |

Interesting. Yes, the possibility of direct key lookup would solve the problem, especially if existing contacts remain connected even after name expiration/revocation.

Perhaps you could do a name presale, although I’m not sure that would be enough funding. I look forward to hearing more about it.

saltcured 1 hour ago | | |

I wonder if something like this could follow the spirit of "tag" URIs where you include a temporal component. A semantics of "namespace Foo as established time T."

Any usage of the name in identifiers includes the epoch. Identifiers from several epochs could be concurrently in use. Identifiers rooted in different name epochs are not be comparable.

Out of band networking effects need to establish the meaning of a specific name epoch. The disambiguation doesn't work if the intended consumers are unwilling to learn and distinguish the epoch component of identifiers they care about.

I suppose, at the limit, this is the same as just having everyone add some numeric suffix to names they try to register? But it would be more constrained, e.g. structurally enforce that the digits correspond to some kind of clock metadata which is a witnessed and signed part of the decentralized ledger.