Isn't this article just incorrect: > Signals database, which we must assume is compromised due to its centralized and US domiciled nature, has a few important

Isn't this article just incorrect: > Signals database, which we must assume is compromised due to its centralized and US domiciled nature, has a few important pieces of data; > Message dates and times > Message senders and recipients ( via phone number identifiers ) Since Sealed Sender[0] in 2018, Signal messages only have the sender available under encryption, in other words the signal server can't store the sender. Conceivably it could store metadata about the sender, but such data isn't authenticated. The article fails to mention sealed sender at all. As a consequence of this, and the way signal bootstraps contacts from phone numbers, you can't actually build any sort of social graph from signal. The signal server doesn't know my contacts on signal. It never has. It knows phone numbers with signal, and I know phone numbers and contact names, and you can join those things to get that my contact, Fred, uses signal. But that's all done on my phone, the signal server never knows that [1]. On the other hand, any system that uses a system of contacts managed within the application (e.x. you have a username and contacts tied specifically to the communication protocol) must maintain a contact graph. It's actually worse from this perspective. I think with a system like Keybase you could attest to having an identity on the platform, and use that to bootstrap but this is no better, and arguably worse, since you still need to publish that identity on the attestation platform. [0]: https://signal.org/blog/sealed-sender/ https://signal.org/blog/sealed-sender/ [1]: Implementation detail I don't know here: whether or not you are sent all signal phone numbers, or if you reveal the contacts you know to the server. The first is obviously more private.