No, that’s just an assumption. It’s very standard. But they do, this is the code for it. https://github.com/signalapp/Signal-Android/blob/main/app/src/main/java/org/conscrypt/ConscryptSignal.java
That doesn’t confirm they send anything extra about your device, that’s an assumption as well.

If I share an IP with 100 million other Signal users

That’s already not very likely, but ignoring IP, you’re the only one with your SSL keys. As part of authentication, you are identified. All the information about your device is transmitted. Then you stop identifying yourself in future messages, but your SSL keys tie your messages together. They are discarded once the message is decrypted by the server, so your messages should in theory be anonymised in the case of a leak to a third party. That seems to be what sealed sender is designed for, but it isn’t what I’m concerned about.

daniel sent a user an image…

Right, but it’s not other users I’m scared of. Signal also has my exit node.

What you’re describing is (not) alarming (…) Signal’s security team wrote.

I mean if strangers can find my city on the secret chat app I find that quite alarming. The example isn’t that coarse, and Signal, being a centralised platform with 100% locked down strict access, they well could defend users against this.

What do you mean when you say “conversation” here?

When their keys are refreshed. I don’t know how often. I meant a conversation as people understand it, not first time contact. My quick internet search says that the maximum age for profile keys is 30 days, but I would imagine in practice it’s more often.

Even if we trust Signal, with Sealed Sender, without any sort of random delay in message delivery, a nation-state level adversary could observe inbound and outbound network activity and derive high confidence information about who’s contacting whom.

That is true, but no reason to cut Signal slack. If either party is in another country or on a VPN, then that’s a mitigating factor against monitoring the whole network. But then if Signal is sharing their data with that adversary, then the VPN or being in a different country factors has been defeated.

Here’s the blog post from 2017

I appreciate the blog post and information. I don’t trust them to only run the published server code. It’s too juicy of an honeypot.

I don’t have any comment on SGX here. It’s one of those things where there’s so many moving parts and so much secret information, and so much you have to understand and trust that it basically becomes impossible to verify or even put trust in someone who claims to have verified it. Sometimes it’s an inappropriate position, but I think it’s fine here: Signal doesn’t offer me anything, I have no reason to put so much effort into understanding what can be verified with SGX.

And thanks for the audits archive.

Okay. But this method doesn’t address that the service doesn’t need the message to include the sender to know who the sender is. The sender ('s unique device) can with 100% accuracy be appended to the message by the server after it’s received. Even if we trust them on the parts that require trust, the setup as described by the blog doesn’t do anything to prevent social graphs from being derived, since the sender is identified at the start of every conversation.

If we trust them not to store any logs (unverifiable), then this method means they can’t precisely know how long a conversation was or how many messages were exchanged. But you can still know precisely when and how many messages both participants received, there’s just a chance that they’re talking to multiple people. Though if we’re trusting them not to store logs (unverifiable), then there shouldn’t be any data to cross reference to begin with. So if we can’t trust them, then why are we trusting them not to take note of the sender?

The upside is that if the message is leaked to a third-party, there’s less info in it now. I’m ignoring the Github link, not because I don’t appreciate you finding it, but because I take the blog-post to be the mission statement for the code, and the blog doesn’t promise a system that comprehensively hides the sender’s identity. I trust their code to do what is described.

How does that work? I wasn’t able to find this. Can you find documentation or code that explains how the client can obscure where it came from?

Your data is routed through Signal servers to establish connections. Signal absolutely can does provide social graphs, message frequency, message times, message size. There’s also nothing stopping them from pushing a snooping build to one user when that user is targeted by the NSA. The specific user would need to check all updates against verified hashes. And if they’re on iOS then that’s not even an option, since the official iOS build hash already doesn’t match the repo.