Self-destructing messages can prevent this, and hostility towards 3rd party apps help in that case since you can be more certain that nobody is using some shoddy implementation that ignores self-destruction or improperly deletes things.

Helps you with local cops for sure. But disappearing messages are also just a false sense of security IMO, there's nothing technically stopping someone from using a modified client like that, in fact some do exist and generally work despite the hostility, and so do screenshots...

If you're somewhere in between like a cartel or terrorist organization, please stay off any app I use to send memes to friends.

I mean yeah, but I don't think this is realistic. If you offer people bulletproof un-censor-able security they're going to take you up on it, even if you don't like them. But signal isn't that

Signal like every mainstream service has some amount of control and uses it to crack down on things like spam. They likely will use that control to censor other things too in the long term. To me that's a bad thing. If it were federated, that power and responsibility would be with the instance/homeserver, not with one centralized organization.

The beauty of E2EE is that you don't need to trust the servers at all, once you verify that you're actually connected to the person you intend to be.

This ties back to my point about metadata. There are plenty of reasons to want to trust the server, and with signal, you can't.

I do agree though, feds doing targeted surveillance have easier ways. The issue is more one of bulk collection, and principle.

And frankly the whole argument about open source safety goes out the window when the source and distribution is centralized, development is done behind closed doors (not sure to what extent this is true of signal clients but it was true of the server), and updates are automatically pushed out.

There are big advantages to the linux-distro-with-maintainers model in that regard, as those are well-versed people who track development and act as a filter between users and a malicious update.