You know what, in my head I think I want a whole new messenger.
There's an indexer that acts as a phone book, but at the same time, people can bypass that by directly adding contacts.
All chat history and groups are peer 2 peer and are stored like torrents with the extended backup being self-hostable.
Recent chat history (up to 30 days) can be stored on the indexer, though they're encrypted and so the server is blind to what's in them. They should explicitly be opt-in.
Whenever a user adds a new client (device), all conversations recipients should have to approve in order for them to see the chat history.
It should also have all the bells and whistles, like emoji, stickers, groups, channels, etc.
I have been thinking of something like this too, the thing in common between us is that neither of us has the competency, the time and the persistence to make this happen.
Sometimes putting the ideas we have out there makes a difference. While we lack the competency, perhaps someone that sees this will and it will inspire them to bring something to life.
Whenever a user adds a new client (device), all conversations recipients should have to approve in order for them to see the chat history.
Why though? In case of a public chat or a chat with at least few dozens of users it'll already be excessive if it could work at all.
All chat history and groups are peer 2 peer
Like really P2P or E2E? Because I know at least one chat app that is serverless but doesn't involve E2E apparently - tox. E2E is an overkill for big group chats because it means you have to re-encrypt every message for every new user for them to see it. Else if you rely on just a fixed shared key it's not E2E anymore (which will make some people sad and hate your app).
For public chats, you wouldn't need to approve, only for private chat groups.
I get that but it kind of defeats the purpose. If your group is so small that it's worth it for every member to approve new ones then it probably doesn't produce enough content for each new member to care about.