▲ | Nathan2055 7 days ago | |
This has been the advantage, and the drawback, of Signal's security model from the start. Everything on Signal (at least the "original" design from a few years ago, this has started to be adjusted with the introduction of usernames and now backups and eventually syncing) is end-to-end encrypted between users, with your original phone acting as the primary communication node doing the encryption. Any other devices like desktops and tablets that get added are replicating from the original node rather than receiving new messages straight from the network. This offers substantial privacy and security guarantees, at the cost of convenience and portability. It can be contrasted with something like iMessage, before Messages in iCloud was implemented, where every registered device is a full node that receives every new message directly, as long as they're connected at the time that it's sent. Today's addition brings Signal to where iMessage was originally: each device is backing up their own messages, but those backups aren't syncing with one another. Based on the blog post, the goal is to eventually get Signal to where iMessage is today now that Messages in iCloud is available: all of the devices sync their own message databases with a version in the cloud, which is also end-to-end encrypted with the same guarantees as the messages themselves, but which ensures that every device ends up with the same message history regardless of whether they're connected to receive all of the messages as they come in. Then, eventually, they seem to also intend to take it one step farther and allow for arbitrary sync locations for that "primary replica" outside of their own cloud storage, which is even better and goes even further than Apple's implementation does. If done well, I actually quite like the vision they're going for here. I'm still frustrated that they wouldn't just port the simple file backup feature from Android to the other platforms, even as just a stopgap until this is finished, but I think that the eventual completion of this feature as described will solve all of my major concerns with Signal's current storage implementation. |