Finally, someone pointed out to me another snag in encryption recently. Encryption doesn't just hide the post from malicious actors, it can also hide it from the server itself: #staff already has difficulty #moderating direct messages, as they require reporting to work, and this adds another layer. Anything that can't be decrypted by the server software would even bypass locally based automation: we couldn't even set up a bot to watch for people saying the N word in the DM system, which is the number one way it's abused.
I don't necessarily think staff should be reading people's private messages without permission, it's definitely not an ability I would want to see implemented, but I worry that encryption could preclude us from implementing more sophisticated #moderation tools. Without it, the best we could do is have some sort of javascript-based system in the viewer itself recommending the user report a post when it detects things.
Maybe we should, but it needs to be considered.