In a form like this, if someone is insulted, it is just the idea and words that have been insulted.
When using your last name, it is the real person's identity that has been insulted. Then it goes both ways in a feedback loop involving two real people's real identity without the constraint that face to face confrontation would impose.
The only way to make that worse then would be to have ML algorithms running on top trying to nudge people to but heads for engagement.
Maybe we could design a system that is worse that in order to join you have tell someone using both real names that their newborn baby is ugly and instead of collecting a list of friends you collect a list of enemies. Short of that though we seemed to have really done a great job figuring out the worst possible form of communication.
