I see it like this: Taking in the totality of the danger, they're right. If the source (social network) and the destination (child brain) cannot be treated as trustworthy, then you must control the content for overall safety. If you could trust either end, then you could dismiss the argument. But you cannot trust children to be cognizant of abuse, and you already know social media literally reinvented abusive behaviors for the 21st century. Do nothing and children will be harmed. Overreach by any amount and you have destroyed freedom. The only middle ground is weaker encrypted E2E comms. Something that creates a forcing function with very high cost (an electric bill or SaaS service) for the sniffer but can be broken with enough horsepower. Think about what millions of dollars per character would do. Good luck codifying that insane compromise into a law.