censor at the client all you want. just let the protocol be permissionless, both in writing and reading (many clients). e.g. email: google can kick me off gmail but no one can kick me off of SMTP (this is actually a bad example because spam prevention wars centralized email too)
Okay I don’t know the details (and I am not technical) but is SMTP hosted on their own servers/cloud? Otherwise they are also at their mercy of censorship via Google cloud, AWS, cloudflare etc.
I second that. Censor/filter at the client level. Let people choose what they want content they want to consume. If course if it's something illegal then the client can be liable for allowing that thru, but otherwise empower people with more choices