You're supposed to do that kind of filtering on the endpoint. If it's possible anywhere else, then it could be used to censor other people's computers.
I'm afraid you only take into account private use.
On-endpoint filtering is not enough on a large scale. In a network with expected high level of security I don't trust the endpoint and censoring them is a feature, there is no moral or legal expectation of privacy, but data integrity would be nice.
If you own all of the endpoints, then it is indeed morally okay for you to do network-level filtering. The problem is that if we design network protocols so it's possible for you to do that, then it will also be possible for people to do network-level filtering that affects endpoints they don't own.
How do you propose that it be possible for you to do that to things like your smart TV without it being possible for you to do that to other people's computers?
On-endpoint filtering is not enough on a large scale. In a network with expected high level of security I don't trust the endpoint and censoring them is a feature, there is no moral or legal expectation of privacy, but data integrity would be nice.