This is just one of the contradictions of having a global medium/platform of communication being controlled by a single operator.
It is extremely difficult to create global policies to moderate the conversations of 3 billion people across different languages and cultures. No team, no document, is qualified for such a task, because so much is dependent on context.
The approach to moderation taken by federated social media like Mastodon makes a lot more sense. Communities moderate themselves, based on their own codes of conduct. In smaller servers, a strict code of conduct may not even be necessary - moderation decisions can be based on a combination of consensus and common sense (just like in real life social groups and social interactions). And there is no question of censorship, since their moderation actions don't apply to the whole network.