I think this is an inherent problem with the kind of approach you’re suggesting (i.e. let people say what they want, so long as they’re not breaking any laws, and let debate/downvoting sort it out).

In my head, it works the opposite. If you have a list of things people can't do, they will wiggle their way around them to still be an asshole. If your rules aren't explicit then they can't be the "actually" guy.

If you just delete any posts saying hateful things, then you get marginalised people trying to participate, they get notifications saying someone has replied with something hurtful, then it gets removed and they just have to accept there are not nice people in the community. Worse, people may turn to DMs.

On the other hand, if someone says something hurtful, and 20 others reply explaining why it wasn't a nice thing to say, then they get to see that the view is not widely held but hidden, and instead most people do not hold that view.

As an example, in one of our first posts here, the OP used a derogatory term in a common colloquial way, someone replied with a link explaining why it was a problem, then the OP apologised and edited their post.

This is how I hope for things to work. In practice, many people will double down or be trolling, and we don't have to stand for that. But with my naivety I'd like there to be a chance for redemption.