You let them follow or block people. You allow them to join open or closed groups, groups which are okay with adult language or groups which encourage manners, some groups might be invite only, etc.
In most instances given those kinds of options people can manage their interactions.
You don’t have to shutdown people you don’t like on the platform, so long as they are not engaging in criminal activity.
Unfortunately it's never framed as such. It's usually of the form "this person's words/ideas are objectively bad and your platform is bad by allowing them to post at all"
Other words and ideas may not be universally bad, but I don't want to host them on my platform. My platform, my rules.
Still other words and ideas are not universally bad, and I don't have a problem with them on my platform, but some of my users will. So my users need the ability to define their own rules and control who they interact with.