[Tomte]

No, my answer is to have a community where you ban those people without discussing the semantics of some sentence.

In the end, it's probably equivalent to your answer. Because when the people in your organization feel someone is harassing people and you decide to kick him out, you're probably simply describing his behaviour and referencing your CoC, but not doing some detailed analysis how and why he did so.

[lillian-lemmer]

Yeah, it's just that it is sad that it is practically required to be spelled out. But even bigger yet, is that we have to find ways to enforce it and its purpose.

[vertex-four]

As somebody who has experience moderating groups without explicit rules - the result of this in a group of more than a dozen people is inevitably that some of them will feel that (a) the banning is unfair, no matter how egregious the behaviour, and (b) there should be a set of written rules for the leaders/moderators to follow, so that people understand clearly what is expected of them. You really can't win.

Personally, I tend to steer on the side of a broadly defined CoC backed up with a very clear dispute resolution policy (there's plenty of examples of both of these - Debian has a pretty decent set for large projects). In a functioning community, reaching for this should only be required very rarely, so they shouldn't affect anyone most of the time - and they quite obviously don't in the case of open-source projects, or we'd hear about actual cases where people have been kicked out of projects for very little a lot more than we do.