We can't create a well defined or objective method of choosing, but we also don't need to. Like most things, it's a judgement call. It's reasonable to add a lab

We can't create a well defined or objective method of choosing, but we also don't need to. Like most things, it's a judgement call. It's reasonable to add a label for basically anything if you know people are affected by it. It's reasonable to omit a label for something you don't think will affect people. It's reasonable to criticize the absence of obvious labels in obvious contexts, and not reasonable to criticize the absence of non-obvious labels in non-obvious contexts (but it's reasonable to request them politely!). We can't define any of that, but it would be a fallacy to use that as an excuse to attack the concept of content warnings.