I find the “where do you stop” argument to be riddled with holes. Laws are essentially written to explicitly outline boundries and moderation policies are basically just internal laws. Like Canadian law has very specific laws regarding what constitutes hate speech, here is what that looks like.

First you outline protected grounds. In Canada this is race, national or ethnic origin, colour, religion, age, sex, sexual orientation, gender identity or expression, marital status, family status, genetic characteristics, disability and conviction for an offence for which a pardon has been granted. (note: pardons are only available via democratic votes or through appeals in Canada)

There’s a stage where you determine what context stuff is in. Like whether it is being performed publicly or privately but marketing a video game is definitely publicly so in tgis context we can skip to it’s last part where you explicitly define hate speech. Hate speech is rhetoric that :

• Describes group members as animals, subhuman or genetically inferior

• Suggests group members are behind a conspiracy to gain control by plotting to destroy western civilization

• Denying, minimizing or celebrating past persecution or tragedies that happened to group members

• Labelling group members as child abusers, pedophiles or criminals who prey on children Blaming group members for problems like crime and disease

• Calling group members liars, cheats, criminals or any other term meant to provoke a strong reaction including usage of known slurs in the context of intended harm to group members.

These rules likely wouldn’t touch some hateful rhetoric that sneaks through under the wire disguised in very abstracted metaphor but it creates a pretty distinct pass fail bar that would catch explicit hate speech on their platform.