Section 230 needs an update. It needs to be made clear that *hosing* speech is not a liability. "*Recommending*" speech with a black box algorithm that the user can't control or select *IS* a liability.