The problem has, in my opinion, become opposite to what Section 230 was meant to solve. The majority of online communication platforms are, in my opinion, “over censoring” exactly as you described. If they can be trusted to over censor, why can’t they be trusted to censor to the extent required to avoid liability? It’s obvious to me that we shouldn’t be expecting any company to have 100% correct censorship against illegal content, so we shouldn’t allow them to police other content to such a great degree.