I think it was inevitable that these platforms would start having content restrictions once they shifted to using machine-learning algorithms tuned to maximize

I think it was inevitable that these platforms would start having content restrictions once they shifted to using machine-learning algorithms tuned to maximize each user's engagement. For a significant fraction of users, the most engaging content will be psychologically manipulative - conspiracy theories, racism, political intrigue, and the like. I see the combination of an engagement machine and such content ethically problematic, but I don't think they're going to turn off the extremely-profitable engagement machine unless forced to, so they restrict content instead. The trouble is the engagement machine is sophisticated, while the content restrictions are crude. This probably isn't sustainable in its current form, but there aren't easy answers to the problem. A sophisticated "bad" content suppression algorithm sounds pretty dystopian to me.