[lordCarbonFiber]

Except, in this system you're relying on users who choose to view unrated content to provide the initial ratings. Even places like 4chan have moderation and most people won't be willing to engage with content at that level to positively rate memes or more general discussion. Add onto that that untrained users are going to be largely garbage at actually classifying your content; this is somethign social media struggles with and they have trained moderators. What's the difference between hate and ironic self ribbing, what's the difference between sexuality, art, and medicine (see tumblr's hilarious attempt at banning "femme presenting nipples"), what kind of content can be legitimately traumatizing or otherwise dangerous (calls to violence; deadly misinformation; etc).

The closest thing to that model is something like the fediverse where you're choosing to federate with instances based on their moderation rules and your users choose to stay on your instance based on yours. Even that has problems with chains of trust and discovery though. At scale you can't assume good faith anymore