It's not dumb and promotes what's engaging for users individually algoritmically without categorising. That's the not dumb part about it.

>They serve you content that they think you will find engaging, and there has to be some automatic categorization as a result.

And this can be completely smart or dumb depending on how you look at it. From my perspective it's a smart algorithm still. If you like or comment on far right or far left ragebait or just ragebait in general then it can serve you ouput from accounts who attract engagement from others who also engage witht hat video. Aka similarly minded people get similarly minded content semi organically as a result of optimising for engagement.

It can do so without genuenly knowing that the video is of category xyz and promoting that specific category but automatically it detects that people who engage with video y are also more likely to engage with video z. The ideological and engagement type bubbles create themselves as a result.

>even if they don't select for rage-bait, they still refuse to take action against accounts that repeatedly post it, making your interjection largely pointless.

The definition of ragebait is vague. Your supressing ragebait is another person's ragebait or supressing of desimination of societal issues. It can be a video of violent crime or of political corruption. Of war or religious extremism. It can be sharing statistics, supporting misogynistic or xyzfobic talkingpoints or exposing them, etc

In a society facing economic decline, corruption or cultural upheaval or what have you you will be stepping on toes whichever way you go. I'd say it's better to either outright ban this kind of social media which will be deeply unpopular or to focus wholely on tackling the divisive issues that lead people to be attracted to some of the dominant topics.