[ynx]

> Don’t you think they would, for example, be able to prioritize content based on reputation/hidden metrics?

Like rank/derank the feed based on contextual clues? No way this wasn't the first thing done.

> Don’t you think, for example, that a fair amount of videos categories can be auto filtered with extremely high accuracy?

They've touted high accuracy in "using AI" for content moderation, like 95% or 99% or something. i.e. using machine learning to automatically tag content.

> Don’t you think, for example, that if they had enough content moderators they would have been able to stop the stream and to take down this video faster?

So your proposal boils down to "hire more people so more content can be filtered faster". GP perhaps didn't steelman your argument for you but if your proposal were to be applied the only real delta would be workforce size. It follows that your proposal in effect reduces to "hire more people".