It's the dreaded "more of the same shit" algorithm that people confuse with deep learning and AI. The sad reality is that recommendation teams in these companies maintain some ugly gobbled together if .. else logic rather than a proper recommendation engine. Proper recommendation engines are hard. That's why recommendations change dramatically if you just use a vpn or travel. The IP address based geo location trumps everything else they might know about you. That's because there's a big if else around whatever other logic they have.

Add promoted content/ads to the mix that they are paid to promote and you end up with a mix of content that is primarily based on where you are, what they want to pimp in that region, and more of the same shit you were clicking on anyway; more or less in that order. The exact same shit quite often actually. Even filtering that out seems hard.

They are incentivized by ad clicks, not by your amusement. That's the business mistake that allows Tiktok to prosper because they are still trying to grow at the expense of their competition which means that they are in fact incentivized by your amusement as observed through your addictive behavior in the form of content clicks.

Instagram was like that once. And then Facebook bought them and started messing it up. Facebook was once like that until Facebook became an advertising company. The mistake that gets made over and over again in this space is that at some point they think they've won and start milking their network for revenue. The process of doing that destroys and erodes that network and something else pops up that is more interesting for users.

Tik Tok is doomed to go there too. Just a matter of time. It's irresistible for shareholders and advertisers. They'll get dollar signs in their eyes and they'll mess it up.