Basically everything I've got that is relevant is about information propagation in networks... It'd probably be easier for you to build an agent based simulation to model this out than to dive into the literature on it, even if you have to learn to code.
Finding a niche should be a natural byproduct of getting value from the system.
But because Twetch is a negative sum game for money, most people will spend more than they earn here. Hence why it's better to focus on value gained from interactions.
For a user to be established, they have to find their niche. They don't have to be interesting or clever, because when they find their niche, they'll be able to produce signal with others who perceive their actions as signal rather than noise.
Signal =/= echo chamber unless that's exactly what you're looking for.
Regarding me having a following, sure, but if my way were the one taken, the provided feeds would be much easier to navigate to find something you find valuable.
In terms of money, it's negative sum because there are fees involved.
But I get the sense that you're attaching far more to the concepts of positive, zero, or negative sum games than actually exists within those concepts.
Look at the quality of your interactions, and improve them through incentives... Then, your signal:noise ratio will improve, and you'll shift the cost/benefit balance of the noise makers, too.
Make a little noise, don't overdo it, be a better troll than the trolls who come your way (and only use those skills on the trolls), and make a couple of valuable points along the way.
I think just a curated feed would do the trick. Follow certain people by topics that they post about (using incentivized tags to determine this,) and individually set rules for posts from those you don't follow.