There's a group of people on X called "this part of twitter" and almost 400 of them uploaded their data to a "community archive," for public experiments like this one.
We wanted to see what would happen if we applied UpTrust's algorithm to interactions there. It's not the same as being on UpTrust bc it's passive, whereas our interaction paradigm is enactive and ongoing, making it more and more rewarding to be prosocial and trustable and bridgey, but we figured we'd find some cool stuff anyway.
And we did! The tldr in the short video; in general is that:
- different topics make a big difference
- viewer relative feeds make a difference
- the people most argued with are often the most trusted
- follower count sometimes overlaps with trust and sometimes doesnt, and in this community it's very topic/group specific
- trust reveals different clusters with different preferences
full writeup here: https://uptrusthq.com/writings/tpot-trust/