They're using a graph database and their web of trust is still naively based on follows. But they have figured out what kind of read queries are useful for a Twitter clone, and that's worth using as a reference.
Discussion
My interesting but does any of the architecture allow for unsilo’d application design that extends beyond the so-me (twitter clone ) scope ? I might have to look into it again 😄 last time it just seemed so small
Their database queries leave much to be desired. It's all Twitter clone stuff. I don't know about the silo part, but servers are a big part of their architecture.
This is incorrect. The graph is based on semantic metadata created by you and your peers.
So while I have you here, what do you think about these weightings?