Cool Farcaster graph visualization from @woj.eth and @nobuilder.eth

customer support @supercast

trying to sound smart /doingnumbers

Building /frametown. Amusement, wonder, and uncertainty ✴︎ libred.org


Cool Farcaster graph visualization from @woj.eth and @rafi 

https://nobuilder.github.io/farcaster-social-graph

just 1 day in, had to beg @rafi to launch fast a shitty concept

what type of connections that are less cluttered could we implement?

On Twitter everything with an edge weight < 3 is randomness. Delete them. Delete all isolates

Let users enter their name & then build out the network in steps

Cluster ppl. show only clusters & their connections. Clicking a cluster opens it up (often used for SNA with orgs)

Graph per channel (Jaccard index correlation btw them)

That’s all what I can think of right now at 10pm after a shitty night of sleep 

For metrics check this. Is based on science & user research. Ask me if you have q

TogetherCrew.com | single parent x 4 | seeker of adventure | survivor of chaos | host /togethercrew /kbc /single-parenting
seemore.tv/free/fc/kbc

Las Palmas de Gran Canaria, Las Palmas, Spain

Graph per channel (Jaccard index correlation btw them)

That’s all what I can think of right now at 10pm after a shitty night of sleep 

For metrics check this. Is based on science & user research. Ask me if you have q

https://github.com/RnDAO/tc-Documentation/blob/master/Twitter-vibes/Twitter_Community_Vibes.md

Is your edge based on following? If yes, get interaction data (recast, reply, watch, like etc). If interactions <3 delete the follower edge. If need more info I need @jam 

Clusters: do bottom-up at the start. SLM is good for large data. You can check sociometrix (uni of Leiden) for ideas

1. entering a name def something i want to do

2. wdym by edge weight <3?

3. what cluster rules should we create? rn fc is small and all power users are all connected to each other, twitter with 1b users has much more visible clusters