Content
@
0 reply
0 recast
0 reaction
Stephan
@stephancill
just put up this prototype of what i call subjective farcaster: a collection of cached endpoints which can be used to calculate the number of likes on a cast or the number of followers of a user filtered based on 2 degrees of separation from the viewer's social graph i.e. only counting interactions from the viewer's follows or the follows of the viewer's follows would like to turn this into a cast action but it's really slow https://github.com/stephancill/subjective-farcaster hosting an instance of the api at https://subjective-farcaster.steer.fun
35 replies
6 recasts
33 reactions
artlu is on lunchbreak!
@artlu
@stephancill I would like to eventually use your cached endpoint for reactions. Today I built channel feeds, with a toggle to show only users I'm following. Imajin scrolling a noisy channel and only seeing ur frens. Then flipping the toggle and seeing everybody. And a toggle for owner-curated feed ("Main Feed") 1/2
2 replies
0 recast
1 reaction
Stephan
@stephancill
Very nice! How do you see feed generation working with this an endpoint like this? Get a feed from neynar and apply filtering retroactively or something more advanced? Not sure if I will end up hosting this as a service but the code will always be open source. Happy to discuss in dm https://github.com/stephancill/subjective-farcaster
1 reply
0 recast
2 reactions
artlu is on lunchbreak!
@artlu
pretty agnostic tbh. Closer to the user for immediate client-side filtering; calcs on server for cost+API usage reasons. Right now, I get a quasi-Neynar feed (via Pinata), and apply filtering on top using the fast far.quest API endpoints. State lives in React Query, not on the server. Holy grail is to identify+elevate casts that a large percentage of my personal graph will eventually like+recast+reply. This probably means identifying skilled tastemakers *that I like* and fast-following their engagement
0 reply
0 recast
1 reaction
