papers-please

Been looking at this for techniques to predict "break out" casts

Been looking at this for techniques to predict "break out" casts 

https://ai.meta.com/research/publications/detecting-large-reshare-cascades-in-social-networks/

Technowatermelon. Elder Millenial. Building Farcaster. 

nf.td/varun

This is the alt account of Sister Hazel. Writings at https://powerlaw.systems. Building https://testnet.uvio.network. 

> only because some click farms in some third world country generate attention

Algo is more sophisticated than that. Not all engagement is ranked equally. Think PageRank

When you guys focus on surfacing what most people like, then you are 

A) doing the same as Twitter (why even bother to do the same?)
B) promoting the most average emotional reaction (how is this useful?)

There is magic on Farcaster. It is not a magical experience for me to see shitposts about shitcoins only because some click farms in some third world country generate attention.

The magic that I value as a real user is to create a connection with an individual, which happened to me more on Farcaster than on Twitter. For me it is possible to meet somebody with the same interest and objective here on Farcaster, and that interest is often rather niche and thoughtful. I think you should lean into that notion more.

I don't know how you measure success, but it shouldn't be recasts. It should be human connections made. You are probably more likely to find signal in sentiment analysis of DCs than whatever feed amplification is happening on the public timeline. Because this is all noise.