ICYMI @dwr.eth dropped tons of alpha on the Official Channels Day livestream 

It’s clear he recognizes channels are a valuable community building tool on Farcaster. However, he believes discovery and retention aren’t working well with the current model. 

He enthusiastically shared his thoughts on how LLMs could help surface conversations via topics which could then be used to drive engagement to likeminded people in subgroups and communities. 

He acknowledged the challenge is unlearning old models and behaviors, helping people on Farcaster feel as part of the process, and moving forward together. 

Do you think AI can help us innovate new models for engagement or are we better off with human curated experiences?

Founder /cryptosapiens, Onchain Creator House architect. Pushing creator ownership forward—media, value, and community onchain.

Maybe a mix of AI and human curation would work best (at least for the foreseeable future). AI to filter through everything and human to filter through what the AI brought up. Not sure though, just thinking out loud here

Collecting stories at Beams https://paragraph.xyz/@beams ✨ 🔦 Always playing, always learning ✌️ https://deca.art/Ashaverse

Yes, it's tricky for sure. If it's within a channel, it's easier I guess. But outside of it how can the algo tell if someone is interested in a topic in order to present that topic to them? If it's only counting on previous engagement we're in for a vicious cycle of being presented the same things over and over again, because that's what we previously interacted with.

I believe Dan’s thinking is that the human curation happens when you share a post on Farcaster, like, comment and recast. The algorithm then would pick up this conversation and present it to anyone who’s interested in that topic. The hard part I think, and it seems the Farcaster team is also considering this, is how this then becomes a community, or a destination for extended dialogue about that topic.