Technowatermelon. Elder Millenial. Building Farcaster. 

nf.td/varun

What are the best technical resources for building algorithmic social feeds? 

Feels like a great opportunity for farcaster to open source their findings and tech alongside the client to help level up the other clients in the ecosystem (including those that don’t exist yet)

cc @v

If open sourcing makes it easy for others to implement, X open source code should make it easy for anyone to implement a world class recommendation algorithm for a text based social network. 

This obviously hasn’t happened, so maybe the question worth asking is why?

they don't have an incentive to help others build good feed engines

What specifically is wrong with their open source code, that you can’t build good feed engines with?

1. their stack is almost 2 decades old
2. it's built on a closed data source
3. the repo was dumped with a single page technical blog post, no other guides, and isn't being maintained in the open
4. i'm not building a client right now, just want to learn – that blogpost is probably a good starting point anyway

you built a pretty good algorithm from scratch on an open protocol, which is not something that is abundant in the online social landscape today – sorry for hoping that the feed algorithm would form part of this effort to open source the official protocol client

The most popular text based social network in the world has open sourced their algorithm that works exceptionally well. 

If you believe that open source algos are the answer, just copy theirs. It’s objectively better than ours in many ways.

Have you tried this? If not, why not?

i'm not sure what X's feed engine has to do with my ask? it should be obvious that seeing the official farcaster client's approach to building feeds would be 10x more relevant than X's