It's like the web — anyone is free to create a website and host it.
There is no moderation at the WWW level. And even in the cases of DNS seizures (extremely rare; multi-jurisdictional), IP address will still work!
Post-permissionless, the raw protocol data from Hubs was never going to be "nice and clean". This is why PageRank was necessary for web search. A useful public API from Warpcast is the Power Badge: https://docs.farcaster.xyz/reference/warpcast/api#get-all-power-badge-users
Would encourage people interested in looking at Farcaster data to move to social graph-based models. Start with a core of 100 “good” users (you get to pick!), see who they follow and engage with, rinse and repeat, etc.
Long-term, the only way to measure relevance and influence of a given Farcaster account is who is following and engaging with an account, not the raw numbers. This isn’t possible on web2 social since the platforms don’t expose the entire graph via APIs. On Farcaster, the graph is public and permissionless.
Why don’t we increase cost? We started at $12 per storage unit per year in October and had a lot of good users complain it was too expensive. So we iteratively brought it down over the last few months to $3 (paid to the protocol). Warpcast charges $5 because of mobile app store fees.
Increasing cost to sign up makes it less likely that good users sign up, but doesn’t deter all spam (just some of it).
All that said, of the 326K FIDs with a paid storage unit, i.e. valid protocol users, ~300K are paid sign ups. So every DAU by definition generates at least $3 in revenue to the protocol. You can see those payments onchain: https://optimistic.etherscan.io/txsInternal?a=0x00000000fcce7f938e7ae6d3c335bd6a1a7c593d
This doesn’t happen in web2 social (since sign ups are free). Although, that’s changing: https://twitter.com/elonmusk/status/1779930065469383166
Finally, just because an FID has a lot of followers and raw engagement numbers on the protocol, doesn’t mean the average user on clients is seeing it. Warpcast and other clients are going to continue to refine quality filtering (in the same way search engine indexes and email spam filters evolved).
Over 700 subscribers
How to think about Farcaster protocol data
Published on Paragraph https://dwr.email/how-to-think-about-farcaster-protocol-data
Bluesky has some interesting moderation systems - I hope Warpcast can borrow from without the complexity
Probably doesn’t scale. Each client should have its own moderation system. Don’t like one, use a different client.
i love this & @dwr.eth nails it as to the possibilities uniquely enabled by onchain we're launching smth later this week along the lines of "page rank for farcaster users" based on this data...it's looking really useful so far in our POCs but users will be the judge, and then we'll learn from it, tweak it, and oss it
It’s like the web, I love how this starts
You and @v are really good explainers
I like to think of my FID activity as an API. La creme de la creme.
I really loved the last part, "SE indexers & email sp...". What I understand from this paradigm is that there will be an ongoing plan of refine & improvement (as there is in SEs). This openness has really opened a massive opportunity of research on social networks that were not accessible to the public before.
adjacent topic but 1) how should we think about the rights to own our data (casts, reactions) and 2) can anyone use Farcaster protocol data for things like training LLMs as the dataset grows?
1) how does that work on ethereum? 2) it's a permissionless public protocol that no one is forcing anyone else to use :)
I was hoping this was the case but didn’t want to bias your answer 😉 as we grow to 1 billion DAU Farcaster will become one of the largest organic open datasets in the world and our conversations will help shape AGI who wouldn’t want to be part of that?