🚨 New case study drop: We’ve been working on a Sybil detection algorithm for Farcaster, with real-world use cases in @optimismgrants Airdrops and Citizen House governance.
Here’s what we found—and why @openrank might be your best friend. 🧵👇
🔗

Account of the Optimism Grants Council.
Independent from Optimism. Does not administer RPGF. 
app.charmverse.io/op-grants/

🚨 New case study drop: We’ve been working on a Sybil detection algorithm for Farcaster, with real-world use cases in @optimismgrants Airdrops and Citizen House governance.
Here’s what we found—and why @openrank might be your best friend. 🧵👇
🔗 https://govgraph.fyi/blog/sybilrank-for-identifying-sybils-in-farcaster-case-study

Social Entrepreneur, Strategist, Researcher and Org Innovator

https://maxsemenchuk.com/

🧠 Why this matters:
Sybil accounts threaten both airdrop fairness and governance integrity.
We analyzed on-chain activity across 50k+ addresses in OP Airdrop 5.
Goal: Detect real humans in a network that’s open by design.

🔍 We used data like:
	•	POAPs
	•	Attestations (e.g., citizens, GitHub-linked IDs)
	•	Farcaster social graph
	•	SAFE multisigs
	•	OpenRank scores

Turns out: 76% of Airdrop 5 addresses showed some meaningful activity 👀

🤖 Are these perfect predictors?
Nope. But they correlate with airdrop rewards:
	•	Attestations: r=0.21
	•	OpenRank: r=0.19
	•	POAPs: r=0.1

These can act as “health checks” for any new drop or governance expansion.

🧪 We built tools for the community to explore the data:
🔬 Test addresses: https://farcaster-sybil-rank-algo.vercel.app/
🛠️ API access: https://farcaster-sybil-rank-algo.vercel.app/api-docs
👨‍💻 GitHub: https://github.com/GeneralMagicio/farcaster-sybil-rank-algo

💡 Interesting finds:
	•	One address had 3059 POAPs 😳
	•	Another had 835 attestations
	•	Only 11 Citizens & 10 Delegates in the entire Airdrop 5 sample

📌 What’s next?
	•	OpenRank + attestations = good start
	•	Exploring @holonym & other Proof-of-Human tools
	•	Possibly extending to multi-chain activities
	•	Would love feedback from Farcaster & OP folks

🗣️ If you’re involved in airdrops, RetroPGF, or just curious about Sybil detection in decentralized social…
We’d love your thoughts & use cases.
Let’s chat.

And thanks for the partner & supporters @generalmagic @emthemaker @griff.eth @openrank @linda @krati @meriembarhoumi.eth

📊 Key Insight:
@OpenRank was the strongest signal for identifying humans.
Why? It’s based on follower-following relations—harder to fake at scale.
Also had the widest coverage (57%) vs. attestations (~43%) or POAPs (~35%).