This document is meant to describe the sybil resistance investments being made for S18/S19/S20, and to begin to describe what a minimum viable anti-sybil future looks like.
This document was prepared by:
The Gitcoin anti sybil ecosystem has many different moving parts.
- The transition from cGrants => Grants Stack/Allo/Passport
- The transition from FDD => ODC.
- The cluster mapping project by Joel Miller.
- The YOLO data analysis that Umar does to support Connor in anti sybil work.
As it stands right now, there is no group that is formally responsible & enabled via budget to deliver solid anti sybil results for the program (especially GG18). Passport is focused on their product. PGF works on fundraising & round ops. Our aim is to build the glue between these groups and enable anti-sybil to be better every round than it was the last round.
Anti sybil at Gitcoin is an ongoing evolutionary game between the anti-fraud folks (blue team) and the sybil/collusion attackers (the red team). We aim to empower a diverse & decentralized ecosystem of blue teams to stay ahead of the red team’s evolving tactics.
Right now, the anti sybil pipeline is pretty barebones. Umar has a series of scripts & data sources that he uses to put together an analysis that he delivers to Connor/PGF.
Our hope is to enable Umar to contribute his data sources, scripts, and methodology into a resource known as RegenData.xyz (formerly Grants-ETL)… which any data scientist has easy access to. By enabling this community of data scientists to have access to all the data they need without writing a bunch of custom ETL scripts + also have access to Umar’s tools very easily, we will empower these data scientists to do what they are good at (finding sybils) and not what they’re not (cleaning data, writing data gathering scripts). This ecosystem of data scientists will be able to do their own sybil/collusion analysis and contribute back to RegenData.xyz, ultimately creating a plurality of approaches & a community of practice around anti sybil/collusion at Gitcoin.
Instead of relying on personal relationships between Umar + Connor, this community of practice will have a diverse tapestry of relationships.
Further, we will be developing this community with community/ecosystem funding to draw in and retain top data scientists.
All in all, this initiative is designed to help the blue team systematically stay ahead of the red team by accelerating the Blue team OODA Loop end-to-end:
What are our goals for the future?
- good sybil resistance for GG18
- get Umar reports on grants-etl
- Good sybil resistance for GG19
- Add in the ability to update these passport scoring algorithms with blocklist + data that we find.
- give all of this to ODC & other groups (publish via OCEAN, etc)
- Co-develop community driven lists (Humans/Good actors, Sybil / Bad actor, Attacker(s), Suspicious, etc)
- Fund Red Team Work Through a QF Round (come take the matching pool if you tell us how you do it)
- Invite outside community contributions + support them with funding.
- Integrate Joel Millers cluster mapping algorithm.
- Nurture a community of blue-team contributors.
- Introduce more governance to the process.
- Make sure Grants-ETL has the latest data + reports in it.
- Iterate + Revise the ecosystem as we learn.
If you are interested in getting access to this toolset / community, please join the RegenData.xyz community on telegram.