On Software Engineering:
- I built the cGrants trustbonus system in 2020-2021. I built it on a shoestring budget nights and weekends based off of this paper.
- The GPC has built Gitcoin Passport in 2022 on a budget of roughly $60k/mo.
With a $400k-$1m/quarter budget at times in 2021 FDD was not able to successfully make any changes to the cGrants system or develop Gitcoin Passport - though it was successful in building data science tools on top of cGrants and a grants review system on top of cGrants + SAAS tools like notion.
Yes, the system is now run by the GitcoinDAO, but that has been enabled a lot by the Gitcoin Foundation and GPC. The FDD had a hand in the grants review & sybil data science pipeline being run by the DAO.
Because of the above, I fail to see where you have a credible claim that FDD can do product oriented software engineering.
On Data Science: It’s great that you’ve got people with PhDs on the FDD, but we should really care about results first & foremost. There have been some successes with data science in the FDD, but there have also been some failures within data science at FDD (which have been compounded by communication issues). Some things I’d like to see
- three cross validation of the models or otherwise providing the validity of the models) ( @kevin.olsen or other engineers might have more details here)
- end to end crypto economic mechanisms to find & catch sybil attackers
- if not end to end systems building, then partnering deeply with the GPC to inform how they build their products.
- extending out of sybil resistence + analysis of collusion attacks in the system.
- solving the communication issues
Sure, Proof of Humanity has built a fairly 13k sybil resistent DAO with 1-3 person dev team before launch. Their dev team is now a bit larger. The cost of forgery in Proof of Humanity is probably about $100.
As has BrightID. Also on a fairly small budget. And they have decentralized their network analysis using their new aura tool. The cost of forgery in BrightID is probably about $10.
Idena is also fairly interesting, though I dont think their method could scale.
Worldcoin has a very sybil resistent registry based on biometrics that is in the hundreds of thousands.
Neither is perfect, but both POH/BrightID gotten closer to solving sybil resistence at scale than almost anyone else (GitcoinDAO included).
None of these other DAOs have solved sybil resistance with large and complicated centralized fraud prevention/data science teams bolted onto software someone else develops. They have solved it with elegant crypto economic mechanisms integrated with bio/digital mechanisms .
I find it inappropriate to deflect constructive criticism in this way - it causes you to lose credibility because it is in direct opposition to your own post about finding & engaging with honest critics.
You asked me for feedback on the post on DMs, which I shared privately. Then I asked you if I should share it publicly, and I did. When the feedback wasnt what you wanted to hear, you accuse me of just being in a bad mood.
I will not be commenting on this post anymore and will instead leave it to others in the DAO who have graciously volunteered their own skills, time, & emotional energy to help FDD retool & steer in a new direction.