This is a huge increase. Were there materially large changes to the Algos that we are using to detect these? Have we gone back through GR11 data to see if the new algo would catch missed sybil in GR11?
This is pretty immense (the number of outcomes you are driving towards). There are only 12 listed in the proposal, are there another 6 missing from the proposal?
There is an interesting evolution of FDD that is unlike other workstreams. When I look at PGF, which is arguably the heart and soul of the DAO, they are modeling much of their work off of a similar ethos to what Peter Pan has talked about.
The entire DAO will be organized as sub-DAOs and ideally two-pizza teams:
PGF has three/four “sub-DAOs” (lets call them sub streams) that have less than 5 people per sub stream who are compensated. This structure lacks the same level of anti fragility, but also offers accountability on outcome with more budget efficiency. (more on this below)
I really like the council idea and commend you for trying this out. The Incentivization on showing up on time is a great way to align participant incentives to the outcomes you want (punctual attendance to run efficient meetings!) This approach seems like it would create an accounting nightmare though. The complexity in tracking and paying people feels rather burden some. This is my opinion on the cumbersome accounting, but I am interested in seeing how this works out.
As I dig into the roadmap you have linked and outlined, this summary at the top really stands out:
The FDD workstream budget was approved and began to decentralize the process by having the community take over responsibility of approving grant eligibility, paying contributors to evaluate grants and accounts flagged as sybil by the ML algorithm, and even paying the contracts to SMEs guiding the machine learning efforts
The future of sybil detection includes a “community data lake” with anonymized data allowing data scientists anywhere to compete to find the best models using an open-source feature engineering pipeline. Sybil accounts will be detected across Ethereum addresses, DIDs, and Gitcoin accounts. Eventually, meta-models will give even greater insights.
This was the dream we had for the workstream in its inception, and reading this blog post was a great and inspiring refresher of the dreams we held when launching the workstream. I am really torn on if now is the right time for us to expand to the great vision this budget is advocating for (and if we should be paying for three teams to build ML models, for anti fragile and evaluation purposes)
Frankly, It’s a moonshot. Going straight from crawl, to run… scaling immensely to hit strides, build out the data lake, scale ML teams, all while creating an operating system and operations layer to manage it all.
when is the right time for FDD to seek additional customers to validate this vision?
This vision was born based on a single team’s needs (Gitcoin’s Grants development team) and continuing to fund it without confirmation on a broader market need feels like we are gold plating a system that is for a single customer. I am interested in acting in good faith and continuing to fund as is for S13, but I would request more feedback before funding in S14. I am also happy to help as much as I can to make intros and help validate this as I love the vision!
this is another example where we are building and scaling without stopping to evaluate how to solve the problem from first principles. For example, have we considered requiring 25 DAI to be staked per new grant request to help increase the fraud tax on this process? If so, apologizes for throwing stones, I do worry that scale has been the default state for FDD (in many cases for good reason, as it reduces fragility).
When I review the budget, It looks like roughly $596K is budgeted for, but your request is asking for twice this (roughly) - why the reserves? Further, there are folks that seem to potentially be getting double paid on the ground control budget.
Here is my take as a single steward who has tried to really dig in and understand the proposal.
- The Ground Control work seems like it may be duplicative. As some of that work is being covered/compensated by DAO Ops. I would strike most of that budget if I am being prudent.
- I would also deny the reserves. FDD has been a great steward of the budgets it has received and the reserves feel unnecessary.
- GIA seems like the wrong solution to the problem, but I would leave that $75K budget intact with the hopes of us partnering with GPG to build in a staking model in GR14, or another software based hurdle (recognizing a change for GR13 is likely too aggressive).
- Sybil Defenders budget seems pretty heavy, but I am not an expert in how to cut there. Knowing BlockScience is at least half the cost here, it would be great to get the model owned by the DAO so that those funds could be re appropriated… right now it feels like we are double paying, but I am fine with that if there is an end in sight. I would advocate strongly that we do not continue to fund the Block Science requirement in S14 as we would have funded all the work necessary for the community to take this over in S13.
- I would also leave Source Council as is as I really appreciate the though that went into crafting the experiment. It seems to be a primary funding mechanism for contributions and I would love to see how it evolves and what FDD learns.
Given this, I would personally like to see a budget request closer to $555K. With $313K in reserves, an actual request of $275K (35K GTC) feels more than sufficient even with some price volatility.
This assumes that in S14, we can reduce costs as the “double payment” for BSCI and Community run models goes away, the GIA work is hopefully greatly reduced (if GPG can help build staking into grant creation or some other software deterrent to the deluge of new spammy grants), and DAO Ops continues to support and fund work that FDD relies on (DevOps, Consulting, Accounting, etc.)