[Cancelled/Superceded S14 Proposal] FDD Season 14 Budget Request

I really appreciate the thoughts from @annika and agree with so many points she made:
1 - Being a cost center is not a bad thing, and shouldn’t be viewed as a criticism
2 - It’s been difficult to navigate the story line on the value being delivered
3 - Sybil defense is important work and needs to be part of any QF round

For my own edification I have tried to break down the core work FDD does
1 - Primary scope:
1.1 - Defend Grants Rounds from Sybil Attacks
1.2 - Review Grants to confirm compliance to our rules/policies
1.3 - Defend against collusion attacks

2 - Secondary scope:
2.1 - Generate ML models to run programmatic sybil detection (using the centralized data set, and a growing on-chain data set)
2.2 - Conduct human review of potential Sybils
2.3 - Build/invest in tooling to make 1.1 and 1.2 more scalable and approachable
2.4 - Invest in OS and working norms (and operations) for the workstream

Is this accurate? I am trying to keep this as “pure” of a scope as possible.

2 Likes

I do not view it as a criticism, but as an inaccurate description or at best, part of the story.

We thought so too, so we added a very small budget to be intentional about this during s14.

This is not only for today, but for tomorrow. The ASOP could run without innovation, but it would cease to be effective. We HAVE to innovate every round to keep up with the red team and larger rounds.

Looking at this from a grants 2.0 perspective, we need to give communities who use or fork grants 2.0 a way to be credibly neutral.

This is a wildly subjective area

I don’t think the grants 2.0 architecture will solve this through “policy” but rather by giving communities tools to provide legitimacy (sybil defense) and credible neutrality (community consent). These communities would then compete for mechanisms that truly work better for their participants. Basically, our longer term strategies for sybil and grant eligibility have a by product of collusion protection, but this has not been a focus in the last few rounds.

For grants 2.0, users will have a dPoPP. By integrating many services, they will be able to participate meaningfully. By only allowing them to advance by integration, we would be putting new users at a disadvantage in the system. How might we provide the new and disadvantaged users the same access to the system. (That is the point afterall)

Our job is to find a way to decentralized the inputs, remove PII from the equation, and still be able to make sense of the rich history of participation users will acquire on dPoPP.

This is done so algorithmic bias does not co-opt our system. We don’t have a steady true state for what is sybil.

If we had zero ethical concerns about the bias of our algorithms, we could remove this, but I think it is best thought of as part of 1.1.

And reduce unwanted bias. Additionally, I think providing a solution which includes legitmacy and credible neutrality to any ecosystem who starts a round is a part of grants 2.0, not just an accessory.

If we hand them the 2.0 rails without this protection, they may use 2.0 and get worse results than legacy systems. The reason a community would want to use grants 2.0 is to allocate better.

It’s like selling a car without tires. Sure, you can do that, but good luck!

Yes. Can we make sure that the machine building the machines of legitimacy and credible neutrality are sustainable without key man issues.

Lastly, the desire to define innovation as secondary for FDD is not realistic. Each round the attackers advance their strategies and we need to innovate ours to be effictive.

2 Likes

I’ve gotten feedback about people not understanding what is going on in the workstream. It seems that there are many impressions about what the stream is, or should be, but they have missed the ongoing effort due to our having too much content to review.

“I feel somewhat DDos attacked when I read FDD budget”

Here are a few ways to look at it.

First, check out this explainer from Octopus octopus Thoughts On FDD - YouTube

Next, you can see how evolution is more about understanding what we did the last round, finding what is and isn’t working, and then converting to a better overall process. It is not redundant with DAOops.

Here we show which squads focus on executing this round vs innovating and if they are about efficiency or decentralization.

This shows our essential intent agreed upon by CSDO and how each FDD squad fits.

Then we connect which squads are related to which aspects of GitcoinDAO.

7 Likes

This is a step in the right direction - I really appreciate the bucketing - but I still feel a little DDoS attacked (to steal anon’s on-point wording) even when looking at this view.

I said this to @kishoraditya just now about something else - “I think the challenge with FDD’s “at a glance” stuff so far is that it’s still quite complex - e.g., the bucketing of what FDD is focused on in the four buckets is a step in the right direction, but understanding 30 broad strokes bullet points and what’s underlying them is still far, far too granular / lacking context to be a summary”

As a Steward coming in to review this, my head goes towards:

  • What are the top 1-2 priority / impact items in each of the four buckets?
  • What is the resource allocation / rationale for each of those?

That is where I would want to start to make sure I’m understanding & agreeing with the most critical items. Beyond that, as a Steward, I care less about understanding the long-tail (unless, of course, it’s taking up substantial budget - in which case I would want to go into its rationale, but only after understanding the core pieces).

1 Like

I know FDD is taking a lot of steps to evaluate where changes can be made. I appreciate the work and contemplation.

I am still not sure how to vote until the final thoughts are shared (I think FDD is going to revise again tomorrow after an internal meeting per my conversation with Joe).

1 Like

\

Thank you on behalf of the GIA :slight_smile: We appreciate your praises and the fact that you have noticed us means we are getting stuff done and that’s nice…we have actually worked a lot and also put in a lot of passion, late hours and brain matter :brain: into defending Gitcoin from threats to its credible neutrality. It’s not easy for us to make friends when our mission is to maintain fairness (in the degree that we can and are allowed). We are decentralizing more and more each round and building, modelling based on our results. We work very closely with the “Sybil defenders” and the data science @omnianalytics have been slowly but surely building on our results. We are working to become a fine tuned machine that has the purpose to integrate with Grants 2.0

It’s not healthy to compare things, but I was want to give this as an example:

I personally help out with the MMM(because I love it) pro bono because I like co-hosting and participating in fun events that are beneficial for Gitcoin and the web3 space overall. I don’t have the deepest understanding on what’s going on over there, but I do find it intriguing that the MMMs budget is a decent amount and it was approved by almost everyone, which signals that the work and outcomes generated by them are visible, while it seems that our collective work in the FDD has not reached the same level of awareness… :frowning:

Tough times, we will make it through. :robot: :blue_heart:

I apologize for all my “emotion” filled posts and thank you for all the work you are doing! Without the collective efforts of the Gitcoin core team we would not be here and we should also remember that
:smiley:

2 Likes

AGREE.

FDD is working on a revised budget. Hopefully to post on Friday 5/13 with input from other leadership in Gitcoin.

10 Likes

We have decided to go back to the drawing board rather than push forward a proposal which doesn’t take the current situation into consideration. We would love to get more feedback from you here:

https://snapshot.org/#/gitcoindao.eth/proposal/0x332237cf15443ea727e914431b9d916aa2b72d421ce1fa2f8165cc7774dbde0e

Simply select the squads or principles you agree with and vote to help us get an understanding of what is most important to you.

We hope to ratify an updated budget request by Thursday, May 19th.

1 Like

hey Joe, voter turnout on this looks a bit low. i wonder how we can get that up.

there are many options in the poll, which can be complicated for stewards to wade through.

from speaking with stewards, i perceive that many of them would be aligned with the following:

  1. lean Grants 1.0/ GR14 sybil protection & grants approvals
  2. Grants 2.0 rebuild for decentralization/protocol-centric fraud defense

i wonder if a more thematic poll with fewer options would yield more turnout?

6 Likes

Hey friends on GitcoinDAO :earth_americas:

Sorry for the late comment. I’ll express my contribution which will make the case towards us exploring what it takes to push for more evolution and how this passes through sketching how a Sybil Detection DAO could look like, as we can and should perform incentive alignment around our ultimate goal of providing Web3’s best Sybil Detection as a Service.


Most of the current function as well as some challenges of FDD has been described in the discussion here. FDD performs a critical function in GitcoinDAO, as Sybil Resistance and Collusion Resistance are critical for the legitimacy of Grants.

This is done through a series of increasingly sophisticated processes that have been developed by the entire FDD group since before the DAO was created, and required because our ecosystem demands ever increasing technical excellence with no compromises whatsoever on transparency, decentralization and community ownership on the input, process and outcome. It has been a process of pioneering both in how we organize (governance of algorithmic processes - i.e. the machine learning pipeline), and how we detect and defend against sybil attacks and collusion (governance by algorithms, in a DAO).

Going deep and evolving this critical function is not easy nor trivial, and it can be far from sexy when scratching the surface of what it should be. After all, it is about the ongoing operation and maintenance of GitcoinDAO as infrastructure. It also requires domain expertise to develop, operate, iterate on, and maintain the machine learning sybil detection pipeline.

Additionally, FDD has arguably managed to create one of the most distinctive initiatives on Web3 so far in terms of real-world Sybil Account Detection - one of the space’s biggest challenges, especially where quadratic voting is employed. The unique blend of experts together with the holistic / integrated processes that were battle-tested on one of the Web3 mostly used platforms is certainly something to be proud of. Something that can also be leveraged and used by all value-aligned platforms on which their success depends on reaching out to real users without compromising on their privacy or incentive alignments.


Given the above considerations, it can be helpful to also share some thinking about what can be a possible strategic direction and a proposal for next steps.

We want to ensure FDD and GitcoinDAO have incentive alignment AND functional alignment to defend against sybils. I have written a tale below that provides an analogy for what is functional alignment in this case.

We think it could be possible for FDD to evolve further from a GitcoinDAO core team, to become its own DAO that is a service provider to Gitcoin, and potentially other DAOs. FDD is a critical function for DAO maintenance (especially with Quadratic Voting), but it also may likely be a useful service to other DAOs, which would reduce cost dependencies on GitcoinDAO.

This being true, then it could be the best long-term sustainable path by both improving the sybil detection (through bi-directional interface points with other data partners) and reducing the cost to GitcoinDAO as various ecosystem partners will be involved in the usage and cost sharing on this new Sybil Detection DAO.


The next step is to Explore. To sketch how it could look like and to map its limitations and opportunities. To discuss as much as possible while there aren’t points of no return. To have a document which allows the community at large to constructively criticize and reject in detail or in its whole form. This is the core intent and function of the SDD work-stream, and it is expected to take the following form:

  • Deliverable: A written briefing report that includes:
    • Definition of a clear purpose and functional remit for SDD
    • Diagrams/sketches of potential SDD token governance and economic systems
    • Summary of the results of discussions with potential data partners and scoping of bi-directional interface points
    • Description of a technical roadmap for making SDD viable
    • Recommended next steps
  • Method: a 3-phase process, on which:
    • Phase 1: Freestyle sketching
      • A strawman sketch will be produced based on the combined knowledge of how the SDD + ecosystem tech & dynamics is playing right now.
    • Phase 2: Intense Iterations
      • The strawman sketch will pass through extensive review from SMEs to ensure that its cohesive and viable
    • Phase 3: Polish & Communicate
      • The iterated document will be transformed and published into an form that invites public-facing iterations & criticisms

I hope these thoughts help share the broader context going into this function, its value to GitcoinDAO and the broader ecosystem, and the potential strategic direction.


Extra: The Tale of the Decentralized Common Goods Factory and its own Sybil Detection Generator

(If is too wacky, replace DCG Factory → Factory, and Sybil Detection → Electricity)

Let’s imagine a fictitious factory. The Decentralized Common Goods (DCG) Factory, which is owned and managed by the diligent Quadratic Bot, which concerns itself only with maximizing the quality and the positive externalities of what’s being produced.

Quadratic Bot knows that the DCG Factory consumes a lot of Sybil Detection energy, and in order to have consistent production results, he decides to purchase and operate its own Sybil Detection Generator, which provides this critical piece of infrastructure to the factory to be autonomous.

At some point in time, Quadratic Bot decides that the factory should be less monolithic and that its core functions should be more decentralized and autonomous. As a result of that, the Sybil Detection Generator now has its own manager: Not-Sybil.

Not-Sybil, which is as diligent as Quadratic Bot, also only concerns himself with the quality and the positive externalities of what’s produced too. The product in this case is delivering the best Sybil Detection to his users.

As the DCG Factory is decentralized and open-source, the commoners of the local village quickly take notice of all the innovation happening inside it, including the Sybil Detection Generator, which is tailored end-to-end for being as optimal as possible for its function.

Not-Sybil, observing the demand for Sybil Detection and taking in mind his mission of maximizing positive externalities, decides that it can be a good idea to expand the Sybil Detection Generator capacity and technology for more deeply integrating with the community demands.

Quadratic Bot is enthusiastic, but this capacity is way more than he needs, and it is not obvious how the payoff is going to be in terms of the quality improval on the factory products. It may even be a distraction.

Quadratic Bot and Not-Sybil are intertwined in a dilemma. There are no questions about the history and the ownership: it was Quadratic Bot who bought and took the risk for the Sybil Detection Ownership. But now that it is being managed by Not-Sybil, the goals are different. Quadratic Bot wants the generator to provide critical services. Not-Sybil wants the generator to provide the best services.

How do we untangle that? One option is to re-absorb the Sybil Detection Generator under the Quadratic Bot guidance and dismiss / re-educate Not-Sybil. Make it provide the critical service again.

Another one is to imagine a future where the Sybil Detection Generator becomes its own autonomous organization. On which Sybil Detection is also a decentralized common good product too, which is used as a critical service by multiple villages and projects.

The future is uncertain and risky. Discovering and mapping it beforehand, and making explicit the landmarks and obstacles can allow for us to have a grasp of the journey without taking points of no return.

10 Likes

Wow, love that post and the storytelling inside…

Feels like Ethereum and Ether, or a car can’t run without gas/electricity.

Nothing much to add here but really enjoyed to read this @danlessa :slight_smile: :star_struck: :clap:

6 Likes

So FDD had some autonomy and realized it could make a bigger impact on GitcoinDAO’s mission by expanding positive externalities outside of Gitcoin, and this is the misunderstanding.

GitcoinDAO stewards may see FDD going outside it’s critical mandate as not within it’s mandate.

While FDD may see it as serving GitcoinDAO’s mission better.

The way we solve this is to transparently and publicly discuss the situation. Perhaps FDD should lose some autonomy and scope as to not confuse the stewards, but then it should probably also spin out sybil detection DAO to serve Gitcoin’s mandate if the Stewards agree that it has value and fits within the GitcoinDAO scope.

6 Likes

We plan on making a new budget request using this framework if ratified by FDD source council tomorrow.

I would say options 4,3,2 are all part of FDD’s critical mandate only on a short, medium, and long term timeline. Option 5 includes the concept covered succinctly by Danilo above.

6 Likes

Translated into hype slang:

Vote for the FDD Starter Pack
Vote for the FDD Premium Pack
Vote for the FDD Full Pack
Vote for the FDD Full + Expansion Pack

:smiley:

7 Likes

One more refinement before we post, hopefully tomorrow.

The voting will ask voters which option they would fund at the highest level. Then votes will be added from higher levels to the lower levels.

If Option 1 has 1 million votes, and option 2 has one million votes, then option 2 will be considered to have 2 million votes.

Then, if option 3 had one million votes, it would be considered to have 3 million votes, enough to pass quorum, therefore option 3 would win.

5 Likes

Started a new thread with the new proposal [Proposal] FDD Season 14 Budget Request

3 Likes

@kyle I’d love to get your thoughts on this. I think the sybil detection DAO plan could potentially be the driver of dPoPP adoption.

2 Likes

I think maybe in the medium term it could be…

We have a need in serving Gitcoin’s use case first (and driving adoption that way). As mentioned previously, I see this as a crawl → walk → run approach. lets solve our need, using the protocol (and developing the protocol) before we turn outward for customers and guidance.

Gitcoin Passport is like a baby bird. We are not ready to throw it out of the nest, but we do see a big blue sky of potential.

2 Likes

IMO this is going to require less data science/ML focused, more devrel, marketing, mechanism design, fraud detection + remediation + governance management. would be interested to build a strategy around it.

3 Likes

Maybe this round isn’t the time to officially put budget towards it, but if we did the initial discovery this round, it could be great timing. We will keep it an option on the vote.

Here is what we would do if it was voted for:

Problem Narrative

In order for a borderless democracy on the internet to succeed, it will require sybil resistance. Gitcoin Grants has shown us that the machine learning detection models can scale, but it does not provide a sustainable source of funding to build a digital public infrastructure.

Solution

A model for launching Sybil Detection DAO (SDD) needs to be designed. This begins with standing up supply side microservice subDAOs specialized in incentivizing one behavior. FDD will use funding from Gitcoin and potentially other grants to fund the contributors to build these microservice subDAOs while earning governance in the future Sybil Detection DAO via an issuance curve.

Season 14 Deliverables

  • Reaching potential data partners and scoping bi-directional interface points
    • Interface point 1: Data flows
    • Interface point 2: Token/governance flows
    • PoC Use Case: Bright ID
  • Sketching and modeling potential SDD token governance and economic systems
    • Include how microservice subDAOs would be embedded in the scheme
    • Include budget estimates for operational costs for 1 year
    • Mapping of microservice protocols to launch prior to SDD with roadmap
  • Description of a technical roadmap for making SDD viable
    • How incentive mining might perform in SDD microservice subDAOs
    • System engineering graphic identifying inputs and outputs
  • Research and recommendations for legal “container”
  • Research and describe integration points between SAD, SDD, and dPoPP
3 Likes