GG23 OSS Program Quadratic Funding Results

GG23 OSS Program Quadratic Funding Results

Our GG23 OSS Program Quadratic Funding results are live! We’ll have one week for review and feedback then, barring any major issues, will proceed with payouts by May 2, 2025.

The OSS Mature Builders Retro Funding round results will be posted to the forum next week in a separate post.

TL;DR

  • In GG23, we moved towards a deeper builder-centric approach in the way we fund public goods, focusing on supporting builders at every stage of growth through a new pluralistic, multi-mechanism design.
  • We implemented the same two-pronged sybil resistance strategy as we’ve been using since GG20. We used a pluralistic variant of QF and we used Passport’s model based detection system. We did not do any closed-source silencing of sybils/donors. Instead, we’re solely relying on our mechanism and Gitcoin Passport.
  • Discussion will be open for one week before payouts are concluded. A GG23 pre-ratification proposal was passed through governance, speeding up the process of payouts to grantees.

Full QF Matching Results

GG23 Overview

Every round sees new developments. Some of the most exciting in GG23 included:

OSS Program Quadratic Funding Rounds: Key Metrics

3 Program Rounds
$600k Matching
$95,278.16 Total Crowdfunded
9991 Unique Donors
235 Projects

:bulb: Round & Project Spotlight

Here are the top five projects by total matching funding (gleaned from the overall matching results). The projects on this list are the ones with the most diverse bases of support, regardless of the size of the base.

GG23 OSS Developer Tooling & Libraries

View this round’s report card here.

Project Name Matching Funds (USDC) Matching Funds (USD)
Human Passport (formerly Gitcoin Passport) 20000 $19,997.88
rekt.news - The dark web of DeFi journalism 20000 $19,997.88
growthepie :pie::straight_ruler: 20000 $19,997.88
OpenZeppelin Contracts Library 20000 $19,997.88
Open Source Observer 20000 $19,997.88

GG23 OSS dApps & Apps

View this round’s report card here.

Project Name Matching Funds (USDC) Matching Funds (USD)
Karma GAP 10000.0 $9,998.94
Treegens DAO🌳 9367.02 $9,366.03
GainForest 9293.43 $9,292.44
Kolektivo Network 8698.34 $8,697.42
$EARTH 8213.56 $8,212.69

GG23 OSS Web3 Infrastructure

View this round’s report card here.

Project Matching Funds (USDC) Matching Funds (USD)
BrightID :high_brightness: Universal Proof of Uniqueness 20000 $19,997.88
Superchain Eco 20000 $19,997.88
ethOS 19148.79 $19,146.76
eth.limo 16004.09 $16,002.39
Deep Funding 12292.52 $12,291.22

:abacus: Round and Results Calculation Details

Before GG20 began, we proposed a two-pronged sybil resistance strategy. To recap it briefly, we would continue to use COCM (Connection-Oriented Cluster Matching) as we had in GG20 & GG22.

As explained in the post, and the paper which introduced this mechanism to the world, COCM is much less vulnerable to Sybil Attacks than ordinary QF because it reduces the matching of donors who look similar.

In addition, over the past few rounds, Passport’s Model-Based Detection system has yielded greater sybil resistance than the stamp-based system without any of the user friction. When used together, we believe these two tools produce the most sybil-resistant results we’ve ever had.

Connection-Oriented Cluster Matching (COCM)

COCM is an approach within Gitcoin Grants that enhances Quadratic Funding by prioritizing projects with diverse support bases, helping to counteract sybil attacks and coordinated groups trying to unfairly influence funding distribution. Key elements include:

  • Diversity Bonus: COCM increases matching funds for projects supported by a wide array of “tribes” or distinct groups, rewarding cross-group support and broad appeal.
  • Markov Chain Enhancement: Recent updates to COCM use a Markov chain approach to gauge connection strength between users and projects, providing stronger sybil resistance by analyzing indirect connections.
  • Funding Distribution: COCM’s design inherently shifts some funds from top projects to those in the “long tail,” enhancing the reach of smaller or niche projects.

Gitcoin Passport’s Model-Based Detection System

The Gitcoin Passport team has implemented a model that evaluates on-chain activity to detect potential sybil accounts, enhancing funding fairness by:

  • Address Probability Scoring: The model assigns each address a probability score, indicating whether it likely belongs to a genuine user or a sybil account.
  • Continuous Improvement: While the model sometimes mistakenly flags new users with limited on-chain history, the team is refining the dataset and expanding to include Layer 2 networks, effectively reducing sybil influence on funding results.

These efforts together strengthen Gitcoin’s defenses against manipulation, helping to ensure fair distribution of funds to genuine community-supported projects.

Code of Conduct Reminder

As a reminder to all projects, quid pro quo is explicitly against our agreement. Providing an incentive or reward for individuals to donate to specific projects can affect your ability to participate in future rounds. If you see someone engaging in this type of behavior, please let us know.

We have removed Pheasant Network & Vine Finance from matching due to the fact that both projects participated in quid pro quo during GG23.

Next Steps

We plan to distribute the matching by May 2, 2025, after the results are open for discussion for one week.

It’s worth noting that GG23 pre-approved the matching fund to be paid out before results were posted. This means that the payout process to grantees will occur faster, after this post has been left open for community input.

We are also hosting an internal retro in the following weeks and will publish further results and learnings. And as always, a detailed blog post will be published on the day that payouts are distributed.

If you know of any projects violating any of our agreements, please do let us know.

We’re also always looking for direct feedback from the community on which improvements would make GG24 even better. Please don’t hesitate to let us know!

11 Likes

Awesome!
This is a very comprehensive report. Congratulations to all projects that came top in this round.
@MathildaDV when would results for Web3 for Universities round be out?

gm @Johnadek, you need to reach out to the Round Operators for the Web3 for Universities Round, as Community Rounds aren’t managed by Gitcoin.

1 Like

Ah I see!
Thank you Wasabi. :pray:
@KarlaGod any update please?

Are projects who have intentionally colluded and participated in sybil attacks excluded? I picked a couple of receiving addresses from the QF calculator and ran them through Arkham. There is a clear intentional pattern of sending large amounts from the receiver address to other passported addresses through the across bridge 2 hops, who then go on to donate more than 99% of the funds forwarded back to the colluding receivers. There are even projects who both forward to the same users who go on to donate back to those projects and just dust a couple other projects.
Looking at the matched donors and amounts its impossible those donations have been ignored, given the full match amount is close to the total the project collected. Those projects are with very high gitcoin matching results and seem to have done a ton of recycling of funds through passported addresses?

1 Like

gm @CheatDetector, please fill out this form with those findings https://forms.gle/gjnz7mtCcd6aRTFw6

2 Likes

This is very detailed

Great job here @MathildaDV

1 Like

Thanks for organizing and the detailed result report!

We are from dDevKit, this GG#23 dev tools round we ranked #5 by Most contributors (168) & #6 by Most donations at the end.

Seems we have mismatched expectations that per above results we end up ranking #22 with ~$1.8k matching funding. Understand the role of COCM, where we tried to attend shill space, mobilize our network and onboarded quite some donors new to gitcoin along the process.

A few things we wonder are

  • did we miss out anything or any patterns we should avoid
  • if dev activities/ contributing metrics of repo is part of the formula? before the round we changed our github repo name, not sure if there is implication.
  • thus we also created a new gitcoin project thus past round is not attached, will that affect the result?

Our project perhaps doesn’t exists without gitcoin in the first place – We’d appreciate it if someone could shed some light on this to help our understanding. We also wonder if gitcoin round is good for us to onboard users, not just rely on highly active wallets Web3 ecosystems in order to align with the formula.

1 Like

“We also wonder if gitcoin round is good for us to onboard users, not just rely on highly active wallets Web3 ecosystems in order to align with the formula.”

I was about to ask the same. Thanks to raise it up. I look forward from any Gitcoin’s team answer.

Thank you for such an extensive report.

As PM at Animal Social Club this has been a first direct experience with Gitcoin QF.

Traveling during the fundraising weeks impacted team’s coordination and thus weakened our marketing approach, but still we’re happy to have learned a lot and for any funds we’ll receive.

Also, it’s been amazing to see many friends met IRL at conferences running their QF campaigns and achieving some great results.

Happy for all. Congrats to Gitcoin. Appreciate for this growing opportunity :pray:

1 Like

Thanks so much for the detailed report!

We’re building Mini Bridge and ranked #5 for both Most Contributors and Most Donations in the Infra round. But our matched result came out at #11, and we’re trying to figure out why there’s such a big gap between the donation rank and matched rank.

Really appreciate your support for our project!

2 Likes

Hi Gitcoin Team,

First of all, thank you for all the hard work in organizing GG23.
I have a quick question regarding the matching calculation under the COCM model.

I noticed a case where two projects have very different outcomes despite differing crowdfund amounts and number of contributors.
For example:

  • ABI Ninja received $79.68 in crowdfunds from 41 contributors, and matched $5,298.
  • SuperUI received $250.45 from 79 contributors, but matched only $356.

Could you help clarify how the diversity score or sybil resistance mechanism affected this result?
Is there anything we could do in the future to better optimize for matching outcomes, apart from increasing total crowdfunds or contributors?

Really appreciate your time and all the improvements made to the grants program. :pray:

3 Likes

I am sure they have logics and good reasoning. And I agree with you that Gitcoin has been supporting web3 community so long I can remember. Personally I don’t really care about the amount/matching fund we have received. Community showed love and support. It’s good enough knowing 1400+ people went to Gitcoin to show support for our project. :pray:

2 Likes

Thank you for the detailed report @MathildaDV !
Can you help me understand where Unlock DAO is in all of this, or didn’t we end up not making it at all? I knew only after that I should have applied for the Infrastructure round but well there is always something to learn …

2 Likes

Check the tabs in the document.

Thanks for the comment @debuggingfuture. So, yes this is COCM in action. With traditional QF, your match would’ve looked different, but due to us using COCM (which is the most sybil-resistant), your match amount would look different.

This is due to the fact that COCM favors projects that have a wider range of donors from various communities. So, if you have a donor base that only donates to your project and not other projects, this is where the difference comes in.

No, COCM takes into account donor behaviour, and coupled with Passport’s MBD (as outlined in this post), the onchain activity of the wallets donating!

Hope that clears things up for you!

Thanks @DeFiTeddy! Yes, the amount of donations is important in traditional QF, but in COCM having a wide range of donors is more important, so for this reason that’s why you’re seeing that gap. As outlined in this post, the two-pronged sybil resistance that we have ensures a fair distribution of funds. We always encourage projects to suggest to their donor base to not only donate to their project, but to others in the round as well!

Hope that clears things up!

I see Unlock Protocol listed in dApps & Apps.

Thank you. This is being investigated, alongside all other reports that we have and that we may receive.

Thank you for the response @MathildaDV it is helpful.

We understand motivations of COCM and have the trust it is well designed & executed.
Thus I think we are aligned with COCM in principle, however we still find it challenging to understand how it actually play out given the algorithm complexity.

It will be great if some heruistic metrics/elaborations are published alongside the matching results (besides on-chain calculations as proposed),
for examples we wonder, for each project

  • how many donors are filtered in the first place (base on Passport MBD)?
  • % of donors donating to other projects?
  • donor similarity for the project and how is that being calculated?
  • visualized comparsion re diversity for projects?

In our case in last round with same COCM mechanism we were fortunately a top matched project (hackathon alumini). While this time similarily we raised donations from quite some different communities and with the large contributors count come organically, we are left feeling “yea we should be diverse with a valuable community” with COCM saying “no not actually diverse enough”.

Anyway just two cents from a project perspective, and we would love to communicate things clearly to our supporters as well.
Still have to say, thank you Gitcoin!