Collaborative Research Round 03 Insights & Learning

$42,000 USDC Distributed In Research Grants

The OpenCivics Collaborative Research Round 03 generated valuable insights into the evolving landscape of on-chain quadratic funding for public goods. This retrospective offers some general reflections with regard to the round specifically as well as a set of potential upgrades and coordination efforts that we hope will improve Allo Protocol and Passport as products and the entire Gitcoin ecosystem as a community exploring similar design challenges.

Overall, the round was the most streamlined that we’ve run thus far. We chose not to implement a cap to the total number of projects included which minimized the deliberative process and resource triage of our previous round. Each round creates an opportunity for on-going learning and development of our Grants Program and we look forward to sharing what we learned!

Results Overview

OpenCivics Collaborative Research Round 03 distributed $38,750 USDC in matching funds to 16 projects. The round raised $3,778.70 in crowdfunding from 141 donors with 337 total donations, averaging $11.21 per donation. Overall, donors generated a 10x multiplier of their donations in matching funds.

Compared with our previous round, both the number of donors and the total crowdfunding amount saw significant decreases.

• Total Number of Donors:

  • Consortium Grant Round 02: 1,062 donors

  • Collaborative Grant Research Round 03: 327 donors

• Total Crowdfunding Amount:

  • Consortium Grant Round 02: $10,306.50

  • Collaborative Grant Research Round 03: $3,778.72

The percentage change between the OpenCivics Consortium Grant Round 02 and the OpenCivics Collaborative Research Grant Round 03 is as follows:

• Total Number of Donors: Decreased by 69.21%

• Total Crowdfunding Amount: Decreased by 63.34%

Overall Data Insights

Key Findings:

1. Top Performing Projects:

   • MycoFi for Municipal Resilience and Stephen Reid: Technological Metamodernism both received the highest matching amount of $5,518.75. These projects also garnered a significant number of contributions (33 and 37, respectively) and large amounts of total received funds. The high matching allocation suggests strong donor engagement and broad support.

2. Moderately Funded Projects:

   • Ethereal Forest’s Open Protocol Research Group and VoiceDeck: A Marketplace for Journalism Impact received moderate matching amounts of $4,271.33 and $4,012.55, respectively. These projects had a reasonable number of contributions (23 and 28) and received significant crowdfunding amounts.

3. Lower Funded Projects:

   • The Bonding Curve Research Group (BCRG) received the lowest matching amount among the highlighted projects, totaling $2,621.20. With 21 contributions, this project still managed to secure a reasonable level of support, but it fell below the others in terms of both contributions and matching funds.

Click here to view the automatically generated Gitcoin Report Card for the round.

For a detailed data, tables, and graphs of the grant round results, check out Collaborative Research Round Analytics on our wiki.

Round Operator Insights

Crowdfunding Decline & Community Engagement

One of the most notable observations during this round was the decline in crowdfunding donations. This reduction might be partially attributed to the broader market downturn, but it is more likely linked to the absence of an open-source software round that typically runs concurrently with key community rounds.

A critical lesson from this experience is the need for the impact-focused segment of the Gitcoin community to extend its outreach beyond the existing web3 space. We see a significant opportunity to engage with other communities that are interested in funding social impact projects but are not currently involved in the space. This challenge could be addressed by coordinating efforts to bring traditional nonprofit ecosystems and climate organizations into the quadratic funding ecosystem.

One of the main barriers to this type of expansion has been the lack of web3 knowledge within these communities. However, the introduction of a fiat PayPal donation option in this round has proven to be a significant step forward. This innovation, made possible through a partnership with viaPrize, allowed us to create a seamless donation process for those unfamiliar with web3, while still ensuring that donations were eligible for matching funds. The success of this experiment suggests that we can now onboard non-web3 participants using traditional payment methods, allowing them to engage in the ecosystem without the steep learning curve associated with purchasing and bridging different cryptocurrencies on different blockchains. For a full retrospective on the viaPrize payment system experiment in GG21, please review founder Noah Chon Lee’s post on the Gitcoin Governance Forum.

Challenges in Matching Calculations

Another key learning from the round was the complexity involved in calculating matching amounts using cluster matching, minimum matching amounts, changes in matching pool size, and whitelisted wallets.

Upon reviewing the cluster matching results, we observed discrepancies between the expected matching results and the actual outcomes. This discrepancy was traced back to a parameter within cluster matching that reduces the matching for wallets that only donate to a single project. While this serves as a useful Sybil protection mechanism, it also unintentionally penalized grantees with unique donor bases mobilized specifically for their projects. To address this issue, we employed a 50/50 split between cluster matching results and quadratic funding results.

Additionally, we had to adjust the results due to a reduction in the matching pool’s size, caused by holding a portion of the pool in ETH rather than USDC. While we immediately swapped ETH for USDC as the market turned, but still lost about $4,000, which we corrected for after the round by manually modifying the matching results to reflect the correct matching pool size.

Lastly, we deployed an experimental $1,000 minimum matching to ensure that all projects selected for the round would receive at least enough grant funds to complete a portion of their research inquiry. Round results were modified by filling the gap in matching for projects that fell below the $1,000 threshold by redistributing matching funds proportional to the amount that each project over $1,000 received.

All of these modifications, the calculations used to derive them, and the original figures are listed on our wiki and can be audited by anyone to determine how those choices were made. In the future, we would appreciate features within Gitcoin Manager that would allow us to reduce the matching pool size, set minimum matching amounts, and modify sybil parameters.

Sybil Protection and Model Transparency

While the cluster matching results in our previous round were intuitively smoother and generally more equitable, cluster matching results in this round were vastly different than the traditional quadratic funding results. Given the large difference, we began to look deeper into what was driving the changes in the underlying model, something we had not thought to inquire into prior. Given what we learned, we want to instantiate a larger public conversation around sybil protection, the tradeoffs between transparency and opaqueness for model-based sybil detection, and other ways we might be able to rethink matching mechanisms.

While the Passport team’s decision to reduce matching for single-project donors is well-intentioned, it highlighted the importance of donors understanding the evaluation metrics of cluster matching. Without this understanding, the algorithm’s outputs may lead to unintended consequences. We believe it is crucial for donors to be informed about the factors influencing their matching scores upfront, enabling them to consciously adjust their behavior. Of course, this transparency could also be used to game the system, but we felt concerned about the unintended negative consequences for some grantees who could have otherwise informed their donors that they would need to also donate to other projects in order for their donations to be eligible for matching. We feel that this kind of information could actually generate positive effects for rounds by encouraging donors to consider all the projects for donation instead of just the projects of their friends. In the future, we may tell our donors something like:

Your donation is a weighted preference vote for which projects should be funded. In order to know how much you value a project, we need to understand your preferences in relationship to at least one other project. If you do find that your preferences are nearly 100% aligned with only one project in the round, please make at least one other small donation to another project to ensure your primary donation is eligible for matching.

We recognize that the previous stamp-based Passport system is cumbersome for most users and were very excited for the new donor UX made possible by cluster matching, but we believe that greater transparency is needed regarding how cluster matching scores are determined. The lack of transparency, which makes sense from the perspective of sybil protection, puts us as round operators in a difficult position, as we must justify round results that stem from an algorithm we do not fully understand.

Furthermore, upon understanding more clearly how the cluster matching method does reflect a set of values and perspectives (ie what behaviors are encouraged or penalized), we believe that such value judgements should not only be made transparent but should also be composable by communities to reflect their values and needs. Moving forward, we aim to make this value judgment more transparent while also exploring the possibility of new features and approaches that empower communities to customize matching parameters. This customization could include incentives and rewards for specific behaviors, creating a more pluralistic approach to sybil resistance and community participation. For example, we are interested in exploring “tunable QF,” an approach to matching that modifies matching amounts based on the particular token holdings of any given address. We would be interested in experimenting with a pluralistic approach to matching modifiers and would love to beta test tooling that enables us to link various ERC-20 and ERC-721 contract addresses as matching criteria, enabling us to increase the matching signal from OpenCivics Network members with membership NFTs, $REGEN token holders, or staked GTC, as a few examples.

Round Innovations and Feedback

One piece of feedback that emerged early on in our application for matching funds was the ease and clarity of attestations in a research-focused round. The binary nature of research publication—either a project publishes its research or it does not—makes impact attestations straightforward to document. This simplicity may make research-based quadratic funding rounds more attractive and easier to manage. We’re incredibly happy with the grantees in this round which demonstrated a wide diversity of inspired, useful, and forward reaching research inquiries into a wide range of domains within the field of web3 civics or DeCiv. To hear from just a few of the amazing grantees in this round, head over to our Round Showcase Space on X. We look forward to sharing the research that’s produced through the funding of this round!

We consider our PayPal fiat experiment with viaPrize to have been a massive success. Thanks to the amazing work of Noah and his team, combined with efforts from the Gitcoin Grants Labs team to enable a feature for white listing the addresses generated by viaPrize, we were able to process fiat donations for the first time while addressing sybil protection. We encourage folks to read more about Noah’s method and results in his retrospective post to learn more about how it works and to decide for themselves if they want to deploy it for their own rounds in the future.

Lastly, we experimented with implementing a minimum matching amount to ensure that all grantees received enough funding to impact their proposed research meaningfully. We set a minimum of $1,000, redistributing matching amounts proportionally from grantees who received a larger percentage of the pool. Given the circumstances we’ve encountered in both of our previous rounds in which some grantees do not receive enough funds to take meaningful action on their grant project, we feel that the additional redistributive burden on the higher performing grantees, as well as the burden on round operators who must review applications much more carefully given the cost of minimum matching per project, is ultimately worth the tradeoffs and we look forward to experimenting with matching minimums again in future rounds, perhaps with different approaches to the threshold and the threshold qualifications.

Conclusion

Overall, this round provided critical insights and valuable lessons that we’re already beginning to build upon for next round. We are committed to continuing our partnership with Gitcoin and the entire Community Round Operator ecosystem, and we’re grateful for the incredible support we’ve received from the Gitcoin and Passport teams as well as other round operators. We look forward to building upon these experiences, refining our processes, and exploring new possibilities for fostering impactful, transparent, and equitable public goods funding.