2024-06-13 RegenLearnings.xyz
Call Summary
Call notes:
Welcome to the latest Regen Learnings call, your monthly dose of fresh insights into public goods funding, Web3 innovations, and community-driven growth. Hereβs what went down in our latest session, packed with updates, demos, and future plans.
-
Intro Kevin set the tone with a warm welcome and agenda:
- Metrics for retro funding for Optimism.
- Easy Retro Pgf used for funding rounds.
- Collaborative efforts with Thrive and other ecosystem players to enhance cross-pollination of ideas.
-
Carl presented his recent work helping Optimism to use metrics for Retro Funding 4: The big difference between the previous round and this new one is: Instead of selecting projects directly, metrics will guide which projects receive support, so basically you can select the metrics you care the most and put them in your portfolio, combine them in your ballot and giving each a weight and see how those metrics apply to the project you are reviewing and supporting, and the script that will decide how to allocate the funding.
- The round has completed the sign up phase::
- Over 500 applications expected.
- Both human and algorithmic checks to narrow down to 250 projects.
- Voting on finalized metrics begins at the end of the month.
Testing App Demo
Users can now build a ballot by choosing metrics that matter most to them, giving more weight to trusted usersβ preferences.
Community Updates
Community members shared their cool projects and updates, fostering a vibrant, collaborative atmosphere.
Data Model for Project Comparison
Weβve built a robust data model to clean and process data from six different blockchain chains, offering comprehensive project comparisons. An API endpoint now allows easy retrieval of project metrics.
Defining Custom Metrics & Trusted User Model
Users can create custom metrics, enhancing our data-driven approach. The trusted user model aggregates various trust signals to identify and give more influence to reliable users.
Also, Carl shared that they are exploring running Curation Algorithms Weβre exploring algorithms like the EigenTrust to help users discover projects based on social connections. This approach promises a more personalized and effective way to find new and exciting projects.
Question from Umar Khan: How do you define trusted users? Carl: Great question, thereβs been a huge amount of feedback around it, there are 4 different trust signals that are being ingested:
- Have a Farcaster ID of 20939.
- Have a Gitcoin Passport of 20 points or higher.
- Have a Karma3Labs EigenTrust GlobalRank in the top 50.000 of Farcaster users, or
- Hold a soulbound NFT (in this particular case Optimist NFT) in their wallet.
For the sake of experimentation we are doing another Data Challenge, he was hoping to be able to announce just applying these kind of reputation models to the on-chain data, if anybody wants to experiment there will be some bounties available. Weβll post the link on the RegensLearnings channel once it comes out.
- Thrive Protocol Overview: Ben West from Thrive shared insights into Thrive Protocolβs mission to attract and reward top talent through structured grant programs:
- Thrive Protocol works with ecosystems with relatively large grant or bounty programs to attract fund and scale top talent in the world at the Milestone of value creation.
- Focus on βTrust but verifyβ to ensure effective use of funds.
- Started working with the Arbitrum community, now we are starting to apply the same thinking with the Polygon community.
- Partnership with Karma to improve impact assessment through human evaluators.
Decentralized Review Process.
Refining our review process to ensure fairness and efficiency:
- Feedback loops and economic incentives for reviewers.
- Two-tiered review system with randomization to maintain impartiality.
- Aggregated feedback processed through AI for grantees.
At the moment, the whole process is being done in a spreadsheet, as they are testing out the methodology and concept behind what it is that they are doing, to get to a place where they are able to build this into a platform and do it in a scalable fashion. Recently finished their stage gate review looking at the grants functionality, that will basically be what replaces these spreadsheets, and they may be able to use it with Polygon, whoβs just came on board a few days ago.
Carl asked: Are the reviewers, is that public information or is that kept private so that people donβt interfere with them? Ben: So, who the reviewers are overall is public, whoβs reviewing which project is not, or what comment was derived by from them is not. The intention is to avoid any kind of Shenanigans, like people bribing people, people attaching people. Also we are not making the scores that one reviewer is giving public available to any other reviewers, this is somewhat different than the way we had been doing eligibility decisions at Gitcoin, where reviewers had a shared sheet and would actually have discussions and sort of shared learnings. I think weβll start integrating a phase after the initial scoring is done, where weβll bring people together and have conversations about lessons learned or key takeaways, and get feedback about the process overall and discuss the decisions that are being made amongst everybody involved once the scores have come in, that is a step we didnβt do this time but weβll have sort of end of the round wrap-up.
Ben invited to anyone who is willing to become a reviewer, they are always open to have more people involved, and if you have any feedback or comments about the process please share them in any of the governance forum posts at Gitcoin, Arbitrum or Polygon, we are really open to any critique and feedback. Here goes a link where to head (opens in a new tab).
- Gitcoin Grants Round GG20 Data Analysis: Armanβs analysis revealed a positive correlation between total donations and commit counts, highlighting the projects with strong community involvement. Future directions include expanding data analysis for deeper insights and trend identification.
Here is the link to his work (opens in a new tab).
Umar Khan thanked for his work and shared heβs especially interested to know what you find, and share it with other folks at Gitcoin, hopefully we can spread some more insights, and opened the door to jam on some stuff and collaborate together.
Before wrapping up the session, Carl launched the invitation to all who are going to be at ETH CC in Brussels next month, there is a little get together that we're putting on, and you can get register for that, and itβll be talking all about "Grant Fuckups", so come and share all your stories: The Good, The Bad and The Ugly of doing grants, and enjoy.
Transcript
Introduction
Welcome to the monthly Regen Learnings call, a place for people who want to learn in public and share ideas related to public goods funding. The intersection of priority theory, empirical data analysis, and program design is where the magic happens.
Agenda:
- Kevin's opening remarks
- Update on the data side by Carl
- Featured guest speaker: Ben West from Thrive Protocol
- Slido for crowdsourcing presentations
- Updates from the community
Background
By default, people would be learning within their own silos, within their own organizations, within their own companies. But because Web3 is a network and a movement, we want to create common knowledge between Gitcoin, Optimism, Giveth, Thrive Protocol, and all of the different organizations. When we learn in public, we force multiply the learnings across the ecosystem and build a stronger regen movement.
Data Side Updates π
Carl presented the following:
- Update on metrics for retro funding for Optimism
- Easy Retro Pgf being used for own rounds
- Conversations with Thrive and others in the ecosystem to take the best of this and use it there
- Cross-pollination happening
Funding Round 4 Preview π
- Difference: instead of selecting projects, people will select metrics, and the metrics will inform the portfolio of projects to support
- Funding allocation will happen through a script
- 500 applications, with human and quantitative eligibility checks
- 250 projects expected to make it through
- Metrics will be finalized, and voting will happen at the end of the month and into July
Testing App Demo π±
- Users can construct a ballot by adding a set of metrics to it
- Users can decide which metrics they care most about and combine them
- Concept of a trusted user: a user whose preferences are given more weight in the funding allocation
Data Model for Project Comparison π
A comprehensive data model was built to clean and process data from multiple sources, allowing for comparisons across various projects. The model consists of:
- 6 different chains: Op Mainnet, Base, Zora, Mode, Metal, Frax
- Data ingestion: Grabbing data from multiple sources and categorizing deployers and factories
- Event types: Identifying and defining event types
- Consolidation: Consolidating data into a few tables with relevant metrics
API Endpoint
An API endpoint was created to provide metrics for any given project. The endpoint allows developers to retrieve metrics using a simple API call.
Defining Custom Metrics
Users can define their own metrics using a calculation formula. For example, the transaction count metric can be calculated and used to power an app or grants program.
Trusted User Model π€
Carl shared a new project: The trusted user model, is based on aggregating different trust signals, including:
- Social graph-based signals
- On-chain activity-based signals
- Graph analysis-based signals The model helps identify trusted users and compare their transactions to the overall project transactions.
Trust Signals
There are 5 different trust signals being used, and a user must meet 2 out of the 5 criteria to be considered a trusted user. The signals are:
- Forecaster Data: Early forecaster user with a top 50,000 IC contrust glob rank
- Linked Addresses: Addresses linked to forecaster ID
- Gitcoin Passport: Score of 20 points or higher
- Optimist NFT: Holder of the Optimist NFT with an attestation-based web of trust
- Badge Holder Web of Trust: Part of the badge holder community with a social graph-based trust network
Curation Algorithms π
Curation algorithms are being explored to help users discover projects and services. One example is the IcanTrust algorithm, which is being protocolized by Karma3Labs.
The IcanTrust algorithm uses a user's social network to identify projects and contracts that are most used by their friends. This can help users discover new projects and services based on their social connections.
Future Plans
Future plans include running different curation algorithms on top of the data and creating a voting system based on social networks. This can help users identify projects that align with their interests and values.
π€ Thrive Protocol Overview
Thrive Protocol is a project that has been around for a couple of years, initially known as Thrivecoin. Its primary goal is to work with ecosystems on relatively large grant or bounty programs to attract fund and scale top talent and reward them at milestones of value creation.
Grants = Growth + Trust but Verify
The Thrive Protocol focuses on building a repeated game where the goal is to get better at setting up grant programs every time. The protocol aims to have the big picture lens on, remembering that this is a repeated game.
Thrive Protocol in Action
Thrive Protocol has been involved in working with the Arbitrum community, applying the same thinking to the Polygon community. The protocol has put 200,000 ππππππ‘ππ¦ πππ‘π πππ’πππ πππ‘π π‘βπ ππππβπ ππ’πππ sπππ‘π€πππ πππ’πππ πππ π‘βπ πππππ’πππ‘π¦ πππ’πππ , ππ π€ππl ππ πππππππ‘πππ ππππ‘βππ 50,000 directly into incentives.
Group Incentives Grantees Program directors, community members, and content creators Incentives Drive increased participation and maximize potential impact and effect of the round
Assessment of Impact
The secret sauce of Thrive Protocol is the assessment of impact, which has now started doing for GG20 in a partnership with Karma. The assessment involves setting up a Karma Milestone for each project, offering incentives to create Milestones, and attaching a pot of gold at the end of the Rainbow ($100,000 bonus) broken down to the top 10%, 20%, and 30%.
The Flywheel
Thrive Protocol creates a layer of incentives and feedback loops to ensure that funding is not just given but also evaluated for its impact. The flywheel idea is to create a cycle of incentives and feedback loops to ensure that funding is effective.
"Trust but verify" is the idea that you're not just funding stuff, but you're seeing what was done with the money, and it's good work getting done.
Partnership with Karma
The partnership with Karma involves building out new infrastructure to apply a rubric and not just a completed or not completed evaluation. Human assessors evaluate the work being done, providing a more accurate assessment of impact.
The Future of Thrive Protocol
The goal of Thrive Protocol is to attract fund, scale top talent, validate that work, and scale what's actually working. The results of the GG20 experiment will largely be seen in the new set of information that people have going into GG21. By providing a system for tiers of reviewers, Thrive Protocol rewards reviewers for their time and energy, giving them higher rewards the more consistently they are. This unique approach creates a more accurate assessment of impact, providing a proof of work 2.0 system.
Decentralized Review Process π
Consensus and Feedback The decentralized review process involves providing feedback that is not limited to the consensus view. This is where the "who watches the Watchman" idea comes in, ensuring that reviewers have an economic incentive to provide good, meaningful reviews.
Feedback Loops and Incentives
Additional feedback loop: reviewers evaluate comments from other reviewers who are not part of the consensus view, ranking the quality of their comments. Feedback is aggregated and processed through an AI model to generate a set of feedback for grantees. Incentivizing the right kind of work: reviewers are paid for their expertise and the quality of their work, not the amount of time they put in.
Reviewer Tiers and Randomization
Two-tiered system: Tier 1 reviewers are paid 100 ARB, while Tier 2 reviewers are paid 10 ARB. Randomized reviews: reviewers are randomly assigned to review proposals, and their scores can move them up or down between Tier 1 and Tier 2.
Impact Certification at Scale
Building teams of reviewers with economic incentives and finding ways to fund that work because of the benefit to the community. Creating feedback loops on various programs to help ecosystems allocate their resources more efficiently.
Pluralistic Grants Program
- Request for proposals process: over 100 proposals were submitted, and 60 were shortlisted.
- Initial review: 20 top candidates were selected, and each received 1,000 ARB to build out a more detailed second-stage application.
Review Process Breakdown Stage Number of Proposals Review Process Initial 100+ Initial review, 60 shortlisted Second 60 40 reviewers, 5 reviews each Final 20 Reviewers scored proposals, and funding decisions were made
The process is currently managed on spreadsheets, but the goal is to build it into the platform for scalability. A stage gate review is underway to improve the grants functionality.
Results Field
- Combined score of each proposal, including median score required and up/down yes/no vote.
- Rubric and scoring system were used to evaluate proposals, with both quantitative and qualitative feedback.
Lessons Learned
- Harnessing the wisdom of the crowd can lead to different decisions than those made by individuals.
- The process identified patterns in how reviewers worked, and adjustments were made to ensure efficient funding allocation.
API Integrations
We're talking about a whole bunch of different API integrations that will drive information directly into the Thrive platform for our human evaluators and assessors. There's also some stuff that we're automating the assessment of.
Reviewers and Scoring
- The reviewers are public information, but what score they gave or what comment was derived from them is not public.
- We're not making the scores that one reviewer is giving public available to any other reviewers.
- We're planning to integrate a phase after the initial scoring is done where we bring people together and have conversations about Lessons Learned or key takeaways.
Community Involvement
They're open to feedback about the process. If anybody would like to kick the tires and try being a reviewer, we'd love to have more people involved. Any feedback about the process can be added directly in the governance forum posts.
Data Analysis of Gitcoin Grants Round 20 π by Armand
Arman took on the task of analyzing the open-source observer data set with the Gitcoin data set to explore and see what insights could be gained. This report is a summary of his findings.
Analysis of Total Donation and Commit Count
- Total Donation: The sum of total amount donated towards each project.
- Commit Count: The number of commits made to each project.
Correlation between Total Donation and Commit Count
The analysis shows a positive correlation between the total donation and commit count for each project. Web.py has the highest commit count and total donation, indicating a strong community involvement.
Analysis of Contribution by Project Name
The graph shows the total donation and contributor count for each project. Web.py and nft storage have the highest contributor counts.
Insights and Observations
"It's interesting to see how many commits people made and how many total donations were made for each of these projects."
- The analysis highlights the potential for growth and development in each project.
- The data can be used to answer questions such as:
- Do projects continue to develop if they continue to participate?
- Does their velocity increase over time?
- Can they grow the number of contributors and maintain support over time?
Future Directions
- The analysis can be expanded to include more rounds of data to identify trends and patterns.
- Interactive graphs and 3D plots can be used to enhance the visualization of the data.
- The data can be combined with other sources to gain more insights.
For those of you who are going to be at ETH CC in Brussels next month, there is a little get together that we're putting on, and you can get register for that and itβll be talking all about Grant Fuckups, so come and share all your stories: The Good, The Bad and The Ugly of doing grants, and enjoy.
Stay tuned for more exciting updates as we continue to learn, share, and grow together in the Web3 ecosystem. π