Current public goods funding mechanisms struggle with scalability and fairness. As the number of projects grows, funders face an overwhelming burden of evaluating each project individually. This either incentivizes public campaigning, favoring well-known projects, or concentrates decision-making in small groups, leading to lobbying and inefficiency.
Pairwise is part of the latest funding mechanism to hit the scene: Deep Funding. Deep Funding, supported by Vitalik Buterin, allocates $250K to impactful Ethereum ecosystem open source dependancies using an innovative approach that combines Deep Graph with Deep Learning AI. This initiative unites leading Web3 organizations, including Eval Science, Pairwise, Drips, Open Source Observer, Gitcoin’s Allo Protocol, and VoiceDeck.
In this blog, we’ll explore:
What Deep Funding is
How it benefits the community
How Pairwise is involved
How you can get involved
Deep Funding is built on two core elements: Deep Graph and Deep Learning.
Most public goods funding mechanisms evaluate projects individually, often overlooking the broader, interconnected network of contributions that amplify their impact. Deep Graph changes this by analyzing how foundational infrastructure, collaborative synergies, and dependencies contribute to the Ethereum ecosystem’s overall value.
This approach highlights the "networked impact" of projects, demonstrating that true value often emerges from the dynamic interplay of multiple efforts working in harmony—like pieces of a complex puzzle.
Deep Learning leverages advanced AI models to assess these connections and allocate funds effectively. Think of it as the "engine" driving the funding mechanism, with regular "checkpoints" provided by expert decision-makers acting as the "steering wheel" to ensure human alignment.
The first pilot of Deep Funding focuses on solving a critical challenge: Which open-source dependencies (OSS) of Ethereum should receive funding, and at what level? Backed by $250,000 in prizes, this pilot is structured to evaluate and reward the most impactful contributions to Ethereum’s ecosystem.
How the Pilot Works:
Mapping Ethereum’s Dependencies:
A deep graph has been created to map Ethereum’s 40,000 open-source dependencies and their interconnections.
Nodes represent individual dependencies, and edges highlight how they relate to one another.
Allocating Weights:
A market of AI allocators assigns weights to the graph, determining the relative importance of each dependency.
The weights answer key questions like: “What is the relative impact of dependency A and dependency B on [project]?”
Spot-Checking by a Jury:
A jury of experts validates the proposed weights by performing spot-checks on random parts of the graph.
This process ensures that funding decisions are aligned with human judgment while leveraging the scalability of AI.
Distributing Funds:
The model that aligns most closely with the jury’s spot-check preferences determines the final allocation of funds.
Deep Funding addresses several limitations of traditional funding mechanisms, including:
AI models reduce human bias, time consumption, and knowledge gaps, ensuring fairer assessments. Projects no longer need to allocate resources to marketing or awareness campaigns, as funding decisions focus solely on their impact.
Deep Funding uncovers underappreciated yet vital contributions, such as cryptographic projects that form the backbone of Web3 infrastructure. These initiatives often remain hidden from mainstream recognition despite their critical importance.
By analyzing all available data, Deep Funding ensures a hyper-aware, data-driven approach to resource distribution—bringing inclusivity and precision to funding decisions.
December 12th: Data on 40,000 Ethereum dependencies released for model building.
January 20th: Sample spot-check data provided by jury members for model training.
January 20th: Early bird prize submission deadline for open-source models (50% reserved for early bird submissions).
February 20th: Model submission deadline.
February 27th: Results announced.
These deadlines incentivize early contributions while providing ample time for refinement and submission.
Pairwise plays a pivotal role in the human “checkpoint” aspect of Deep Funding, helping jury members compare repositories.
Jury members will assess questions like: "What is the relative impact of the developer contributions on [project]?"
Results will be a reference point for when jury members spot check AI models and choose the model with the closest to the human approach.
Evaluating projects on a long list is overwhelming. Pairwise simplifies the process by presenting users with side-by-side comparisons, enabling quick, relative judgments. This method reduces cognitive load and transforms governance into an engaging and straightforward experience.
Pairwise has proven its effectiveness in Retro Funding 6 on Optimism, where 2,660,000 OP (out of a total 3.5M OP) was allocated using the tool. Initially tested in RetroPGF 3, Pairwise has continuously improved through feedback from each funding round. In Retro Funding 6, features like liquid democracy were introduced, enhancing decision-making flexibility.
We’re incredibly grateful to Devansh and the Deep Funding team for inviting Pairwise to play an essential role in advancing capital allocation in Web3.
Contributors can win a share of the $250K prize pool provided by Vitalik:
$170K: Funding Ethereum GitHub repos based on edge weighting by the winning model.
$40K: Prizes for models aligning with jury spot-check evaluations.
$40K: Prizes for innovative open-source models, with early bird rewards.
Explore Deep Funding in detail at deepfunding.org.
Pairwise is redefining governance by making decision-making intuitive and enjoyable. Want to integrate Pairwise into your governance processes? Contact us here.
Deep Funding represents a groundbreaking evolution in funding impactful projects. With its combination of Deep Graph, Deep Learning, and guidance from Vitalik, this initiative is set to revolutionize how we allocate resources in Web3. Don’t miss your chance to be part of the future!
