A brief analysis of Deep Funding: Using AI to reshape the future of public goods fund allocation

In the blockchain and open source fields, efficient fund allocation has always been a problem. Today, an innovative project called Deep Funding is trying to use artificial intelligence and decentralized review to solve this problem. This project, supported by an initial capital of US$250,000 from Vitalik Buterin, not only plans to solve the resource allocation problem in the current Ethereum ecosystem, but also creates a new model for the future allocation of funds for public goods.

01. Deep Funding

What is Deep Funding?

Deep Funding is an innovative project that optimizes the allocation of public goods funds through AI and a decentralized review mechanism, aiming to solve the problem of inefficient resource allocation in the Ethereum ecosystem. The project goal is to build a fair, transparent, and efficient fund distribution system to support Ethereum and the key open source projects it relies on to achieve long-term sustainable development.

Official website: https://deepfunding.org/

What problem do you want to solve?

Currently, there are the following problems with the allocation of public goods funds in Ethereum:

1. The irrationality of human decision-making: When faced with complex and abstract problems, it is often difficult for humans to make reasonable judgments.
2. Favoring surface-level projects: Election-based funding mechanisms tend to fund projects that are superficially obvious, while ignoring deep technical dependencies and complex contributions.

This results in some "hidden" infrastructure that is critical to the Ethereum ecosystem not receiving adequate support, and may also waste resources on projects that may seem important in the short term but have limited long-term value.

What kind of thinking is used to solve the problem?

Solutions proposed by Deep Funding include:

1. Build Deep Graph

Deep Graph is a dynamic dependency graph that displays dependencies between projects and assigns weights to each dependency. In this way, the contribution and actual value of public goods can be visualized, solving the problem of "invisible contributions" that are difficult to measure.

2. AI model weighting and evaluation

• Data input: Various information based on open source projects (such as number of stars, contributor activity, update time, etc.). This requires using your imagination and understanding of the value of open source projects.
• Weight allocation: The AI ​​model allocates weights based on the importance and actual impact of dependencies, and dynamically adjusts fund allocation.
• Verification and optimization: The jury conducts spot checks on the model to ensure the rationality of the weights.

3. Jury review mechanism

• The jury consists of experts who provide training data for the model by answering questions such as "Which is more important, project A or B?" This type of question was chosen because it is easier for humans to identify and answer.
• Cooperation model between humans and AI: humans are responsible for direction and value judgment, and AI provides data analysis support. After that, multiple models that are good and more in line with human consensus will be selected for application.

4. Distribute funds equitably

Funds are allocated according to the contribution ratio of the project, and there will also be some incentives for the winning models.

Deep Funding will not only be used for weight construction and distribution of open source software, this model will be used in any scenario with dependencies and distribution. For example: papers, music, film and television works, etc. Open source software is just an initial attempt, and Deep Funding hopes to turn it into a solution suitable for various scenarios.

02. Deep Funding Competition

Currently, Deep Funding's first competition focuses on GitHub repo and open source projects. It constructs a weighted Graph through the dependencies of open source projects to obtain the donation amount that each warehouse should receive. Then focus on the open source projects under the Ethereum label, especially the client.

Current progress of the Deep Funding project includes:

1. Sponsorship and Funding: Vitalik Buterin provided an initial sponsorship of $250,000.
2. Data preparation: Collect Ethereum dependency graph, involving data of approximately 40,000+ edges. It's ready now.
3. Mechanism design: Launch an AI model competition (will be held on Kaggle), and are currently recruiting AI models.
4. Pilot evaluation: Verify the effectiveness of the model through jury spot checks; apply the dependency weight model to Ethereum-related projects and see the actual effect.

Of the 250K bonus, $170k will be allocated to projects based on the weight of the dependency graph, $40k will be awarded to the model with the best performance in the review spot check, and $40k will be awarded to open source submitted models. The innovativeness of these models will be reviewed by experts. Decision will be made after evaluation by the group.

There are still many challenges that need to be addressed

1. Jury fairness and incentive mechanism: How to ensure the neutrality and long-term participation enthusiasm of the jury? How to build a fair and effective jury?
2. AI model effectiveness: How to accurately weight deep dependencies to avoid model abuse or gamification?
3. Dynamic adjustment mechanism: How to balance self-assessment and external review to avoid bias?
4. Funding sources and incentives: How to attract more funds to participate in distribution, especially for non-code contributions?

We will discuss and explore step by step.