a3na GCP
Enhancing Gitcoin’s Knowledge Base and Citizen Engagement
Summary
a3na (Consensys) proposes to unlock funds for Gitcoin Citizens to be rewarded for content that they contribute by leveraging a developing protocol named a3na. We also wish to work closely with Gitcoin to ensure that what we are building is for communities like Gitcoin, helping us to make informed decisions along the way about our application design, architecture decisions, smart contract design, and tech stack in order to ensure that Gitcoin’s knowledge base and community contributions are best positioned for scalable growth and usefulness.
Abstract
a3na is a new concept by the team behind the community support product VillageDAO (Consensys) which looks to reward content contributors for their contributions to a shared, community knowledge base. Our research has found that communities find it difficult to keep their knowledge bases up-to-date, which results in gaps. Additionally, individual community member’s knowledge is not optimally utilised inside of a given community. The recent development of LLMs means that knowledge bases have become more valuable, as LLMs are able to deliver community knowledge to individuals at scale.
Inside of Consensys, we built a proof-of-concept, an LLM bot which references a specific knowledge base only (when answers do not reside in the KB, the LLM responds to say there is no answer available). Once one (or more) articles are referenced, a token is passed to the wallet of the contributor(s). Through this method, members of a community can be incentivized to contribute knowledge to a community knowledge base.
By expanding on and further developing this proof-of-concept we aim to revitalise the Gitcoin knowledge base (support.gitcoin.co) and provide a system for ensuring its sustainable long-term growth and management. We are excited to work with @Sov on this project given his expertise and familiarity with the Gitcoin ecosystem landscape and knowledge base.
Demo of the proof of concept built during the hackathon.
Benefits & Motivation
Maintaining a repository of knowledge for individuals to reference and use is a key component to enabling the successful growth of an ecosystem. However, it can come at a high cost in terms of time and effort when it is owned by only a select few. This proposal addresses this cost by providing a means for the community to contribute its collective knowledge for the benefit of the community, fostering ecosystem knowledge in a scalable way, and incentivising community efforts in return.
The benefits are multiple fold;
-
Gitcoin will now have a way to catalogue and access all of the knowledge in the community. Which can include support use cases and beyond.
-
Many of the conversations happening in support can be automated.
-
We wish to create a legal framework for the content contributor (Gitcoin Citizen) owning their own content, which means that they will put more effort into the content which results in better content being added
-
Any content that a Gitcoin Citizen publishes which is applicable to other ecosystems can be bootstrapped in those other communities, if they use this solution.
-
Which also means that any content referencing Gitcoin, can be propagated in those communities too, which is good for Gitcoin’s exposure.
-
Humanity will benefit, as this is a model for paying people for their knowledge, as opposed to training LLMs with stolen content.
Project Scope
The proposed scope of the application and solution is planned to encompass the three experiences described below:
- Support seeking user experience
In the first instance we plan to implement the service as an API that could be integrated with tools such as Intercom. The result shown to the user is an answer from the LLM with references to any articles, and their authors. Ideally the references will be links to the full article.
- Content contributor user experience
The content contributor will be able to log into an a3na application (connect wallet) where they can publish content they believe to be useful and subsequently rewarded for it.
- Content curator user experience
The content curator will have the ability to select individuals (wallets) and/or specific content and associate them to specific topics of a knowledge base, so they can own a domain or “library”. Additionally, by leveraging keywords generated from the content submitted, we can expose gaps in coverage of the content being leveraged.
In the first instance, the end user (Support seeker) experience will be delivered as an API, planned for Intercom integration or other similar tools. After developing this initial concept, we plan to consider a more robust user experience for querying an LLM and browsing content which is indexed.
Project Breakdown
Below is a high level overview of how the system will function for contributing and using content provided by the community:
Content Contributor experience eg Gitcoin Citizen
We are striving to design and implement a user-friendly experience for uploading and submitting content for community members. This will ensure that community members can provide useful information to answer the questions of other community members by providing a display of their submitted content and the rewards earned towards each piece of content contributed.
Content Curator experience eg Gitcoin Community Manager
The goal of the curator experience will be to provide a simple and clean means of browsing contributed community content to select and build out a robust set of libraries or collections of content that are targeted towards answering and/or documenting specific domains. Additionally, we want to explore leveraging generated keywords from the contributed content to provide a breakdown of areas that are under/overrepresented in the content being curated.
Content usage
Once the content has been curated, we intend on providing it via an API that will process the data and deliver it in a consumable manner for tools such as Intercom, LLMs, and, ideally in the near future, an embeddable search that can be plugged into various experiences.
File/content storage
We are investigating onchain as well as offchain file storage solutions and will need to consider cost. We can likely allow for intermittent onchain file publishing or some combination of solutions between key data onchain and more costly data offchain.
LLM design
We have previously used OpenAI service to build a proof of concept, but wish to explore more web3 aligned LLM services in our tech stack, such as the Venice.ai API, for privacy and censorship resistance. These services are pretty simple to swap out too.
Token/Rewards design
We are considering a two token model, where a citation token or attestation is sent to a content contributor each time their content is cited. This is then used as a calculation for payment (which can happen in a communities own token, on that communities chain, in this case $GTC on Allo).
Risks
We have identified three potential risks to the success of this project in detail below:
-
Usage and adoption
This can be mitigated by ensuring easy integration with the existing experience. -
Content moderation
This will be mitigated by the Content Curator role who will serve as the moderator and filter for content being selected for usage. In the future we would like to explore more automated means of filtering content contributed to a3na. -
Sybil resistance/spam attacks
There are a number of methods that can be employed (see below).
One note on sybil resistance / spam attacks, is that we can use many different ways to determine whether a user is a bot or genuine. These methods range from bot detection (e.g. cloudflare in use on perplexity.ai, or discord bots to detect real users ) to even the possibility of a kind of “connect wallet to ask the LLM experience”. Different clients of the protocol will have different appetites to prevent bot attacks from earning citation tokens. However, we will be cautious and monitor the activity, and act accordingly.
Also, we can plan the rewards such that they reward better behaviour, e.g. the possibility of future rewards will be removed from bad actors, separation of citation and rewards (i.e. payment for citation).
Additionally, we have prevented any sybil attacks inside of VillageDAO clients due to recruiting trusted community members.
Budget
[Below from the RFP]
The recommended total budget for the enablement of usage and adoption of this platform is 20,000 GTC. The recommended breakdown of monthly allocations is as follows.
Monthly Allocation
1,666.67 GTC per month (20,000 GTC / 12 months)
Breakdown
- Content Contributor Compensation: 15,000 GTC
- Allocated for community members contributing high-quality content
- Distributed based on contribution quality and usage metrics
- Monthly pool: 1,250 GTC
- Technical Development and Maintenance: 3,000 GTC
- For ongoing development, updates, and technical maintenance of the knowledge base platform
- Monthly allocation: 250 GTC
- Community Engagement and Moderation: 1,200 GTC
- To support community managers and moderators overseeing the knowledge base
- Monthly allocation: 100 GTC
- Incentives and Rewards Pool: 800 GTC
- For special initiatives, contests, or bonus rewards to drive engagement
- Can be distributed as needed throughout the year
Additional Considerations
- LLM Usage Costs: To be billed separately based on actual usage
- Recommend setting aside a separate budget for this, not included in the 20,000 GTC
- Flexibility: The allocations above can be adjusted based on actual needs and performance throughout the year, ensuring optimal use of the budget.
- Reporting: Monthly reports will track GTC distribution, content creation metrics, and overall knowledge base performance to ensure the budget is used effectively.
Milestones
We present below an outline of the four major milestones we are planning to hit in order to achieve the initial version of this project and their descriptions. Currently, we estimate achieving these milestones over a time period of six months.
Milestone 0: Design and Content Review
Description: We will work closely with Gitcoin to map out the existing content library, and understand a framework for closing the gap between the missing, and outdated content, and adding new content. Content is a living thing! As a result, the system which keeps the content up to date with the workings of Gitcoin will be important. Through a3na (rewards and citations) we hope to enable a living knowledge base for Gitcoin. Something which many LLM systems struggle with, is how to keep the references up to date.
Milestone 1: View Content Library
Description: We will build a system for storing and retrieving contributed content to a3na that will ensure content ownership is verifiable and lay the groundwork for the rewards and citation system.
Milestone 2: Content Contributor Experience
Description: Completion of this milestone will enable Gitcoin Citizens to begin contributing content to a3na for usage by Content Curators, building the body of community knowledge.
Milestone 3: Content Curator Experience
Description: With the foundation in place for community sourced content, achievement of this milestone will mean that Gitcoin Content Curators will be able to browse and select contributed content for use in their knowledge base via API, enabling integrations into existing systems (e.g. Intercom) and LLM usage.
Milestone 4: Content Usage Rewards
Description: A smart contract, which, when implemented, means each time a content contributor’s (Gitcoin Citizen) content is referenced, rewards are calculated for distribution.
Acceptance Criteria
-
Knowledge Base review and gap analysis: As part of the initial review of the current state of the Gitcoin knowledge base, the gaps in information will need to be surfaced and communicated.
-
Interface for Contributors and Curators: The platform should have an intuitive interface for community members to contribute content and for Gitcoin to browse and curate which content is relevant for the knowledge base based on needs and quality.
-
Content usage mechanism and rewards: The platform should enable Gitcoin to be able to reference and leverage the curated content by means of a permissioned API that can be integrated into existing tools.
-
Onboarding and training materials: To ensure smooth migration and adoption of the platform, onboarding and training materials should be published and provided to the community.
Relevant Metrics
-
Number of content contributed: Track the total amount of articles contributed to the platform to assess level of engagement and variety of content provided by the community.
-
Number of content referenced: Track the total number of times individual articles are referenced (through distributed citation tokens) when providing users solutions to their questions/requests for information to ascertain the level of usage.
-
Number of rewards earned: Track the total amount of rewards being earned for the usage of community content to ensure appropriate amounts of rewards are being distributed for the community’s efforts.
-
Content gaps identified: Published information that highlights the domains where knowledge is out of date and/or lacking so that it can be proactively addressed.
Team
We are a team situated within the Customer Success department at Consensys who have been building web3 support tools to improve the overall user experience in the ecosystem for the past two years. Our current products include 0xplain, a tool designed to improve the readability of transactions and relevant content for users. We also built activity.metamask.io (which leverages 0xplain API) and VillageDAO, a platform that enables communities to reward their members for solving user problems.
Conclusion
We are very excited about this project and the opportunity to enable the Gitcoin community to own and expand their collective knowledge. This project represents a continuation of our efforts to create a web3 suite of tools that fully enable communities to empower themselves and create a more sustainable, transparent and open ecosystem for all to engage with.
We welcome any feedback, questions, and suggestions. The input provided by Gitcoin is key to guarantee that the project appropriately addresses the needs of the community and meets expectations. All of this proposal is up for discussion, if there is any part that does not fit with your needs, we would love to know and iterate!
FAQ
- What is the model for a3na to charge money?
Currently, we are looking to take a % of the token rewards which flow through the system, about 7% to begin with. But we will work with you, and hopefully more communities, to understand how this price discovery works best for everyone.
- Do Intercom have a similar feature?
Intercom have Fin, an AI chatbot which can provide an LLM service. However, the Intercom model assumes that there is a centralised entity maintaining a knowledge base of answers for the bot to draw upon. The a3na concept is a way for a community of individual contributors to earn rewards for their contributions to a collective knowledge base, and see their citations onchain. Also, we will build an integration with an LLM service in order to consume the knowledge base.
