This document is designed to give the Graph Foundation an insight into our technological capabilities and how we can align. Our vision is to move towards a network that builds out decentralised approaches to artificial intelligence that hybridises human collective intelligence and AI capabilities, by leveraging decentralised governance and token based incentive schemes.
A decentralised knowledge graph mechanism has been a core component of this vision for a long time. It is a crucial component and we believe we can provide a pipeline of high quality validated knowledge to Geo Browser and synergise Factory Labs development trajectory with The Graph's.
There are a few important contextual issues that I think make what we're working on here relevant, important and timely:
Generative AI is making content production essentially limitless. The internet is filling up with 'slop', amplifying the need for high quality content curation.
This same effect is limiting the quality of training data for AIs. Standard scraping techniques are leading to less effective, but more expensive, training runs.
The 'war for truth' is heating up, with centralised AIs being pressured to shape inference outcomes towards policy goals, ideological outcomes and shaping inference towards their own benefit. We need a decentralised and open data commons.
However, the AI x Crypto market remains wide open, with trivial agent + memecoin mechanisms dominating the market.
This is a huge opportunity, but the window is closing fast.
Factory Labs is a London based Research and Development company focused on the intersection of AI x Crypto. Factory Labs will have an associated DAO, (FactoryDAO), that will operate as a hybrid decentralised organisation with the Labs entity.
As an entity, we embody a values driven thesis, opting to take a slower pace and more difficult path if it means we can launch systems that are non-extractive, positive sum and move the industry towards systems that benefit humanity.
We have a suite of double audited smart contracts and front ends that are ready for launch and can be combined as modular primitives to build out custom dApps. We can quickly iterate into production interventions and can realise impactful outcomes quickly (in weeks).
There are four co-founders of Factory Labs.
Dr Nick Almond - CEO, PhD in Physics, Expert in Governance and Organisational Design, Token Economist and Mechanism Designer.
Matt Haynes - Head of Coordination and Ops, Computer Scientist, Recruitment and Project Management expert.
Naomi Dara - Creative Director and Designer, crypto UX expert with 10 years of crypto experience.
Chris Smith - CTO, cryptographer and blockchain architect, 10 years of crypto experience, early Ethereum developer.
We leverage a scalable development footprint with developers we have worked with for years and operate with effectively. We can scale up and move quickly, budget dependent.
Our current roadmap is:
#VOICE is designed to be an application that incentivises people to create and curate data in specific knowledge domains. It is ideal for capturing community sentiment, raw insights and facilitating the curation of data towards validated agreement. It is designed to provide the data pipeline for community controlled domain specific agents.
The #VOICE app is 80% completed and needs further funding to finalise an effective onboarding flow that is suitable for non-crypto and crypto native participants.
It can be deployed with 3-4 weeks of development time. It will produce new data sets, which we can then build flows for into the GRC20 standard, connecting the standard to our hypergraph based data structure.
Our intention is to build out a series of "London VOICE" interventions, iteratively improving the application, using the application itself for user research and feedback.
Suggestion: We hone the app in London and then support Geo Browser in enhancing the San Francisco Space.
The data that #VOICE generates is a semantic hypergraph, with nodes as identity and meaning, with edges as weighted preferences generated by curatorial voting.
FACT is our end game target, likely an appchain or L2 with a governance system that can evolve the network towards an approach for building emergent collective intelligence.
We call our hypergraph structure an "opinion graph". It is deliberative, dynamic and is primarily designed to capture raw insight and preferences.
The "FACT Graph" is the output of deliberative decision making, voting and governance. We think this could be where GRC20 comes in.
Voice uses an iteration of quadratic voting we call Semantic Ballot Voting. This generates our data and supports in the open curation of knowledge.
Moving beyond this mechanism, we can weight voting power individually to identities within the system, moving from "attention to influence" as a core dybamic for structuring knowledge.
We have developed a design for paying participants in tokens for contributing data to a shared data commons. It utilises n-player peer review, probalistic payouts and longitudal reputation based payout profiles.
$50k USD Grant: First wave development of #VOICE app, new data sets and experimentation with GRC20.
$100k USD Grant: full scale deployment across London with a view to be ready for deployment in San Francisco. Includes event co-marketing and "powered by the Graph" branding.
$250k USD Grant: On chain reputation weighted voting module for integration into Geo Browser via Aragon OSx. Token based incentive mechanisms and the Crypto #VOICE intervention.
$500k USD+ Investment: Via a SAFE in Factory Labs, to build out the above and build a strategic long term alignment.