The Gradient Papers
A nine-part working dissertation on the architecture, economics, governance, and biological foundations of the Citrate Network.
These papers document how Citrate was designed and why. They cover the protocol specification, consensus theory, federated learning architecture, engineering methodology, analog hardware research, bridge economics, cooperative ownership structure, DAO governance, and the biological design principles that inspired the whole system. They are working papers. They distinguish between what is implemented, what is fully specified, and what is planned for future development. Where earlier versions contained fabricated metrics or unverified projections, the v2 revisions note the corrections explicitly.
Citrate Technical Paper
The complete protocol specification. GhostDAG consensus with k=18, BFT finality committee of 100 validators at 67% threshold, 10-block checkpoints, Lattice Virtual Machine with seven AI precompiles, MCP orchestration layer, and SALT tokenomics. This is the reference document. Every parameter used across all other papers traces back here.
RevisionParaconsistent Consensus
Formalizes how consensus and federated learning are unified in one protocol. Uses Belnap FOUR-valued logic to handle contradictory model outputs without discarding them. Defines the mentor-mentee learning protocol that runs at every BFT checkpoint. Specifies LoRA adapter rank 16, embedding dimension 768, and the slashing parameters for embedding manipulation.
RevisionThe Mentorship Protocol
Grounds the federated learning architecture in organizational learning theory. Documents how mentor selection works, how knowledge transfers are structured, and how the network avoids the mode collapse that plagues competitive learning systems. Maps directly to the mentor-mentee agent swarm in the CVM.
RevisionBehavioral Issues
The engineering methodology paper. Establishes Behavior-Driven Development with Gherkin specification as the required methodology for building agentic systems on Citrate. Every feature is specified before it is coded. Red tests come before green tests. This discipline is what keeps the agent swarm predictable at scale.
PublishedATIS
Analog Token Importance Scoring. A research proposal for using field-programmable analog arrays (FPAAs) to pre-filter transformer attention before digital compute receives it. If this works at scale, inference energy costs drop by orders of magnitude. This is the most speculative paper in the series and the most clearly labeled as such.
RevisionFollow ATIS Research ↗The Memetic Money Portal
The bridge architecture between Ethereum and Citrate. ERC-6551 NFTs with Token Bound Accounts give holders oracle node seats and perpetual fee revenue. The $SNAP fundraise hard cap is 2,000 ETH. Bridge oracle threshold is 2/3 consensus of registered nodes. Fee split is 70% to $SNAP holders, 30% to treasury.
RevisionThe Mozi Cooperative
Proposes cooperative ownership as the economic model for distributed AI infrastructure. Draws on Mondragon cooperative precedent and modern platform cooperative theory. Defines how infrastructure participants earn equity proportional to contribution rather than capital. The Mozi Cooperative is the institutional vehicle for school and community node operators.
RevisionMozi Cooperative ↗The BR1J Constitution
The DAO governance framework for Citrate. Establishes Wyoming DAO LLC structure, constitutional amendment process (50% quorum, 67% approval, 14-day timelock), human-AI symbiosis as a founding principle, and ethical AI development guidelines. The BR1J Constitution is the capstone document that references all prior papers.
RevisionThe Medusa Paradigm
The biological inspiration paper. Documents eight design principles derived from cnidarian species: Aurelia aurita (nerve net consensus), Physalia physalis (colonial modularity), Turritopsis dohrnii (state reversal), Tripedalia cystophora (distributed observability), Cassiopea (symbiotic compute), bloom dynamics (threshold scaling), strobilation pipelines (deployment lifecycle), and nematocyst defense (Byzantine fault tolerance).
RevisionCnidarian Foundation ↗Reading Paths
Engineers
Technical ... Methodology ... Consensus ... Mentorship ... Hardware ... Bridge
Software engineers and protocol developers
Researchers
Biology ... Hardware ... Consensus ... Mentorship ... Technical
Academic researchers and scientists
Investors
Technical ... Bridge ... Cooperative ... Governance ... Consensus
Investors and fund managers
Schools
Governance ... Cooperative ... Bridge ... Biology ... Technical
Schools and community builders