Research Overview

We published the Gradient Papers to show our work -- every design decision in the protocol traces back to these documents. Citrate's protocol design is grounded in a series of nine research papers known as the Gradient Papers. These papers span the full arc of the project's vision, from foundational AI-blockchain integration theory through consensus mechanisms, tokenomics, cooperative structures, and constitutional governance. This page provides a guide to each paper and links to additional research resources.

The Gradient Papers

The Gradient Papers are numbered I through IX and follow a spectral gradient from sage (protocol specification) to lavender (biological foundations), reflecting the project's progression from technical foundations to design principles.

Paper 0: Series Index and Glossary

The index paper provides a comprehensive glossary of terms used across all Gradient Papers and a reading guide for navigating the series. It defines canonical terminology for concepts like Paraconsistent Consensus, blue score, LoRA adapters, and the Mentorship Protocol. Start here if you are new to the Citrate research literature.

Key contents: 52 glossary terms, 55 cross-references, 4 recommended reading paths

Paper I: Citrate Technical Architecture

Color: Sage (#7EC8A0)

The primary technical reference for the Citrate protocol. This paper covers the full stack: GhostDAG block structure, Lattice Virtual Machine, AI precompiles, Model Context Protocol, bridge architecture, and the client software design. It is the most comprehensive single document for understanding how Citrate works at an implementation level.

Key concepts: GhostDAG, LVM, precompile design, MCP architecture, bridge protocol, client internals

Paper II: Paraconsensus -- A Paraconsistent Approach to Consensus

Color: Periwinkle (#6B9AC4)

Paraconsensus proposes a novel consensus mechanism that tolerates contradictory states without system failure. Drawing from paraconsistent logic (a branch of logic where contradictions do not trivialize the system), the paper describes how Citrate's BFT layer can process conflicting information from AI models and human validators simultaneously, resolving disagreements through structured deliberation rather than forced majority rule.

Key concepts: Paraconsistent logic, contradiction tolerance, BFT with deliberation, meta-model consensus

Paper III: The Mentorship Protocol

Color: Aquamarine (#7EC4CF)

This paper describes Citrate's approach to network onboarding and knowledge transfer. The Mentorship Protocol pairs experienced operators with newcomers, creating economic incentives for teaching. Mentors earn bonuses when their mentees achieve operational milestones, aligning individual incentives with network growth.

Key concepts: Onboarding incentives, knowledge transfer, mentor-mentee bonding, milestone rewards

Paper IV: Behavioral Issues in Decentralized Systems

Color: Wisteria (#9B7FC4)

This paper examines the behavioral economics of decentralized networks, focusing on incentive misalignment, free-rider problems, and the unique challenges that arise when AI agents participate alongside human actors. It establishes the theoretical basis for Citrate's incentive design, including the Mentorship Protocol and reputation scoring.

Key concepts: Game-theoretic analysis, AI agent incentives, reputation systems, behavioral penalties

Paper V: ATIS -- AI-Transparent Inference Serving

Color: Mint (#6ABFA0)

ATIS defines the transparency and verifiability framework for AI inference on Citrate. The paper introduces the three verification tiers (Signature, Optimistic, ZK-SNARK), the attestation protocol, and the model card requirements that ensure inference consumers can assess the trustworthiness of AI outputs.

Key concepts: Verifiable inference, attestation tiers, model cards, inference transparency

Paper VI: Memetic Money -- Token Economics for AI-Native Networks

Color: Chamois (#D4A76A)

Memetic Money develops the economic theory behind SALT tokenomics. The paper argues that AI-native networks require a different approach to token design, where the token captures value from both computational work (mining, inference) and cognitive work (model training, adapter composition). It describes the supply schedule, fee burn mechanism, and staking economics.

Key concepts: AI-native tokenomics, computational value capture, fee burns, staking game theory

Paper VII: The Mozi Cooperative

Color: Peach (#E8A87C)

Named after the ancient Chinese philosopher Mozi (who advocated for universal welfare), this paper describes Citrate's cooperative economic model. It introduces the $SNAP NFT mechanism, the DAO treasury structure, and the cooperative governance model that distributes protocol revenue across all participant classes.

Key concepts: Cooperative economics, $SNAP NFTs, ERC-6551, treasury management, revenue sharing

Paper VIII: The BR1J Constitution

Color: Rose (#E88B8B)

The final paper establishes the governance constitution for the Citrate network. It defines participant rights and responsibilities, the amendment process, the ethics framework, and the relationship between human governance and AI-assisted decision-making. The BR1J Constitution is the living governance document that evolves through the amendment process described in its own provisions.

Key concepts: Constitutional governance, AI ethics, amendment mechanisms, distributed sovereignty

Paper IX: The Medusa Paradigm

Color: Lavender (#B497D6)

The Medusa Paradigm introduces the foundational thesis: that AI and blockchain technology are not separate domains to be bolted together but are fundamentally complementary systems. The paper draws on the biological metaphor of the medusa (jellyfish) -- a decentralized organism with distributed intelligence -- to argue for a new class of blockchain where AI is a first-class citizen of the consensus layer.

Key concepts: AI-native architecture, distributed intelligence, the case against "AI as a service" bolted onto existing chains

Benchmarks

Benchmark reports for the Citrate protocol are published as the network progresses through testnet and mainnet milestones. These reports cover:

Consensus throughput: Transactions per second under various validator set sizes
Inference latency: End-to-end latency for AI precompile calls across verification tiers
GhostDAG performance: Block production rates, blue/red ratios, and finality times
Bridge throughput: Cross-chain transfer speeds and attestation latency

Benchmark reports will be published here as they become available.

Audit Reports

Security audits of the Citrate protocol, smart contracts, and bridge infrastructure are conducted by independent firms and published for community review. Audits cover:

Core consensus implementation
AI precompile contracts
Bridge and oracle contracts
Governance and treasury contracts
$SNAP NFT and ERC-6551 implementation

Audit reports will be published here as they are completed.