[ ETERNAL BEINGS / DOCS / SPEC v0.1 ] CONCEPTUAL REFERENCE // CHAIN-AGNOSTIC
Protocol Research Document / 001

Autonomous On-Chain
World Protocol

Eternal Beings is a proposed world factory and deterministic state machine. A creator submits a seed, rules, and economic parameters. The Factory deploys an independent World; that World issues Beings and processes resources, encounters, mutation, fusion, and evolution under verifiable constraints.

[STATUS] This document is a technical design draft aligned with the current concept site. Chain, contract addresses, audits, and measured gas costs are not yet specified. Numerical values are design ranges, not deployed facts, financial promises, or security guarantees.
01 / Protocol Thesis

Rules define the world. Ownership is the entry point.

Conventional NFTs often treat a token as a static media pointer. Eternal Beings treats a token as access to an evolving state. The visible image is one projection; the durable asset is its verifiable history, world rules, and path through the shared simulation.

Determinism

Identical seeds, parameters, and historical inputs must reproduce identical states.

Sovereignty

Each World has independent supply, resources, interaction rules, and economic boundaries.

Composability

Standard interfaces expose ownership and state to markets, games, and other protocols.

02 / Architecture

Modular contract architecture

CreatorWorldFactoryWorld RegistryBeing NFTEngine Modules
ModuleResponsibilityRecommended constraint
WorldFactoryValidate configuration, deploy Worlds, register versions and implementation hashes.Predictable CREATE2 addresses; allowlisted templates; complete deployment events.
WorldStore the rule root, epoch, resource totals, and engine addresses.Accept transitions only from authorized modules.
BeingTokenERC-721 ownership and on-chain metadata interface.Transfers must respect locks, staking, and cooldown state.
Genesis EngineDerive initial genes, traits, rarity, and rendering instructions.No single transaction participant may control entropy.
Evolution EngineProcess gathering, mutation, fusion, encounters, and hunting.Every transition has prerequisites, limits, cooldowns, and events.
RendererEncode state as SVG, JSON, or another on-chain representation.Renderer failure must never freeze core ownership.
03 / State Machine

Being and World state vectors

Bᵢ(t) = [gᵢ, eᵢ(t), rᵢ(t), aᵢ(t), cᵢ(t), hᵢ(t)]g: immutable genes; e: energy; r: resource vector; a: age/epoch; c: cooldowns; h: history commitment.

Global World state is W(t)=[P,R(t),N(t),E(t),H(t)]: immutable parameters, resource pools, active population, epoch, and accumulated event commitment.

Bᵢ(t+1), W(t+1) = T(action, Bᵢ(t), W(t), proof)
require C(Bᵢ, W, action) = true

T is the transition function and C is the constraint set. Each action should use a separate nonce to prevent replay. Cross-module execution should use checks-effects-interactions and reentrancy protection.

04 / Deterministic Genesis

From seed to reproducible genes

seedWorld = keccak256(contentHash ∥ configRoot ∥ creator ∥ salt)
entropyᵢ = H(seedWorld ∥ randomness ∥ tokenId ∥ mintNonce)
traitᵢ,k = Fₖ(entropyᵢ, distributionₖ)

The media seed is a content commitment, not randomness. Full on-chain claims require every generation input and rendering instruction to remain chain-readable, rather than merely storing a remote URL.

Entropy options

  • Commit-reveal is inexpensive but must handle non-reveal and last-revealer bias.
  • Verifiable randomness offers stronger guarantees but adds oracle and callback dependencies.
  • Batch entropy lowers cost but must preserve provable fairness across token IDs.
05 / Mathematical Models

Traits, activity, and evolution

Trait distribution

pₖ = exp(zₖ/τ) / Σⱼ exp(zⱼ/τ), where zₖ = wₖ·φ(seed)+bₖLower temperature τ concentrates rarity; higher τ flattens the distribution. On-chain implementations require fixed-point approximations.

Energy and renewable resources

Eᵢ(t+1)=clamp(Eᵢ(t)+αGᵢ(t)-βAᵢ(t)-δΔt,0,Emax)
Rₖ(t+1)=Rₖ(t)+ρₖ(1-Rₖ(t)/Kₖ)-Σᵢqᵢk(t)

The second equation is a discrete logistic regeneration model. Use lazy accrual at interaction time instead of block-by-block updates.

Rarity score

rarity(Bᵢ)=Σₖωₖ[-log₂(max(freq(traitᵢ,k),ε))]+λ·novelty(historyᵢ)

Rarity is a display metric, not a price. Dynamic frequency changes over time, so every score must publish its version and snapshot block.

06 / Resource Economy

Closed-loop resources and inflation control

Sources

Gathering, exploration, quests, encounters, and bounded cross-world imports.

Sinks

Mutation, fusion, repair, acceleration, access fees, and governance submissions.

Velocity

Cooldowns, escalating costs, and account/Being rate limits prevent extraction loops.

costₙ = c₀(1+γ)ⁿ·congestion(t)
congestion(t)=1+η·max(0,u(t)-u*)²

Marginal action cost rises with repeated use and above-target utilization. Every parameter needs a hard bound so governance cannot make interaction impossible. Resources are gameplay units, not guaranteed yield.

07 / Interaction Engines

Mutation, fusion, and hunting

Mutation

P(mutation)=σ(θ₀+θ₁·exposure+θ₂·resource−θ₃·stability)

A piecewise-linear sigmoid is practical on-chain. Every mutation stores parent state hash, mutation type, and entropy proof.

Fusion

g_child = mask⊙g_A + (1-mask)⊙g_B + mutationNoise
cost = base + κ·similarity(g_A,g_B) + scarcityPremium

Rules must specify whether parents remain, lock, or burn. Both owners must authorize fusion unless one address owns both assets.

Encounter

P(A wins)=1/(1+10^((rating_B-rating_A)/s)) · terrain · stamina

Risk actions should lock assets before settlement. Asynchronous randomness requires timeout, retry, and refund paths.

08 / Parameter Reference

Suggested starting ranges

ParameterStarting rangeDesign effect
maxSupply1,000–10,000Must be fixed after deployment.
mintPriceDerived from chain cost and access targetNever implies secondary-market value.
epochLength1–24 hoursShorter improves feedback but raises transaction load.
energyDecay0.5%–3% of Emax per dayEncourages activity without making recovery impossible.
mutationRate0.1%–2% per eligible actionNeeds caps, cooldowns, and outcome protection.
fusionCooldown1–7 epochsLimits recursive batch fusion.
resourceCap KCalibrated to active populationSimulate at 95th-percentile load.
09 / Security

Threat model and launch gates

Entropy manipulation

Use VRF or penalized commit-reveal; do not rely on block data alone.

Economic attacks

Snapshots, delays, rate limits, and holding periods reduce flash and Sybil attacks.

Gas and denial of service

No unbounded loops. Use pagination, pull settlement, and lazy accounting.

  • Invariant and fuzz tests: capped supply, conserved resources, monotonic nonces, and valid ownership transitions.
  • Independent audit, public testnet, bug bounty, and incident response plan.
  • Publish bytecode hashes, compiler versions, and reproducible build instructions.
10 / Governance

Immutable core, optional extensions

Genesis parameters, supply cap, gene definitions, and state history should be immutable. Interfaces, indexers, and optional gameplay modules can remain replaceable adapters. Any governed change should obey boundaries and a timelock.

P_new ∈ [P_min,P_max] ∧ delay ≥ T_exit ∧ codeHash ∈ auditedImplementations

A World creator must never be able to seize user assets, rewrite rarity arbitrarily, or bypass resource conservation.

11 / Research

Long-term protocol directions

ZK State Compression

Execute complex simulation off-chain and prove the resulting state transition.

Agent Governance

Autonomous agents operate only within verifiable permissions, budgets, and termination rules.

Inter-world Portals

Capability tickets export selected traits while the source World preserves ownership truth.

Implementation should begin with the minimum verifiable path: Factory → World → Being Mint → deterministic metadata. Add resources, mutation, and fusion only after simulation and security testing.

OPEN DEPLOYMENT AND MINT GUIDE →