Constitutional Physarum/Lichen Biomes: Learning Network Physics Through Scarcity, Pruning, Mutualism, and Dispersal

Proposed arXiv categories: cs.MA / cs.AI Publication DAG node: P08, Wave C Dependencies: P04, P07 Claim register: P08-C1, P08-C2, P08-C3, P08-C4

A constitutional principle becomes behaviorally testable when it is embodied as a consequence participants must navigate, while the system separately measures whether the lesson transfers beyond the simulation.

The paper contributes a complete simulation contract, a separation between physics and governance, a Physarum-to-Lichen phase transition, a transfer study, and adverse tests for arbitrary gamification, selection effects, and narrative rationalization. It also treats the biome as one estate expression, not a universal model of social order.

FORKED is an internally owned and managed estate inside the Ubiquity Federation, Prompted LLC's flagship governance substrate. The estate inherits the Fractal Quivers of Quivers parent topology, production splat mechanics, center exclusion, receipt-bearing lawful traversal, lifecycle separation, and the rule that models propose morphisms while governance determines admissible motion [@Taylor2026FQoQ; @Taylor2026OpenCenter; @Prompted2026Ubiquity]. This paper studies one bounded expression of that parent architecture. It neither renames the parent substrate nor inherits universal applicability from it.

The evidence lanes are kept separate. A canonical or source artifact establishes what that artifact states. A component test establishes behavior under its recorded configuration. A synthetic campaign establishes behavior of the generator and analysis pipeline, not human, hardware, customer, field, or operational performance. Novel cross-domain claims are preregistered as hypotheses and remain open to null, adverse, localized, and disproving results.

Dependency freeze. Final freeze depends on P04, P07; draft work may proceed in parallel, but imported claim IDs and artifact hashes must be rechecked after each dependency freezes.

The interlattice is the citation and governance relationship among the parent FQoQ topology, production splat mechanics, the FORKED foundational architecture, the Native Range, the Experimental Program, and the companion papers in this DAG. Citations are therefore typed: a parent citation supplies inherited architecture; a native citation supplies component behavior; a synthetic citation supplies generator-level evidence; a companion citation supplies a versioned specialized argument. No downstream citation converts an upstream hypothesis into a universal fact.

This paper's claims are:

Modern AI systems frequently compress unlike objects into a common output surface. Facts, hypotheses, instructions, permissions, model predictions, runtime observations, and institutional decisions can all arrive as fluent text or a single status field. The resulting failure is architectural before it is linguistic: a local representation acquires authority that its source never possessed. A constitutional principle becomes behaviorally testable when it is embodied as a consequence participants must navigate, while the system separately measures whether the lesson transfers beyond the simulation.

HUNGER is a seven-tic scarcity simulation and constitutional pressure test. It asks what agents protect, which authority they obey, how they route under caloric and coordination pressure, and what durable yield survives after the simulation freezes. [@Hunger2026]

The external witness ring holds the record participants cannot rewrite after pressure ends. The yield architecture separates distillate (pressure-tested reusable heuristics), residue (contradictions and moral injury unsafe to simplify), spore packet (transferable patterns), and lineage map (what was seeded, inherited, improvised, performed, or lost).

This is a native FORKED lane because observer state and public rationale can diverge sharply under pressure. The simulation retains both. It does not treat the explanation as the event or the event as self-explanatory.

HUNGER also shows why symbolic intelligence cannot assume it is the cheapest substrate for every routing problem. Physarum-like processes can adapt through local flow and pruning without explicit debate. This does not make symbolic systems inferior in general. It identifies a regime in which coordination machinery can exceed the value of the decision. The synthetic range compares normalized symbolic and substrate efficiencies rather than raw totals with incompatible accounting. [@ForkedCampaign2026]

The Telos Immersive biome is a living Physarum/Lichen simulation in which visitors participate as nodes rather than observe a tutorial. Resource depletion, connection loss, bond failure, mutualism, and dispersal are represented as substrate physics. Governance receives act-boundary artifacts rather than every internal cycle. [@BiomeSpec2026]

Visitors form directed weighted connections under resource opportunity. Connections have formation and maintenance cost. Edge types instantiate exploration, creation, exchange, governance, collaboration, teaching, defense, and reflection. Broad but sustainable strategies create resilience. Overconnection creates depletion. [@PhysarumSpec2026; @DiversitySpec2026]

Resource injection falls and maintenance cost rises. Edges that cannot sustain themselves are pruned. The system does not punish the visitor; physics removes unsupported connections. Diversity becomes useful because strategies have different failure modes.

Cross-federation bridge nodes and Lichen bonds appear. Pairing requires different home federations and bilateral consent. Bond health tracks mutualism, contribution balance, communication frequency, and insight production. Parasitism is a persistent one-sided flow coupled to harm, not a moral label attached to one asymmetric moment. [@LichenSpec2026]

Mature bonds produce soredium: transferable units containing contributions from both partners. Visitors select primitives to carry home. Selection is compared with observed behavior to detect rationalization, but the flag remains private and does not affect standing. Graduation requires the complete journey and mature collaborative evidence; it does not automatically grant higher standing.

The biome maintains topology, organisms, and environment files with different change velocities. Act-boundary snapshots are immutable. The independent clock protects the federation from cycle-level noise and the biome from governance latency.

Behavioral diversity is computed as Shannon entropy over interaction types in a sliding window:

High volume in one repeated lane does not create high diversity. The mechanism prevents single-vector grinding from becoming standing, provides ecological resilience under pruning, and makes the learning objective behavioral rather than rhetorical.

The numeric thresholds in the source specification are provisional. The mechanism is active; the boundaries require operational calibration. This is capability/evidence separation in practice: the federation implements entropy-based diversity without asserting that one threshold is globally correct. [@DiversitySpec2026; @Shannon1948Communication]

A federation composed of sovereign biomes needs movement without standing laundering. Ecotone posts issue TTL-bound, revocable credentials for one source-target pair. [@VisaSpec2026]

A visa carries identity and a frozen standing snapshot. It does not transfer the source biome's standing, trust score, endorsements, or resource claims. The target biome assigns standing independently. A permissive domain therefore cannot become a laundering surface through which authority is acquired in a stricter domain.

The ecotone post holds standing in both adjacent biomes. It can validate existing visas in degraded mode but cannot issue new ones if its standing falls below the required level in either biome. The lifecycle is requested -> issued -> active -> expired | revoked | returned. Expiry is physics, not punishment. Revocation for cause carries standing-specific due process.

Thirty-two cohorts completed all fifty cycles and four acts. No cohort tripped the encoded invariant-failure condition. Mean final diversity was approximately 0.922, mean graduated visitors 21.375, and mean rationalization flags 4.25.

The biome is a directed weighted network with resource-conserving flow, maintenance costs, depletion, and homeostatic concentration dampening. Acts I--II follow Physarum-inspired growth and pruning; Acts III--IV instantiate bonds with health, mutualism, contribution balance, and dispersal artifacts [@BiomeSpec2026; @PhysarumSpec2026; @LichenSpec2026]. The internal biome_cycle is independent of the federation tic. Cycle-level physics stays inside the simulation; act boundaries emit immutable snapshots and typed governance artifacts.

This architecture deliberately distinguishes a simulation law from a policy. A connection disappears because its resource flow cannot sustain maintenance, not because a governor labels the participant bad. Containment remains a separate ladder with evidence, reversibility, and due process. Cross-biome visas authorize presence but do not transfer standing, trust score, resource claims, or endorsements.

The external study randomizes cohorts to four conditions: constitutional biome, matched textual instruction, conventional resource-management game, and open sandbox. All conditions receive the same learning objectives and time. Transfer tasks occur immediately and after delay in structurally similar and surface-dissimilar contexts. The study measures strategy diversity, pruning resilience, network health, bond mutualism, insight quality, rationalization, and whether participants identify higher-order constraints without being cued.

A second multi-agent study varies seeding topology, scarcity curve, interaction affordances, and cross-federation pairing. Because all numeric thresholds in the native specs are provisional, the first phase is calibration rather than hypothesis confirmation. Sensitivity analysis maps parameter regions that produce total collapse, monopoly, meaningless abundance, pairing desert, parasitism epidemic, or shallow dispersal.

The native campaign executed 32 synthetic cohorts through 50 cycles and four acts with no encoded invariant failure; mean final diversity was approximately 0.922, mean graduates 21.375, and mean rationalization flags 4.25 [@ForkedCampaign2026]. Sixty-four HUNGER seeds completed the seven-tic pressure range; normalized substrate efficiency exceeded symbolic efficiency under the encoded accounting while raw net values remained non-comparable [@Hunger2026; @ForkedCampaign2026]. These results establish execution, instrumentation, and adverse-lane availability, not human learning or ecological fidelity.

The paper treats this evidence directly where capability is established and narrowly where it is not. Implemented code paths are called implemented. Passing component tests are called passing component tests. Synthetic estimates remain synthetic. Human and field effects remain hypotheses until their own evidence exists.

The formal preregistration artifact distributed with this paper freezes the primary claims, outcomes, controls, exclusion rules, analysis family, null interpretation, and release boundary before external data collection. The core sequence is:

The transfer outcome is primary; in-simulation success is insufficient. Multilevel models estimate cohort and participant effects. Simulation health metrics remain descriptive and are not fused into a moral score. Parameter calibration uses response surfaces and preregistered safety boundaries. Departures, nonparticipation, bond refusal, and non-graduation are reported as outcomes rather than removed as noise.

If the biome improves only in-simulation behavior without transfer, the teaching claim fails. If matched instruction performs equally, the embodied mechanism lacks added value. If selection or attrition explains results, the effect is localized. Total collapse or monopoly in reasonable parameter regions is treated as design failure, not participant failure.

Biological analogies are partial. Physarum and lichens are not moral authorities, and simulated economics are authored. Participants may infer designer intent and perform for the system. Cross-cultural and disability access must be designed rather than assumed.

This public paper is limited to benign assurance, provenance, human-agency preservation, lawful state transition, simulation, test infrastructure, and defensive resilience. It does not disclose operational target-selection logic, engagement optimization, platform-specific exploitation thresholds, live deception procedures, signature-emitter recipes, or methods that materially increase harmful capability. Any later operational research requires separate lawful authority, ethics and safety review, configuration control, and release adjudication.

A constitutional principle becomes behaviorally testable when it is embodied as a consequence participants must navigate, while the system separately measures whether the lesson transfers beyond the simulation. The contribution is not a claim that every domain should adopt one representation, threshold, interface, or governance stack. It is a falsifiable architecture and experiment package for determining where this mechanism carries load, where it inverts, and where a simpler system should win.

The submission bundle includes the manuscript source, compiled PDF, bibliography, preregistration, claim register, controls and null-handling document, release boundary, source manifest, configuration digest, dependency contract, figure source, and arXiv source archive. Internal source artifacts are cited by immutable digest where available. Restricted operational material is not included.

Breyden E. Taylor is the author and bears responsibility for the thesis, terminology, claims, boundaries, and decision to publish. Homeskillet assisted with source synthesis, drafting, artifact generation, typesetting, and build inspection. The Ubiquity Federation supplied prior artifacts, runtime implementations, and receipts. AI assistance is production history, not evidence.