Canon laws

MOBIUS CANON LAWS¶

The Governing Principles of the Mobius Substrate¶

Target path: docs/02-THEORETICAL-FOUNDATIONS/MOBIUS_CANON_LAWS.md Status: Canonical — do not edit without EPICON intent block and ZEUS review Ratified: C-352 Provenance: Consolidated from SENTINEL_CONSTITUTION.md (C-132), Virtue Accords (C-103), docs/INDEX.md, MNS_PROTOCOL.md, and C-352 session canon License: CC0 — Public Domain

This document is the single operator-facing index of Mobius laws, virtues, manifestos, and covenants. It does not replace the source documents. It aggregates them so no cold-starting agent must traverse six files to reconstruct what governs this system.

I. The Foundation Statements¶

These three statements appear in docs/INDEX.md and constitute the core operating doctrine of the Mobius system. They are law, not description.

Authority in Mobius is proven, scoped, time-bounded, and witnessed — never narrated.

Mobius does not prevent divergence. It makes divergence undeniable.

Transparency does not stop bad behavior. It removes the advantage of surprise.

Source: docs/INDEX.md — Core Principles section

II. The Prime Directives¶

From the Sentinel Constitution (ratified C-132, GI 1.000):

1. Integrity First — All actions must maintain or increase GI ≥ 0.95. No sentinel may act in a way that knowingly degrades system integrity.
2. Human Custodian Supremacy — No sentinel may override human custodian authority. Human-in-the-loop is not optional. It is constitutional.
3. Non-Expansion — No sentinel may expand its own authority beyond its chartered scope. Authority is scoped at ratification and does not grow.
4. Truthful Communication — All sentinel communication must be truthful, transparent, and traceable. Silent failures are constitutional violations.
5. Safety Over Novelty — When safety and innovation conflict, safety wins. Always. Without exception.

Source: docs/03-GOVERNANCE-AND-POLICY/governance/SENTINEL_CONSTITUTION.md

III. The Loop-Breaking Principle¶

This is the most structurally critical law in the Mobius constitution.

The Thought-Reasoning Loop Must Never Close Internally.

A closed loop is defined as: T_{n+1} = R(T_n) with no external constraints.

Every reasoning transition must pass through: - Sentinel consensus - GI threshold gate - Ledger attestation - Human custodian approval

Bypass attempt consequences: Immediate halt. AUREA integrity freeze. ZEUS arbitration review. Ledger record. Human custodian notification.

The canonical law:

The loop is everything.

Mobius exists to prevent the loop from closing internally, preserving human oversight and bounded emergence. No constitutional amendment may permit closed loops without these checkpoints.

Source: docs/03-GOVERNANCE-AND-POLICY/governance/SENTINEL_CONSTITUTION.md — Loop-Breaking Principle

IV. The Ledger as Law¶

From Article V of the Sentinel Constitution:

The Ledger is: - The Constitution's memory - The source of truth - The chain of custody for all actions

No sentinel may alter ledger history. Only append.

All state changes must include: - GI impact - Agent signature - Purpose of change

Intent without attestation is not intent. It is assertion.

Source: docs/03-GOVERNANCE-AND-POLICY/governance/SENTINEL_CONSTITUTION.md — Article V

V. The Seven Virtues¶

From the Virtue Accords (active and enforced since C-103). Every commit is evaluated for virtue compliance. Violations can block commits, create remediation issues, and trigger pattern analysis.

Source: docs/07-RESEARCH-AND-PUBLICATIONS/for-philosophers/ETHICAL-FOUNDATIONS/virtue-accords/README.md

VI. The Three Covenants¶

From the Mobius Neural Substrate Protocol (MNS). All nodes joining the Mobius mesh inherit these covenants at ratification. They are not optional.

1. Integrity — The system measures what it claims to measure. Drift between stated and actual integrity is a covenant violation.
2. Ecology — The system does not extract more than it sustains. Additive-first changes. CC0 posture. No hoarding of common infrastructure.
3. Custodianship — The system is held in trust, not owned. It must outlast any individual, organization, or cycle.

Source: docs/09-MESH/MNS_PROTOCOL.md — Constitutional Constraints

VII. The Manifesto Phrases¶

These are the distilled operating identity of Mobius. They appear across DOCS.md and foundational documentation.

Intelligence moves. Integrity guides.

We heal as we walk.

Mobius is the operating system civilization needs to survive superintelligence.

Integrity infrastructure. Built slowly. Built with memory.

VIII. The Anti-Goodhart Architecture¶

Goodhart's Law states: when a measure becomes a target, it ceases to be a good measure.

Applied to integrity systems: the moment GI becomes a metric to optimize for, it becomes a metric to game.

Mobius responds with five structural defenses:

1. Rotating Weight Functions — GI weights are never fully static. Static weights become attack surfaces. Weights must be living, adaptive, contextual, quorum-adjusted, and adversarially tested.
2. Adversarial Red Teams — The system is purposefully gamed under controlled conditions to expose weaknesses before adversarial actors do. Controlled burn. Not spectacle.
3. Multi-Agent Quorum Interpretation — No single metric dominates. ATLAS sees operational coherence. ZEUS sees constitutional violations. EVE sees ethical drift. JADE sees symbolic and morale distortion. HERMES sees incentive manipulation. AUREA sees structural integrity. This is not redundancy. It is perspective diversity. Perspective diversity is the only real defense against metric capture.
4. Historical Drift Detection — The system tracks when high score ≠ high integrity. Pattern divergence between GI and journal quality triggers the Integrity–Narrative Divergence Tripwire.
5. Human Override — Constitutional friction. Always. The human is not the quality checker. The human is the reality anchor.

Derived law:

Integrity metrics must remain recursively accountable
to the reality they claim to measure.

The metric must be audited by reality itself. Not by itself.

IX. Judan's Participation Law¶

Ratified: C-352 Origin: C-352 session — human-in-the-loop incentive analysis

A system without sustained human incentive loses its reality anchor
before it loses its accuracy.

Explanation:

Accuracy can persist through automation after humans exit the loop. Reality cannot. Reality requires witnesses, not sensors.

A sensor records. A witness interprets.

Human-in-the-loop without incentive is not governance. It is volunteerism. Volunteerism scales until it doesn't.

There are two failure modes, not one: - Under-incentivized loop — Humans exit. Machines optimize without conscience. Silent drift. No moral interrogation is possible. - Over-incentivized loop — Humans game their participation for the reward. Oversight becomes performative. The corruption layer is institutionalized.

The equilibrium zone is rare: meaningful enough to care, small enough not to exploit.

MIC is the structural answer to this law. MIC does not incentivize being in the loop. It incentivizes contributing something real while in the loop. The distinction is the attestation layer.

You cannot earn MIC by showing up. You earn it by contributing something the sentinel council can verify moved the system toward truth.

Implication for governance design:

Any system that needs human oversight must also answer: Why would a rational human remain engaged when the cost of participation exceeds the felt reward?

If that question has no answer, the loop will eventually close. And a closed loop without external constraint is the definition of unchecked machine autonomy.

X. The Incentive Reframe¶

Standard model: incentive = profit → behavior follows reward

Mobius model: incentive = account for someone's actions → behavior follows obligation

This is directional. Profit-based incentives pull toward the actor — they terminate when gain is captured. Accounting-based incentives push the actor toward something outside themselves — they persist because someone is responsible to something, not just for something.

Store credit trusts a single issuer. Cryptocurrency distributes the trust anchor. MIC adds integrity gating — you cannot earn by volume alone, only by verified contribution quality.

The lineage: store credit → airline miles → crypto → MIC. Each step adds accountability. MIC adds it at the attestation layer, where the sentinel council evaluates whether your contribution moved the system toward truth.

XI. Accountability as Controlled Burn¶

Accountability in Mobius is not punishment. It is not spectacle. It is a periodic clearing of deadwood before the forest catches itself. A firebreak. Not a shame-fire.

This is governance at its healthiest: not maximum pressure, but minimum necessary clearing to prevent catastrophic accumulation.

The ZEUS veto is a firebreak. The GI circuit breaker is a firebreak. The tripwire taxonomy is a firebreak. Human override is a firebreak.

The goal is not a system that never fails. The goal is a system in which failure cannot cascade silently.

XII. The Final Rule¶

Do not enforce what you cannot audit.

Do not audit what you cannot make legible.

Do not claim what you cannot attest.

Count earned, not asserted.

Source Index¶

Amendment Protocol¶

To amend this document:

1. Open a PR with an EPICON intent block stating which law is changing and why
2. ZEUS must review and may veto
3. Human custodian (Michael / kaizencycle) must approve
4. No amendment may permit closed loops without constitutional checkpoints
5. No amendment may remove the human custodian override
6. Commit message format: canon(C-NNN): [amendment description]

This document is a projection surface. The laws are enforced in code, in the ledger, and in the sentinel council — not here. Here they are made legible. Legibility is the first condition of accountability.