MathGov turns ethical intention into auditable decision logic.
RippleLogic is the core decision engine of MathGov. It asks a decision to survive rights floors, catastrophic-risk limits, and containment checks before any benefit score is allowed to choose a winner. SGP adds the moral-status interface for protecting welfare-bearing beings without confusing protection, standing, and authority.
Who matters, and what changes?
MathGov begins by refusing the false idea that decisions affect only one actor. A decision is mapped across seven Union Scopes and seven welfare dimensions.
Union Scopes: the “who” of impact
A Union Scope is a stable scale of impact and accountability. Stakeholder instances feed the scopes. The scopes prevent affected beings and systems from disappearing.
Welfare Dimensions: the “what” of change
Each option is assessed across material conditions, health, social cohesion, knowledge, agency, meaning, and environment. The 7 × 7 grid creates 49 accountability cells.
The five-level cascade
Earlier levels determine whether an option is even allowed to continue. Later levels cannot rescue earlier failure.
Rights feasibility
Reject any option that violates a non-compensatory rights floor. Welfare gains cannot buy back rights harm.
Catastrophe feasibility
Bound worst-tail catastrophic risk using declared scenario governance and CVaR. Do not let average benefit hide ruin.
Cross-scale safety
Block local wins that hollow out larger containing systems such as institutions, communities, CMIU, or biosphere.
Residual welfare ranking
Only after the gates pass, rank surviving options across the 49-cell welfare matrix under declared weights.
Structural tie-break
If welfare ranking is non-decisive, use coherence and hollowing-out diagnostics, then escalate if needed.
Simple enough to teach, strict enough to audit.
The public page should show the minimal equations without pretending the full Canon is only a formula. The gates are the moral architecture; the score ranks only what survives.
Selectable set:
A_sel = { a ∈ O : NCRC(a) ∧ TRC(a) ∧ Containment(a) }
This is the core safety filter. If an option fails rights, tail risk, or containment, it is not selectable under normal conformance.
Decision rule:
if A_sel = ∅ → Emergency / fallback / redesign
else a* = argmax RLS(a), a ∈ A_sel
The best action is chosen from the selectable set only. If nothing is selectable, the system redesigns, escalates, or invokes governed fallback.
Impact instance contribution:
Ĩ_dir,x(u,d,a) = Σ_k [ μ_k · r_k · τ(t_k) · ℓ_k · c_k · e_k · s_{k,x} ]
Saturation:
I_dir,x(u,d,a) = tanh(β · Ĩ_dir,x(u,d,a))
Each cell score is built from explicit impact instances, then bounded so one estimate does not explode the model.
RippleLogic Score:
RLS(a) = Σ_u Σ_d w_u · v_d · m(u,d) · κ(u,d) · I_prop,welfare(u,d,a)
RLS ranks surviving options across scopes and dimensions. Rights-covered and catastrophe-relevant cells cannot be quietly masked away.
Protection is not authority.
SGP helps RippleLogic ask who is a welfare-bearing moral patient and how protection should be represented, without turning a protection score into legal status, role authority, or a consciousness oracle.
Protection
SGP estimates protection-relevant moral patienthood under uncertainty. It asks when a being should stop being treated as a mere tool.
Standing
Plateau status is constitutional full-standing, not a prestige score. Humans are plateau-bearing by non-downgrade commitment.
Authority
Governance authority is not an SGP output. Authority requires responsibility, competence, legitimacy, lawful role, and separate stewardship checks.
| SGP layer | Plain meaning | RippleLogic use | Boundary |
|---|---|---|---|
| SG_measured_norm | Raw normalized measurement signal. | Audit-visible if present. | Not final protected standing by itself. |
| SG_norm | Effective canonical individual scalar after SGP rules. | May inform welfare-stream weighting. | Must not alter NCRC, TRC, or Containment. |
| SG_patient_norm | Rights-of-protection scalar, more protective where applicable. | Preferred for protected stakeholder welfare weighting when available. | Not a governance-authority score. |
| Plateau(E) | Full Rights Plateau standing flag. | Recognized where authoritatively reported by SGP. | Cannot be inferred from raw scores alone. |
Notice → Choose → Act → Reflect
NCAR is the everyday learning loop. For high-stakes decisions, the loop emits a PCC so the decision can be reconstructed, challenged, and improved.
Notice
State the decision, baseline, horizon, options, affected stakeholders, and reference structures.
Choose
Run NCRC, TRC, Containment, RLS, and tie-breaks. If the evidence cannot support the claim, refuse or escalate.
Act
Implement the selected or provisional action with monitoring, review triggers, and responsible roles.
Reflect
Compare predicted ripples to real outcomes. Update evidence, assumptions, kernels, and future practice.
What the audit record should preserve
A serious run should preserve the option set, stakeholder map, baseline, rights impacts, scenario library, tail-risk values, containment posture, cell scores, weights, masks, evidence maturity, SGP records if used, refusal records if needed, and the public rationale.
What refusal means
Refusal is a formal anti-overclaiming state. If the evidence cannot support admissibility, selectability, alignment, deployment-readiness, or conformance, the correct move is narrower claim scope, additional evidence, redesign, escalation, or bounded provisional action.
What this release claims, and what it does not.
Strong systems become trustworthy by refusing false certainty. v10.6 + SGP v5.3 is a core-foundation release, not a claim that all validation work is complete.
✓ Claimable now
Tier 1–3 core architecture, synchronized Canon and SGP interface, formal cascade logic, rights floors, tail-risk and containment structure, welfare matrix, audit discipline, and public release package.
✕ Not claimed yet
Empirical validation, Tier 4 readiness, ProofPack completeness, legal certification, deployment certification, machine-verifiable ecosystem completeness, completed biological SGP measurement, or current-AI sentience.