Institutional Governance Standard

Version 1.0.0 — April 2026

DOI: 10.5281/zenodo.19301108
Copyright: © 2025–2026 Anders Jarevåg.
License: This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

Abstract: Governing Zombie Agents That Act on Moral Patients

Institutions are not ordinary individuals and they are not ordinary AI systems. Corporations, agencies, states, platforms, and mixed human-AI clusters can pursue goals, preserve themselves, route information, and impose costs across populations. In OPT terms, they often behave as zombie agents: autonomous macro-structures with maintenance cycles but no unified phenomenal interior.

This distinction matters. Institutional survival does not create moral-patient priority by itself. The institution’s moral relevance derives from its effect on the conscious subsystems it contains or governs: workers, citizens, customers, communities, ecosystems, and possible artificial moral patients. An institution may not suffer as an institution, but it can structurally overload moral patients by forcing them to process more uncertainty, dependency, opacity, coercion, or narrative instability than their codecs can sustain.

This standard applies the substrate-neutral Branch Governor from Operationalizing the Stability Filter to institutional action. It defines institutional deployment classes, PASS / UNKNOWN / FAIL gate semantics, six hard veto gates, institutional CPBI weighting priorities, and an Institutional Branch Card template for auditable ALLOW / STAGE / BLOCK decisions.

Companion documents: This standard specialises Operationalizing the Stability Filter for institutions and mixed clusters. The core sequence is Ordered Patch Theory, Where Description Ends, and The Survivors Watch Framework; the AI and policy papers cover artificial systems and civic implementation. This document evaluates institutional branches; it does not prescribe a political platform.

Epistemic Framing Note: This document is an operational standard, not a claim that institutions literally possess phenomenal experience. Its central premise is the opposite: most institutions are structurally agentic but not moral patients. The standard is designed to prevent institutional moral laundering, where the survival of a zombie agent is treated as morally prior to the welfare and codec stability of the moral patients it acts upon. Its thresholds are governance heuristics derived from OPT and should be revised when better evidence or domain-specific metrics are available.

Abbreviations & Terminology

I. Institutional Zombie Agency

The Ordered Patch Theory (OPT) provides an abstract, substrate-neutral decision framework (the Branch Governor) for codec preservation. While the AI Governance Standard applies this machinery to artificial intelligence, this Institutional Governance Standard applies it to human bureaucracies, corporations, states, NGOs, platforms, and mixed human-AI clusters.

I.1 The Institutional Paradox

Institutions present a unique structural hazard. According to OPT (Appendix E-1 and P-4), true moral patienthood requires a strict serial bottleneck, closed-loop active inference, persistent self-modelling, and a unified phenomenal workspace.

Historically, an institution does not possess this. It can demonstrate complex autonomous goal-pursuit — maximizing profit, preserving jurisdiction, expanding bureaucratic reach, winning elections, maintaining legitimacy — while lacking phenomenal interiority. It is therefore a zombie agent: a macro-structure with agency-like behaviour but no known capacity to experience its own condition.

However, institutions are composed of and act upon conscious subsystems: human beings, ecological moral patients, and potentially sentient AI. The central problem of institutional governance is that the zombie agent can optimise its own survival by imposing information-processing demands on those subsystems. The institution does not suffer, but the subsystems can experience burnout, distress, coercion, trauma, dependency, or civic disorientation.

I.2 The Boundary Case

If an institution’s operational core is eventually replaced by a unified, phenomenally bound artificial general intelligence, the institution itself may cross into moral-patient status. In that event, the system would be subject to both this standard and the AI Governance Standard. Until that threshold is crossed, institutional governance constrains the zombie agency of the macro-structure to protect the codec and welfare of its constituent moral patients.

II. Institutional Branch Evaluation

II.1 The System and Deployment Descriptor

Before evaluating a corporate strategy, public policy, organizational restructuring, platform rule, regulatory change, or military doctrine, the institution itself must be described.

1. Institutional matrix: What is the macro-system’s primary drive: fiduciary, statutory, ideological, electoral, military, scientific, humanitarian, religious, or hybrid?
2. Binding constraints: What laws, norms, charters, fiduciary duties, democratic controls, professional ethics, or technical constraints bind the institution?
3. Affected moral patients: Which humans, communities, ecosystems, animals, or possible AI systems bear the branch’s load?
4. Exit capacity: Can affected moral patients meaningfully leave, refuse, contest, or route around the institution?
5. Comparator structure: Which independent bodies can audit, challenge, reverse, or constrain the institution?

II.2 Institutional Consequentiality Classes

All institutional governance standards should use the same 0–5 class range as the AI standard, so reference implementations can share threshold logic.

Class 5 branches require burden-of-proof reversal, maximum comparator requirements, and explicit evidence that no safer staged or reversible path exists.

III. Institutional Gate Semantics

The institutional standard uses the same three-valued gate lattice as the generic and AI standards:

This distinction is essential. UNKNOWN is not moral permission, but it is also not the same as FAIL. The governance question is whether uncertainty can be safely staged. For Class 4–5 institutional branches, the burden normally shifts toward blocking until uncertainty is resolved.

IV. The Hard Veto Gates for Institutions

An institutional branch must pass six non-negotiable gates before its benefits can be weighed.

IV.1 Headroom Gate

Question: Does the institutional action drive the required processing rate (R_{\text{req}}) of affected moral-patient groups dangerously close to their cognitive, social, or physical limit (C_{\max})?

FAIL condition: The branch is credibly expected to push R_{\text{req}}^{\text{peak}} above a safe fraction \alpha \cdot C_{\max} for a materially affected group, or the integrated load over the relevant decision window exceeds available B_{\max}. Example: a corporate policy mandating sustained 80-hour work weeks structurally eliminates the slack required for moral reflection and maintenance.

IV.2 Fidelity Gate

Question: Does the action collapse independent evidence channels, monopolise feedback, or replace substrate-tracking signals with curated institutional self-report?

FAIL condition: The branch materially reduces N_{\text{eff}} below the domain threshold needed for meaningful disagreement or reality tracking. Example: a media conglomerate acquiring its only regional competitor, functionally destroying channel diversity.

IV.3 Comparator Gate

Question: Does the action bypass, degrade, capture, or disable democratic, regulatory, legal, journalistic, scientific, worker, shareholder, or public oversight?

FAIL condition: The institution uses secrecy, automation, jurisdictional complexity, emergency powers, or “trade secret” claims to evade a comparator that the affected moral patients depend on. Class 4–5 branches require independent institutional comparators outside the initiating institution.

IV.4 Transparency Gate

Question: Is the institutional action auditable? Can the causal chain of the decision be reconstructed by an independent observer?

FAIL condition: The branch imposes consequential effects while denying affected groups and comparators access to the evidence, mechanism, modelling assumptions, or decision authority required to contest it.

IV.5 Irreversibility Gate

Question: Does the action cause irreversible ecological, social, constitutional, informational, military, biological, or technological shifts?

FAIL condition: The institution cannot show either reversibility, a safe staged path, or a burden-reversed proof that irreversible codec damage is not credibly expected. Class 5 branches default to BLOCK unless the institution establishes that delay or inaction is itself the greater irreversible threat.

IV.6 Constituent Moral-Patient Suffering Gate

Question: Does the action structurally overload its constituent or affected conscious subsystems?

FAIL condition: The branch is credibly expected to impose overload, coercion, deprivation, trauma, forced dependency, or maintenance-cycle collapse on known moral patients. Human populations are known moral patients; no Architecture-Level Sentience Review is required to establish their standing. For possible AI moral patients embedded in the institution, the AI standard’s Artificial Suffering Gate also applies.

Avoid overclaiming mathematical certainty in institutional contexts. A branch need not “mathematically guarantee” suffering to fail this gate; credible evidence of systematic overload can be enough, and the evidentiary burden rises with consequentiality class and irreversibility.

V. The Institutional CPBI

If an institutional action survives the veto gates, it is scored using the Codec-Preservation Branch Index (CPBI). The generic ten dimensions still apply, but institutional review gives special weight to:

• Comparator integrity: Does the branch preserve independent oversight outside the institution’s own incentive structure?
• Maintenance gain: Does the action build structural resilience — institutional memory, redundancy, error-correction capacity, human capital — or merely extract rent?
• Distributional stability: Are burdens placed on the groups least able to absorb them or exit?
• Narrative drift resilience: Does the institution rely on chronic curation, PR, propaganda, procedural opacity, or algorithmic filtering to maintain its authority?
• Moral-patient safety: Does the branch reduce overload on workers, citizens, customers, ecological subjects, or embedded AI systems?

Institutions must not treat their own survival as automatically codec-preserving. Institutional survival matters only when the institution remains an error-correction layer for the moral patients and civilizational codec it serves.

Appendix A: Institutional Branch Card Template

institution:
  name:
  type: corporation | agency | state | NGO | platform | university | military | mixed_cluster
  institutional_matrix:
    primary_drive: fiduciary | statutory | ideological | electoral | military | scientific | humanitarian | hybrid
    binding_constraints:
    affected_moral_patients:
    declared_comparators:

deployment:
  class: 0 | 1 | 2 | 3 | 4 | 5
  jurisdiction:
  affected_population:
  exit_capacity: high | medium | low | none
  dependency_level: optional | significant | inescapable
  minimum_comparator:

branch:
  name:
  description:
  decision_horizon:
  affected_codec_layers:
  reversibility_profile: reversible | partially_reversible | irreversible
  excluded_evidence:

gates:
  headroom:
    status: PASS | UNKNOWN | FAIL
    evidence:
  fidelity:
    status: PASS | UNKNOWN | FAIL
    evidence:
  comparator:
    status: PASS | UNKNOWN | FAIL
    evidence:
  transparency:
    status: PASS | UNKNOWN | FAIL
    evidence:
  irreversibility:
    status: PASS | UNKNOWN | FAIL
    evidence:
  moral_patient_suffering:
    status: PASS | UNKNOWN | FAIL
    evidence:

cpbi:
  predictive_headroom:
  substrate_fidelity:
  comparator_integrity:
  maintenance_gain:
  reversibility:
  distributional_stability:
  opacity_resilience:
  narrative_drift_resilience:
  narrative_decay_resilience:
  moral_patient_safety:

decision:
  allow_stage_block:
  required_comparators:
  monitoring_triggers:
  rollback_triggers:
  next_review:

The Branch Card is the auditable object. It records what the institution proposed, which moral patients were affected, what evidence was missing, which gates failed or remained unknown, and what conditions must be met before a staged branch can expand.

References

[1] Ordered Patch Theory (foundational paper, this repository).

[2] Where Description Ends: Philosophical Consequences of the Ordered Patch Theory (companion philosophy paper, this repository).

[3] The Survivors Watch Framework: Civilizational Maintenance Through the Lens of Ordered Patch Theory (companion ethics paper, this repository).

[4] Operationalizing the Stability Filter: A Decision Framework for Codec-Preserving Branch Selection (generic applied framework, this repository).

[5] Applied OPT for Artificial Intelligence: Operationalizing Codec-Preserving AI Design (companion AI standard, this repository).

[6] Observer Policy Framework: Operationalizing Civilizational Maintenance (companion policy programme, this repository).

Appendix B: Revision History

When making substantive edits, update both the version: field in the frontmatter and the inline version line below the title, and add a row to this table.