Purpose: This document is a pre-P3 filter for identifying, assessing, and establishing minimum governance structures across your organisation's critical commitments. Load this guide into your AI system alongside the companion Minimum Viable guides for Principles, Products, Strategy, Patterns, and Maturity as part of a coherent P3 bootstrapping toolkit.

Governance is not compliance, bureaucracy, or approval chains. It is the integrity function — the organisational mechanism that ensures the system does what it claims to do.

The integrity question at the heart of every governance structure: Are we doing what we said we would do, in the way we said we would do it?

A pilot's pre-flight walkaround provides a useful frame. The pilot circles the aircraft, works through the checklist, confirms all items are ticked. The checklist is not governance. Airworthiness is governance. Airworthiness is the systemic property that emerges when every component, every maintenance action, and every operational decision exists in a traceable relationship to the engineering principles that make flight possible. A pilot can complete every item on the checklist and still board an aircraft whose turbine disk has an invisible fatigue crack. The walkaround checks the clipboard. Airworthiness checks the aircraft.

This distinction is constitutional. Most organisations govern their clipboards. Very few govern their aircraft.

Done vs. Valid. The governance distinction that matters most at pre-P3 maturity is between Done and Valid. Done is a stakeholder-facing claim: the work satisfies the criteria agreed with those who requested it. Valid is a system-facing claim: the work maintains the integrity of the commitments that shaped it. A product can be Done — accepted, signed off, delivered — without being Valid. Every governance failure that makes the news is a Done Without Valid failure: the paperwork was complete, the approvals were collected, and nobody verified that the system was actually doing what it claimed.

At pre-P3 maturity, the minimum governance question for any delivered output is: Can we trace this back to a governing commitment? If we removed that commitment tomorrow, would this output need to change? If yes to the second question, the output is Valid. If you cannot answer the first question, governance is absent.

The P3 minimum governance structure requires three elements for each critical commitment:

Before assessing your governance structures, identify which of the following pathologies are present. Each one silently converts the form of governance into its absence.

Governance Theatre. Governance Theatre is the pathology where the form of integrity checking is preserved while the substance is absent. Pre-flight walkarounds that tick boxes without inspecting anything. Architecture reviews where no one challenges a proposal. Change approval boards that have not rejected a request in eighteen months. Compliance audits that arrive six weeks after the risk they were designed to catch has already manifested.

Theatre is seductive because it is measurable: velocity metrics look healthy, approval rates reach one hundred percent, dashboards glow green. The system appears governed. It is not. The clipboard is clean. The aircraft is not airworthy.

Diagnostic question: Has this governance mechanism stopped, modified, or escalated anything in the last twelve months? If the answer is no, it is Theatre.

Governance Habituation. Habituation is the pathology of approval fatigue. When a human approver reviews their seventh context-free deployment request of the morning, they are not performing the same cognitive operation they performed on the first. The mechanism of genuine oversight degrades to reflex. Research shows it takes an average of twenty-three minutes to recover full cognitive focus after a context interruption — meaning each nominally thirty-second approval costs the reviewer a twenty-three minute recovery period, while providing no meaningful verification.

Habituation is not negligence. It is a rational response to a governance system that has placed a human in a role no human can sustain at the demanded volume. The individual is not defective; the system is. But the outcome is identical to negligence: approvals are granted without scrutiny, and the normalisation of rubber-stamping redefines "reviewed" to mean "signed."

Diagnostic question: Could the person giving this approval meaningfully evaluate it in the time the process provides? If not, the system is producing signatures, not oversight.

Leading vs. Trailing Governance. Trailing governance checks for compliance after work is completed — a post-hoc audit, a retrospective review, a quarterly committee that examines decisions made months ago. Trailing governance is better than nothing, but it has two structural weaknesses: by the time a violation is detected, the damage may be irreversible; and the interval between violation and detection — Feedback Latency — allows the violation to compound into the context for subsequent decisions.

Leading governance embeds the governance check into the work itself, before the point of irrevocable commitment. A deployment pipeline that cannot proceed without passing security verification is Leading governance. A code review policy that someone might or might not apply is Trailing. Leading governance makes non-compliant behaviour structurally difficult rather than merely inadvisable.

Feedback Latency is the diagnostic metric: when the interval between an event and its governance visibility exceeds the system's rate of change, governance degrades from regulation to archaeology. A quarterly architecture review that discovers a violation introduced three months ago arrives after the operational domain has moved on, compounding the deviation into subsequent work.

At pre-P3 maturity, target at least one Leading governance check for each of your three most critical commitments. Everything else can be Trailing — but critical commitments need a structural check, not a periodic one.

For each critical commitment, document it using this six-element structure. All six elements are required. A governance structure missing any one of them is incomplete.

The Sentinel Triad. Every sentinel must possess three attributes to function. Observability: the capacity to see whether the commitment is being honoured — not based on reports, but on direct visibility into the work. Authority: the capacity to act on what is observed — to stop work, request rework, or escalate — without needing permission from those being governed. Accountability: the bearing of consequences when the sentinel fails to detect a violation. Without observability, the sentinel is blind. Without authority, they are impotent. Without accountability, they have no incentive to exercise either.

Displacement: from ritual to structure. The strongest form of Leading governance makes non-compliance architecturally difficult rather than merely inadvisable. A CI/CD pipeline that cannot proceed without passing defined quality gates has displaced the manual approval ritual — the governance intent is preserved; the ritual is structurally unnecessary. At pre-P3 maturity, full displacement may not be achievable, but for your top three commitments ask: Can this check be embedded in tooling, so that violation is detectable at the point it occurs? Even partial displacement — automated detection plus human review of exceptions — dramatically reduces Habituation risk.

Constitutional independence. The sentinel cannot answer to production pressure. When the person responsible for governance oversight reports to the person whose work they oversee, governance will systematically defer to velocity under pressure — not from weakness of character, but from structural inevitability. The builder cannot hold the level. At pre-P3 maturity, constitutional independence means the sentinel's role is explicitly distinct from delivery responsibility, and escalations go to someone outside the delivery team's chain.

Governance must be calibrated to both the criticality of the commitment and the maturity of the practice it governs. Miscalibration in either direction produces failure.

Over-governance of low-maturity practices creates paralysis. Applying constitutional enforcement to ad-hoc work — demanding full audit trails for practices that are not yet documented — generates friction practitioners will route around, producing Theatre by another path.

Under-governance of critical commitments creates systemic risk. Leaving safety, compliance, or architectural integrity on an honour system is deferred liability. Every organisation has commitments whose violation produces consequences that cannot be undone. These need governance regardless of maturity level.

The pre-P3 calibration test — three steps:

Governance Friction cost. Every governance mechanism carries a productivity cost. Research indicates theatrical governance — compliance rituals that produce signatures rather than integrity — accounts for fifteen to forty percent of engineering productivity losses at scale. The discipline of minimum viable governance is as much about what you do not govern as what you do. Define a governance structure for every commitment you cannot afford to have violated. Monitor the rest informally until evidence warrants escalation.

The governance posture test. For each governance mechanism in your current practice, ask: Does this protect against a real violation, or against the appearance of one? The former is governance. The latter is Theatre. Audit your rituals with this question and dissolve what you find.

Connecting to the framework. Governance calibrates to Maturity (current practices constrain governance complexity), protects Principles (the sentinel watches that governing commitments are not violated), and validates Products (Done vs. Valid is governance's contribution to every product decision). The companion guides for Minimum Viable Maturity and Minimum Viable Principles feed directly into governance design: Maturity sets the ceiling; Principles define the commitments governance exists to protect.