A board meeting · a technical panel · an audit committee. The agenda moves along, items are dealt with, someone presents, someone comments, no one objects. At the end of the agenda, the committee adjourns with a sense of having done its duty. And the organisation resumes operating exactly as before.


What happened there — what did not happen there — defines the organisation more than any documented decision.

The committee's silence is a political decision about what to avoid.

The question that inhabits the room

The auditor also reads what was left out of the minutes. An isolated omission says little. The same question missing across several meetings, while the problem resurfaces in incidents or corridors, forms a pattern that warrants examination. Silence is a signal; the conclusion requires additional evidence.

The three types of unasked question

The uncomfortable question. Asking it would reveal incompetence · one's own or that of someone at the table. Whoever asks it risks their political standing; that is why no one does. The system learns to operate by dodging it · it manufactures euphemisms, indicators that skirt around it, metrics that measure it obliquely. Incompetence disappears from the documented system and continues operating in reality.

The operational question. Asking it would reveal that something is being done badly through carelessness, and that the chain of command should have detected it. The system then adopts a tacit rule · what is already known to be broken is kept quiet, because asking about it means having to fix it.

The critical question. Asking it would reveal a structural failure in the system. Its answer requires redesigning the organisation's understanding of itself. It is the most politically costly and therefore the most avoided · and it is the one that best predicts future incidents.

The documented case

The first volume of the British public inquiry into the Horizon system concluded that Post Office and its technology supplier knew — or should have known — that the system produced accounting errors, while more than nine hundred branch managers were prosecuted using evidence from that same system; the inquiry links at least thirteen possible suicides to the scandal (Post Office Horizon IT Inquiry, vol. 1, 2025). Decades of panels, committees and minutes. The question — what if the system is wrong? — was available at every meeting. It was the critical question · answering it required redesigning institutional faith in its own infrastructure and reviewing convictions. The organisation chose silence, meeting after meeting, until the question returned as a crisis of state.

The everyday version of the same mechanism is less tragic and equally instructive. An organisation commits to a single AI supplier for convenience and a discount. When the technical committee finally reviews the market, far superior models are available · the decision had already been made. The question — against which alternatives did we decide? — came too late because no one wanted to reopen a comfortable arrangement.

The committee that provides cover and the committee that deliberates

The committee that provides cover can be recognised by its symptoms · the agenda proceeds without friction, items arrive with prior consensus, disagreement is handled in corridors, the minutes record decisions and omit debate, the signature belongs to the committee and to no one.

The committee that deliberates leaves other traces · the agenda allows room for unwritten questions, silence is read as space to articulate what is difficult, the chair creates conditions for the sensitive question to be asked without unnecessary personal cost, and the minutes have identifiable individual signatories.

The difference between the two lies in institutional culture, and culture leaves traces · complaints processed, dissent recorded, decisions changed in response to new evidence and no observable retaliation. None of these traces is sufficient on its own; together they allow an assessment of whether the committee listens or merely documents.

This decade's question

One concrete question for 2026 is this · what artificial intelligence are we using that is not listed in the inventory? It may reveal a lack of awareness, unrecorded use or a poorly defined boundary. Criterion 10 explains how to investigate the answer.

The doctrinal relationship is explicit · silence at the table is the deliberative version of the dark zone. What the committee avoids asking and what the clean cycle stops examining are the same phenomenon at two scales · Criterion 11 develops the second.

The criterion test

Two checks, one documentary and one personal.

Check one · take the recent minutes and look for questions, alternatives considered and changes in decisions. There is no healthy quota of disagreement per quarter. A total absence of written deliberation is cross-checked against interviews and incidents before drawing a conclusion.

Check two · interview participants with different roles separately and ask what relevant issue was left out. Independent points of agreement guide the audit sample. Divergent answers do not invalidate the exercise; they show that silence can be interpreted in different ways.

When an organisation stops recognising its omissions, an independent perspective can reopen questions that routine has closed. That is one function of external auditing, not a guarantee that it will always find them.

If you want to know an organisation's true state, listen for the questions missing from the minutes. The minutes record what was decided · silence records what was avoided.