DiagnosticMind · ← All editions · PT

What the Runbook Assumed

Documentation captures the steps. It cannot capture the judgement that knows when a step is wrong — and that judgement leaves with the person.

Every runbook is written by someone who already knows what it does not say. The document records the steps in order. It cannot record the judgement that decides, in the moment, that the order is wrong this time — that this is the exception, that the usual fix will make it worse, that the alarm is the symptom and not the fault. The author supplies that judgement automatically, which is precisely why they never write it down. It does not present itself as a step. It is simply what they know.

The steps and the silence

A procedure is a list of actions. Around every action sits a silence: the conditions under which the action does not apply, the signal that says stop before the final step, the exception handled so often it stopped registering as an exception. That silence is invisible from both ends. The author cannot see it, because they fill it without noticing. The reader who has never needed it cannot see it either, because nothing in the document points to where it is missing. The runbook reads as complete to everyone — except in the one situation it was never going to cover, where it reads as complete right up until it doesn't.

The interchangeability assumption

An organisation that treats the document as the knowledge will treat the seat as fillable by anyone holding the matching title or credential. The reasoning is implicit but firm: the capability has been written down, therefore it transfers with the document, therefore the person is a slot. The role is costed as fungible. What this cannot see is the layer that was never written — the tacit judgement that lives in having done the thing across years, not in having read about it. That layer is built the slow way, by attending to the system as it actually behaves rather than as it is described, and it appears on no handover checklist, no organisation chart, no succession plan. The instrument that would have recorded it is the same instrument that omits it.

What leaves with the person

The assumption holds, visibly, right until it fails. The experienced holder departs; the runbook stays; the judgement does not. The system keeps running, and the continuity is read as confirmation that the role was interchangeable after all. Then arrives the first situation the runbook did not anticipate — which is, with grim reliability, the precise situation the departed person had been retained to handle. The cost of the assumption is paid in a single moment, long after the decision that incurred it, by people who inherited a document and were assured it was the knowledge.

The runbook was never the knowledge. It was the part of the knowledge that happened to be writable — the residue left after the judgement had been poured off. Treating the residue as the whole is the interchangeability assumption in its quietest form: not a claim that people do not matter, but an accounting in which the part of them that mattered most was never entered, because it could not be. The assumption was there from the start. It becomes visible only at the moment it breaks — which is the moment it costs the most to have believed.