The gap in the middle, which is worse than a gap at the end

Two bundles for the same quarterly control, over the same twelve-month period.

The first holds two artefacts. One dated the eleventh of January, one dated the nineteenth of December. Both complete, both signed, both exactly what the control is supposed to produce.

The second holds two artefacts as well, dated the sixth of April and the fourteenth of July. Nothing before, nothing after.

Same count. Same quality. The first one reads much worse, and the reason is not a rule anybody wrote down.

Predict

Why does the bracketed bundle read worse than the clustered one?

Choose one answer

It covers less of the period than the two in the middle doThe pattern is the signature of a control that ran when watchedEleven months is long enough for the control to have changedA gap in the middle is harder to explain than a gap at the end

Lock in answer

Why it reads worse than having less

Put the bracketed bundle next to a third one: a single artefact, dated the twentieth of August, with an honest note saying the control ran once that year and the organisation knows it.

The third bundle holds half as much and is in a better position, for three reasons that have nothing to do with volume.

It makes one claim rather than two. It says the control ran once. The bracketed bundle implies a quarterly cadence and then contradicts itself with its own dates, and a bundle that argues with itself is worse than one that does not try.

The open question has an answer. What happened in the other three quarters? It did not run. That sentence is available, it is true, and it can be said out loud. The bracketed version leaves the question hanging in a way no artefact resolves.

The remediation is credible. One instance with a named reason is a control that somebody understands. Two instances at the edges is a control nobody has yet admitted is not running, and no plan written on top of that admission is worth much.

The gap that is genuinely explicable

None of this makes a middle gap automatically damning, and treating it that way pushes people towards the one response that is actually dangerous.

Gaps have real causes. A platform migration that moved where the control happens. A regulatory change that suspended the activity being controlled. A reorganisation where the owner changed and the handover was documented. A control whose cadence was formally changed, with a date, by somebody with the standing to change it.

In every one of those the gap is a fact about the period with an artefact of its own: the migration plan, the decision to change the cadence, the handover note. The move is to put that artefact in the bundle beside the gap, in the same breath, rather than waiting to be asked. The gap stops being a silence and becomes a described event, which is the only transformation available once the months have passed.

›The one response worth refusingOptional

There is a response to a middle gap that fixes the arithmetic and should not be reached for: producing the missing instance now and dating it then.

It is tempting precisely because it is easy. The review would have reached the same conclusions. Nobody is being deceived about the state of the system. The record is accurate about everything except when it was made.

And it converts a problem about a control into a problem about the trustworthiness of every artefact you hold. An organisation with one backdated record has, from the outside, an unknown number of them, because there is no way to tell from the inside of a document when it was written. The gap costs you a conversation. The backdated record costs you the evidential value of the whole bundle, including all the parts that were honest.

The available alternatives are all worse-sounding and all survivable: record the gap, record why, record what changed so it does not recur, and date the new record today.

Check

A monthly control has artefacts for every month except June and July.

Choose one answer

A two-month gap mid-period, so the whole bundle is underminedName the two months, say what happened, and keep the other tenIt is fine, since two months out of twelve is a small shortfallProduce the two missing records now so the set is complete

Check answer

Worth remembering

• A cadence is a claim about regularity, so the shape of a bundle carries information its count does not.
• Artefacts at the start and the end of a period are the signature of a control performed when somebody was watching.
• A bracketed bundle argues with itself: its dates contradict the cadence its presence implies.
• One honest artefact with a stated reason is in a better position than two that imply a cadence nobody kept.
• A gap at the end has explanations available; a gap in the middle has already happened and only a described event closes it.
• Real gaps usually have artefacts of their own — a migration plan, a cadence change, a handover note — and those belong in the bundle beside the gap.
• Producing a missing instance now and dating it then trades a bounded question for an unbounded one, and cannot be undone.