The evidence of absence

We are taught, early and thoroughly, that absence of evidence is not evidence of absence. It is a fine rule for the natural world, where the universe keeps no minutes and a thing unobserved may simply have gone unwitnessed. But institutions are not the natural world. An institution that makes a consequential decision has a hand on the pen. It can choose to write the decision down or choose not to. And once choice enters, the old caution inverts. When an organization that could have kept a record did not, the missing record stops being a neutral gap. It becomes a fact about the organization — a fact it produced on purpose, or by a negligence it was free to avoid.

This matters because so much of accountability now turns on what can be produced after the fact. Someone is harmed by a decision. They ask the obvious question: how was this decided, and on what. The institution reaches for its files, and either the account is there or it is not. We have trained ourselves to read the second case as a kind of zero — unfortunate, perhaps embarrassing, but not itself incriminating. The decision still happened; we just cannot see it. That reading is a gift to whoever wanted not to be seen. The honest reading is that the inability to produce an account is, in matters of accountability, one of the most legible accounts there is.

Silence is a choice, and choices have authors

Consider what it takes for a record not to exist. At some point a process was designed, or allowed to run undesigned. Someone decided which steps would leave a trace and which would pass through unmarked. A retention period was set, or defaulted to nothing. A logging switch was left off, or turned off, or never wired in. None of this is weather. Every gap in the record is downstream of a decision about what to remember, and that decision had an author and a moment, even when no one will admit to either. The absence is not the absence of a decision; it is the visible shadow of a decision to keep the original decision out of view.

And the shape of the absence is informative. Institutions are rarely uniformly forgetful. They remember what protects them and lose what exposes them; they keep the timestamp and lose the reasoning, keep the approval and lose the dissent, keep the outcome and lose the inputs that would let you test whether the outcome was warranted. When a record is complete everywhere except at the one joint where the decision was actually made, that is not a coincidence to be excused. It is a pattern to be read. The selective gap is the tell.

We treat silence as the absence of testimony. In an institution that could have spoken, silence is testimony — it is the institution declining to be cross-examined.

What the missing account says

There is a clean test for whether a decision was made in a defensible way, and it is uncomfortable precisely because it is so simple. A decision that could survive being seen tends to get written down, because writing it down costs the decider nothing and protects them entirely. A decision that could not survive being seen tends not to get written down, because the record would be the first thing handed to anyone inclined to object. So the presence of a usable account is weak evidence that the decision was sound, and the absence of one is strong evidence that someone, somewhere, sensed it would not hold and arranged for there to be nothing to hold up. The asymmetry is the whole point. Keeping the record is cheap when you are right and expensive only when you are wrong.

This is why provenance is not a clerical virtue but a moral category. To say where a decision came from — what evidence was in front of it, which rules were active, who set the threshold, when each thing happened — is to make the decision answerable. To leave those things unrecorded is to make it unanswerable, and unanswerability is not a side effect. For a decision that someone would rather not defend, unanswerability is the goal, and the missing file is the instrument that achieves it. The gap does work. It is not the failure of an account; it is an account of a different kind, written in what is not there.

None of this requires assuming bad faith in every case. Plenty of records go missing through ordinary disorder, and not every gap is a confession. But the burden has to sit in the right place. An institution that decides about people, and cannot show how, is not entitled to the benefit of the doubt that the natural world receives. It had the pen. It could have kept the account and chose, by act or by neglect it was free to correct, not to. The person harmed should not have to prove what the record would have said. The institution should have to explain why there is no record to consult — and "we did not think to keep one" is not an answer that should comfort anyone, because the not-thinking was itself the decision under review.

So when an organization spreads its hands and says there is simply nothing to show you, resist the reflex to hear nothing. A refusal to account is not the absence of a position; it is a position, and a revealing one. The decision was made somewhere, by someone, on some basis. That all of it is now unreachable is not an accident that befell the institution. It is, far more often than we let ourselves notice, the one part of the decision that went exactly as intended.

— Dispatches · Summit Cognitive