The first question after

After something goes wrong — a decision that should not have been made, an outcome that lands on the wrong person, a result that nobody intended and everybody now has to answer for — there is a first question, and it is always the same one. Before blame, before remedy, before anyone decides what to do next, someone asks what happened. Not whose fault it was; that comes later and depends on this. Not how to fix it; that comes later still and also depends on this. Just: what actually happened here. It is the question that every subsequent question is built on, and until it has an answer, nothing else can move. The investigation, the apology, the correction, the assurance that it will not recur — all of them wait on the answer to the first question.

And here is the thing that is easy to miss in the urgency of the moment: whether that question can be answered at all was not decided in the moment. It was decided earlier — sometimes much earlier — by what the system that made the decision happened to keep. By the time anyone is asking what happened, the answer either exists or it does not, and which of those is true was settled before the asking began, on a day when no one was asking anything, when the decision that would later go wrong looked exactly like every other decision and seemed to warrant no special attention at all.

The answer is built before the question is asked

This is the inversion at the heart of accountability, and it is genuinely counterintuitive. We tend to imagine that the work of accounting for a decision happens after the decision goes wrong — that when the bad day comes, we will reconstruct what happened, dig through what is there, piece together the story. But you cannot reconstruct from what was never kept. The reconstruction can only work with the material that exists, and the material that exists is whatever the system retained at the moment of deciding, when it had no idea this particular decision would ever be questioned. The answer to what happened is not produced on the bad day. It is merely retrieved on the bad day, from a record built on an ordinary one. If the record was not built then, there is nothing to retrieve now, and no amount of urgency or goodwill or investigative effort can conjure it back.

So the quality of your worst day — how well you can answer for it, how much of it you can actually account for, how much you are reduced to guessing about — is set on your ordinary days. It is determined by what you chose to keep when nothing seemed to be at stake, when every decision passing through the system looked routine and unremarkable and not worth the trouble of a real record. The trouble is that the decision which later goes wrong is, on the day it is made, indistinguishable from the thousands that do not. You do not get a signal that this is the one to record carefully. You only learn which decision mattered after it has already happened, by which point the choice of whether to record it has long since been made — and made, almost always, by default.

You cannot reconstruct from what was never kept. The answer to your worst day's first question is built on an ordinary day, or it is not built at all.

Why the ordinary days carry the weight

This is why accountability cannot be a posture you adopt when trouble arrives. By the time trouble has arrived, the relevant decisions about what to keep are already in the past, frozen, unalterable. The only decisions that genuinely shape whether you can answer the first question are the ones made far upstream, on the quiet days, about which decisions get an account and which pass through unrecorded. An organization that records carefully only the decisions it suspects might matter is recording carefully almost none of the ones that will, because the ones that will are not the ones it suspects. The whole point of the first question is that it falls where you did not expect it — and the only defense against a question you did not expect is a record you built before you knew you would need it.

There is a discipline in this that has nothing to do with caution and everything to do with seeing clearly. The temptation on an ordinary day is to treat the record as optional, because nothing is at stake and the cost of keeping it is immediate while the benefit is hypothetical and remote. But the benefit is not hypothetical; it is merely deferred and invisible until the moment it becomes the only thing that matters. The record you build on a day when nothing is at stake is not insurance against a risk that might not materialize. It is the literal substance of the answer to a question you will certainly be asked, about some decision you cannot yet identify, on a day you cannot yet see. The stakes of the ordinary day are not zero. They are simply hidden, the way the stakes of every preparation are hidden until the thing you prepared for arrives.

So when the bad day comes and someone asks what happened, the honesty of your answer was already written long before, in the accumulated habit of what your systems did and did not keep when no one was watching. You can build that answer now, on the unremarkable decision in front of you that will probably never be questioned — or you can decline to, and gamble that this one is not the one. The first question always comes. What it finds when it arrives is the only thing you ever actually controlled, and you controlled it earlier than you think.

— Dispatches · Summit Cognitive