The record outlives the reason

At the moment a decision is made, the reason for it is the most vivid thing in the room. Everyone present can feel why it is obvious. The context is in the air, the constraints are understood without being stated, and the urgency that makes one option clearly right and the others clearly wrong is so present that no one thinks to write it down. Why would you? It is self-evident. That is precisely the trap. The reason is vivid exactly because it is temporary, and the more obvious it feels in the moment, the more completely it will vanish once the moment passes. Obviousness does not survive. It is the first thing to go.

What follows is a slow and total evaporation that no one notices while it happens. The people who held the reason move to other teams, other companies, other lives. The context that made the constraints legible dissolves into the background noise of a world that has moved on. The urgency that organized the whole decision is gone within weeks, and with it goes the felt sense of why this and not that. Within a year, the reason exists nowhere except in the memories of a shrinking group of people, and memory is not storage — it is reconstruction, edited each time it is recalled, until what remains is a story about the decision rather than the decision itself.

The questions do not keep the same schedule

If consequences arrived on the same timetable as reasons, none of this would matter. The reason would fade just as the need for it faded, and we could let both go together. But consequences keep their own clock, and it runs much longer. A decision made in an afternoon can surface a question five years later — from a person it affected, a regulator, a court, a successor trying to understand why the institution is shaped the way it is. The question arrives long after the reason has gone, and it arrives addressed to an institution that no longer contains anyone who was in the room.

So the institution does the only thing it can. It reconstructs. Someone is asked why the decision was made, and they assemble a plausible account from what survives — the outcome, the present-day rationale that would justify it, the general way things were done back then. This reconstruction is offered, and received, as memory. It is not memory. It is a present-day guess wearing the clothes of a past fact, and the institution cannot tell the difference, because the thing that would let it tell the difference — the actual record of what was known and weighed at the time — was never made. It is defending a past decision with a present invention and calling that defense an answer.

An institution without records does not remember its decisions. It re-imagines them, on demand, and cannot tell the difference between the two.

The danger here is not only that the reconstruction might be wrong, though it often is. The deeper danger is that it is unfalsifiable. There is nothing to check it against. A guess about why a decision was made, offered years later by people who were not there, cannot be confirmed or refuted, because the only thing that could confirm or refute it — the contemporaneous record — does not exist. So the institution is left in a position that looks like accountability and is in fact its opposite: it can always produce an answer, and it can never produce evidence. It has fluency where it needed provenance.

What a record is actually for

This is the case for capturing the reason at the moment it is still present, not because anyone will need it tomorrow, but precisely because no one will be able to recover it the day they finally do. A record made at the time of decision — the evidence in front of it, the rules it was bound by, the state of the world as it then stood — is the one artifact built to outlast the reason that produced it. Everything else about the decision is perishable. The record is the part designed not to perish.

And the value of that record is not retrospective comfort. It is the difference between an institution that can answer a question from five years ago with a fact and one that can only answer it with a story. When the consequence finally surfaces and the question finally lands, the institution with a record can say what was actually known and weighed, and can be checked on it. The institution without one can only say what now seems likely to have been the case, and cannot be checked at all, because it has nothing to be checked against. One of these is memory. The other is reconstruction from nothing.

The reason will fade. That is not a failure of attention or discipline; it is the nature of reasons, which are made of context and urgency and both of those are made to expire. You cannot keep the reason. What you can do is leave behind the one thing that does not fade — a record made while the reason was still in the room, waiting, patient, for the question that will not arrive until everyone who knew the answer is gone.

— Dispatches · Summit Cognitive