The right to the inputs

When a decision goes against someone, the first thing we tend to offer them is an explanation. We have built a whole discourse around this — the explanation as the centerpiece of fairness, the account of the reasoning as the thing the affected party is owed. And there is something right in the instinct; a decision delivered with no account at all is plainly worse than one delivered with words attached. But I have come to think the explanation is the second thing, not the first, and that treating it as the first quietly hands the deciding party a way to satisfy the letter of fairness while keeping its substance entirely out of reach. There is a more basic entitlement underneath, and it is not the right to be told why. It is the right to see what.

Consider what an explanation actually is. It is the deciding party's own narration of its reasoning — a story the decider tells about how it got from the facts to the outcome. Even at its most sincere, an explanation is a secondary artifact, produced by the same party whose judgment is in question, describing a process you cannot independently inspect. You are being asked to evaluate the soundness of a conclusion on the basis of the conclusion-maker's own summary of how it reached it. This is not nothing. But it is structurally compromised, and the compromise is exactly the kind that fairness is supposed to guard against: the party being challenged is also the party narrating the grounds of the challenge.

You cannot argue with what you cannot see

The inputs are different in kind. The inputs are not the decider's story about the decision; they are the actual material the decision rested on — the facts asserted, the records pulled, the data fed in, the specific evidence that, if any of it is wrong, makes the whole conclusion unsafe regardless of how impeccable the reasoning was. And here is the point that I think reorders the whole question of fairness: you can contest a flawed input without any access to the reasoning at all, but you cannot contest flawed reasoning if the inputs are hidden from you, because you have no way to know whether the reasoning ever met reality in the first place.

Suppose a decision rested, in part, on a record that says you were somewhere you were not, owed something you did not owe, or were someone you are not. If you can see that input, you can refute it directly — here is the proof it is mistaken — and the entire decision built on it collapses, no matter how elegant the chain of inference above it was. But if you are given only an explanation, the false input is buried inside it, doing its quiet damage, and the most sophisticated account of the reasoning in the world will not surface it, because the reasoning was internally valid; it simply ran on a falsehood. The explanation cannot rescue you here. Only the inputs can, and only if you are permitted to see them.

An explanation lets you dispute how the decider thought. The inputs let you prove what the decider thought with was wrong. The second is the deeper power, and it is the one that gets withheld.

The order of the rights is the whole argument

This is why I want to insist on the priority — not merely that both matter, but that one comes before the other and grounds it. The right to the inputs is prior to the right to an explanation in the strict sense that the explanation is worthless without the inputs and the inputs retain their force without the explanation. Give me the evidence the decision rested on and withhold the reasoning, and I can still mount a genuine challenge by attacking the evidence. Give me the reasoning and withhold the evidence, and I can do nothing but admire or resent the prose, because I have been handed a conclusion about facts I am not allowed to examine. The asymmetry is total. One of these is a real entitlement and the other, on its own, is a courtesy.

There is a reason institutions prefer to offer the explanation and not the inputs, and it is worth naming plainly. The explanation is cheap and safe; it can be written carefully, scrubbed of anything embarrassing, made to sound complete while revealing nothing checkable. The inputs are expensive and dangerous, because they can be wrong, and once you have seen them you might be able to prove they were wrong, and that proof would bind. To disclose the inputs is to expose the decision to the one form of challenge it cannot talk its way out of: the demonstration that its evidence was false. Naturally, then, the inputs are the thing most often kept back, dressed up as confidentiality or proprietary method or simply not mentioned at all, while the explanation is offered generously in their place.

So when we ask what a decision owes the person it landed on, we should start one step earlier than we usually do. Not "can you explain your reasoning" but "will you show me what you decided on" — the evidence as it actually stood, available to the party with standing to contest it, before any narrative is layered on top. This is what it means for a Decision Receipt to carry its inputs and not merely its account: it makes the deeper right real, the right to see the material and test it against the world. The explanation can come after, and it should. But the inputs come first, because contestability lives in them, and a fairness that offers the explanation while withholding the inputs has offered the affected party everything except the one thing they could have used.

— Dispatches · Summit Cognitive