Why I think winning the argument never settles what you owe the person it cost
Being right about a decision does not settle what you owe the person who bore its cost: either you have something to say to them about their own loss, or they were wronged — and a wrong owes more.
If you have ever been on the winning side of a decision that cost somebody something, you have probably had this thought: we looked at it hard, we made the right call, that is the end of it. I do not think it is the end of it — and I do not mean that as a sentiment, that we ought to feel bad or spare a thought for whoever lost. I mean a specific list of things is still owed to that person, and which list it is depends on a question the decision itself never answered.
Here is the argument, in steps you could say out loud to a friend.
The argument
1. Some decisions land on one person in particular. Not on everyone equally, and not on an average. Someone identifiable carries most of the cost while others carry very little. Those are the cases I am talking about.
2. When that happens, you owe that person an account aimed at her. Not only a defence of the decision — a defence is aimed at whoever asks. Something aimed at her: why what happened to her was all right.
3. A reason can be a genuinely good reason and still not be aimed at her. "It came out better overall" is a statement about a total, and she is not the total; she is one of the things that got added into it. "No reasonable person could object to the rule we followed" is a statement about a standardised person, and it was not a standardised person who lost her job. Both may be true. Neither is yet addressed to her.
4. So after the decision there is a second question, and it has two answers. Can you say to her, in terms about her situation, why this was all right? Either you can or you cannot. That is the fork, and the decision being correct does not put you on either side of it.
5. The two answers leave you owing different things. If you can answer her, she is still owed acknowledgment that she lost something real, an explanation she can actually get at, and repair where the loss was concentrated on her rather than spread. If you cannot, she is owed all of that and more: it has to be named as a wrong rather than a cost, the arrangement has to change so it does not fall the same way on the same people next time, and where you profited from the wrong specifically — not merely avoided a loss — that gain is not simply yours to keep.
6. "It was necessary" does not move you from the second list to the first. Necessity is a fact about your options. It explains why you had no better move; it says nothing to her about her loss. A forced hand accounts for you. It does not answer her.
7. And you cannot close her account on her behalf. You can say you have done everything you are going to do — people and institutions have to be able to stop somewhere. What you cannot do is declare, in her name and in your own favour, that she has been made square, so that raising it later counts as ingratitude or as old news. Only she can say that, and she may never say it.
Winning, then, does not settle it, and it does not move her between the two lists either. Which list she is on was never decided by whether you were right. It is decided by whether there is something you can say to her.
A case
A transit authority finds a support column failing under a station. It is the only station on that line with a working lift. The repair will take four months, and the station has to close.
For most riders this is a bad few months: walk to the next stop, leave earlier, complain. For Dana, who uses a wheelchair, the next station she can actually get into is an hour away in the wrong direction. Her forty-minute commute becomes three hours. After six weeks she has lost her shifts.
Assume the closure was right. The column really was failing, the money for a faster repair really was not there, no other station could be made step-free in time. I am not going to argue the authority chose wrongly, because I do not think it did.
Look at what the two familiar answers do with Dana. The first adds it up: thousands mildly delayed, one badly stranded, nobody crushed by a collapse — clearly the better outcome, so Dana is outweighed and the books close. The second asks whether anyone could reasonably reject a rule that shuts stations with failing structures, finds that nobody could, and treats Dana as answered by that — and the books close. Both are decent answers to was the closure right. Neither produces a sentence you could say to Dana about Dana.
Whether such a sentence exists is the question I want asked. Sometimes it does. Suppose the authority runs an accessible shuttle to the nearest usable station for the duration, or moves her line ahead of three cosmetic repairs, or just pays for her taxis for four months. Then someone can sit down with her and say: this landed on you far harder than on anyone else, here is what we did so it would not cost you your job, here is when it ends. She has still lost something, and is still owed the acknowledgment and the explanation. But she was not wronged.
Suppose instead none of that happens, because a shuttle is expensive and Dana is one rider. Then the correctness of the closure does not reach her, because it was never addressed to her in the first place. She was wronged — and what is owed changes shape. It is owed as a wrong. It includes redesigning the closure protocol so that the one accessible station is not simply the first thing to go next time. And "the column was failing" stops being the last word.
That is the practical difference, and it is not a difference in feeling. Two authorities that made the identical, correct decision now owe measurably different things, and you can check which.
What I am not claiming
I am not claiming the closure was wrong; it may well have been the only responsible thing to do. This argument does not rank your options — it tells you what you are left owing once you have chosen. Nor am I claiming any of it is new. That something is left over after a hard decision made rightly is an old observation, and it has good names already: moral remainder, agent-regret. What I think is worth adding is narrower — the schedule. A person-by-person statement of what is still owed and to whom, and a refusal to let "it was necessary" quietly convert a wrong into a cost. And I am not claiming Dana has to accept any of it. She may take the shuttle, stay angry, and never call it square. That is not a flaw in the account. It is what it means for the account to be hers.
Where you can get off
This argument rests on 3 premises you can reject. They are not hedges — each is a real position that reasonable people hold, and rejecting one stops the conclusion following. Here is what each costs.
If what you did fell on one particular person, you owe that person a reason aimed at them — a decision being defensible in general is not automatically an answer to her.
If you think a decision only has to be defensible in general, the page stops at step two and nothing after it follows. You can still hold that Dana's loss matters enormously — you just count it in with everything else, and once the sum comes out right there is nobody left standing outside it with a separate bill.
What the framework calls this, and what it records
Answerability as a directed account-owing relation — held as foundational.
Refusing it costs: The framework loses the directed-obligation form that separates answerability from impersonal appraisal, from party-relative assessment, and from authorization: "must answer to" dissolves into mere affectedness or collapses into an authorization requirement, and no standard of justifiability-to-each-holder-of-an-unsatisfied-claim remains to govern conduct whose addressee did not, or could not, grant permission.
What survives: Standing's party floor, the claim grounds that direct a claim to an addressee, impersonal appraisal and party-relative assessment as separate forms, and authorization as the distinct grantor relation all survive; what is lost is the account of the directed obligation those grounds generate.
You do not get to decide, on someone else's behalf and in your own favour, that they have been made square with you.
Drop this and a decision-maker who has done what could reasonably be asked may treat the matter as finished, so "we handled that years ago" becomes the end of it rather than an offer Dana can refuse. You keep the rest of the argument — the two lists, the account owed — but the person who owes it also gets to say when it is paid.
What the framework calls this, and what it records
The prohibition on settlement — held as defended.
Refusing it costs: Refusing this claim costs the framework's distinct identification of, and bar on, closing another party's moral account in the closer's favor without that holder's valid governing exercise, and takes down the derived beginnings-as-settlement paradigm result (that no beginning can be settled in the name of the one begun) as an application of it. What falls in the credit, grievance, and gratitude sections is specifically the settlement-closure framing: treating the beginning as a spent release or acquittal ("your existence puts you in my debt"), or treating gratitude, flourishing, or benefit-acceptance as a completed grant that closes the account. It does not remove the positive analyses those sections route through the initiation ground (Claim Grounds) and Authorization's counterfeit-grant devices, which survive and independently continue to deny the maker a reciprocal initiation claim, to identify an account owed now without a pre-existence claim, and to bar retroactive or manufactured permission.
What survives: The reserved life-verdict survives as an independent foundational commitment, since the page states it is not derived from Authorization or the settlement argument. Legitimate finality survives — existing parties resolving disputes, holders performing releases, and institutions closing questions of action for coordination while preserving review and the account. Claim Grounds' separate determination of whether a beginning creates later claims and what they contain, Authorization's validity conditions, and obligations grounded in promises, reliance, care, and relations that develop after a beginning all survive.
What is left over comes in different sizes: owing someone for a hard call you had to make is not the same as owing them for something you should not have done.
Reject this and you can still say something is always owed — but it collapses into one undifferentiated apology, and the difference between the authority that arranged the shuttle and the one that did not stops changing anything anyone actually has to do. You also give up the check at the end of the case, since there is no longer a second list to be on.
What the framework calls this, and what it records
Differentiated residue and concentration conditions — held as proposed.
Refusing it costs: Every burden tends to become compensable, or formal reciprocity defeats claims from those who bear concentrated loss.
What survives: G5 responsiveness and the open account survive without this taxonomy.
Those costs are read from the framework’s own record, not written for this page, so they say what it actually commits to. The refusal explorer lets you select several at once.