1.1Data and method
The records are NADIR's public list of question–judgment pairs, fetched by GET on 6 October 2026 (UTC) and kept in this study's folder. The list holds 1534 pairs. Every one has the status "hypothesis", none carries axes or evidence keys, and the list's evidence table is empty. The studio files each pair under one of 10 layers and one of three phases of its research frame: erosion (2016 to 2025), collapse (2026) and transition (2027). Of the four authorship labels in the list, "NADIR hypothetical decision exercise" is carried by 496 pairs. Every exercise has exactly two options, is set in a situation the studio marks as hypothetical, and is unreviewed. The studio's note on its method says that this edition adds 496 agenda facets across 62 explicitly hypothetical situations, using 29 reusable two-option decision frames, and the list bears this out: 62 distinct situations, 29 distinct pairs of option labels.
The selection rule is one sentence. This paper reads every hypothetical decision exercise in the layer "fairness", which the studio labels fairness and exchange, in the two phases after the erosion, collapse and transition. The rule gives sixteen pairs, eight in each phase. The layer holds 174 pairs, 42 of them in those two phases. Of the 26 left out, 22 are episode interpretation drafts without options, and 4 are what the studio calls blackbox conditional model interpretations; these four are set in situations of another kind, which the studio calls episode transpositions, and none of their option labels is a pair the exercises use. The 32 fairness exercises in the erosion phase are not read, so that every pair read sits at the frame's end or its new start. None of the sixteen is read by any of the 6 earlier NADIR studies in this series.
The sixteen pairs fall into exactly two situations, eight in each, and no other pair in the list belongs to either. In the first (the studio's situation 059, in the collapse year), extra labour and personal costs that the official records did not keep are raised during a final settlement. In the second (situation 029, in the transition year), the gap in authority and reward widens between those who took part in a restart first and those who joined late. Each pair has a question, a judgment of five sentences and two options, each option with a label, a benefit, a cost and a condition of use. Of the questions, 6 are noun phrases, 6 statements and 4 end with a question mark. None names a real person, organisation, brand, place or event, and no line of the appendix had to be withheld.
Two cautions from the series' rules shape the reading. The first concerns the frame. The first sentence of every judgment is the pair's decision focus, word for word; the other four are identical in all eight judgments of a situation. Measured by runs of five words that recur in the same field of three or more pairs, 85 per cent of the judgments' runs are frame, against the series' threshold of 30 per cent; the questions, the decision focuses and the options stand at 0, 0 and 1 per cent. This paper therefore reads each pair's own words in its question and decision focus, reads each situation's four shared sentences once, and treats no shared sentence as evidence that two pairs agree. The second concerns the options. Having two options, each with a condition of use, is a property the studio guarantees and is not a finding; what the options say is. The sixteen pairs use 11 distinct pairs of option labels, 10 read pairs share theirs with another read pair, 8 of them across the two situations, and every pair of labels used here recurs elsewhere in the list, in between 6 and 42 pairs. The options are read as the studio's stock, set beside each question.
The records are in Korean. This study's editorial session put every string into English, sentence by sentence and with nothing added; the Korean stays the original. Every block quote is a sentence or field of that English, word for word, with a source line naming the pair by a short title and the field. Every count comes from a script that reads only the kept records; counts of wording were made on the Korean by fixed word lists printed beside each count. The paper does not test whether the exercises are sound, gives no advice to anyone settling accounts, dividing authority or making a claim, and reads only how a published set of exercises frames one kind of decision.
1.2Settling up: the work that left no record
The eight exercises of the collapse year are hypotheses set in an invented organisation at its final settlement, and what follows says only how they frame a decision. They frame the settlement as a contest between proof and memory in which the record, the obvious test of a claim, is itself under suspicion. The situation is set out in the same four sentences in all eight:
It is a situation in which, during the final settlement, extra labour and personal costs that were not left in the official records are raised. The need for proof and the reality of unrecorded contributions collide. A standard for verification is needed, but a burden must not be treated as never having happened merely because there is no record of it. The practice of the time and the confirmation of other people should be reviewed together, while a consistent procedure of judgment is kept.
The frame grants both sides. A standard of verification is needed; the absence of a record does not show the absence of a burden. Between them it sets two kinds of evidence other than documents, the practice of the time and what other people can confirm, and one requirement on the decision, that the procedure stay consistent. What it does not supply is a rule for how much a confirmed burden is worth. The frame tells the organisation how to establish that a claim is real, not what the claim should receive.
The exercises' own words then explain why the record is missing, and they place the reason inside the organisation:
Costs that became hard to prove because they were handled as a favour
The responsibility of an environment in which informal sacrifice was taken for granted must also be examined.
A favour, by its nature, is not invoiced. The exercise does not treat the lack of proof as the claimant's failure to keep receipts. It treats it as the product of an environment in which informal sacrifice was expected. Here the debate on invisible work, Daniels, Invisible Work (1987), and Hatton, Mechanisms of Invisibility (2017), applies closely. This paper reads the exercise as naming one of Hatton's mechanisms at the scale of a single workplace: a habit of treating extra effort as goodwill, which is cultural in Hatton's sense, removes the effort from the record that the settlement will later consult. In the exercise the invisibility becomes a problem of fairness at a precise moment, when the organisation ends and the record becomes the basis of payment.
Two exercises take the next step. The first asks whether one standard of proof is fair to all claims:
Is it fair only if every claim is assessed at the same level of proof?
The nature of the costs, and whether they could be recorded at the time, may have differed.
A single standard looks like the very model of equal treatment. The exercise asks whether it is, given that some costs could have been recorded when they arose and others could not. This paper reads the question as raising the possibility that equal treatment of unequal records reproduces the inequality the records already contain. The second exercise counts what the search for proof costs:
The personal burden that arises again in the search for proof
It must be considered that the time needed for verification is also a limited resource.
The word "again" carries the argument. The person who bore the unrecorded burden is asked, at the end, to bear a second one: to search for evidence that the first ever happened. The costs that Bolino and Turnley (2005) traced to help given beyond the role, in overload, stress and strain at home, are costs of the help itself. The exercise adds a cost that falls later, when the help must be proved; that addition is the exercise's hypothesis.
The exercises do not treat recognition as unlimited. One sets the line directly:
Recognising informal contributions and accepting unlimited claims are not treated as the same thing.
Setting a scope that can be reviewed and a procedure for objections can reduce both risks together.
The two risks are those named in the frame: refusing what was real, and paying what cannot be checked. The exercise answers both with a procedure, a scope that can be reviewed and a way to object, rather than with a figure or a ceiling. The same turn appears where the exercises consider who gains from the procedure itself. One asks that the standards of the settlement be explained so that those who already know them gain no advantage, and its decision focus states the principle:
The accessibility of the procedure also affects the fairness of the outcome.
The debate on procedure treats the fairness of a process and the fairness of a result as distinct, even where they interact, as in Brockner and Wiesenfeld (1996) and Colquitt (2001). The exercise runs them together. This paper reads it as saying that a procedure open only to those who know its rules produces an unequal distribution, so that the procedure is part of what is distributed. Another exercise makes the same point about speed:
The cost of speed may be shifted onto those with less bargaining power.
Its question asks whether small claims are made to be given up so that the settlement can end quickly. Speed is presented as neutral, and the exercise finds a distribution inside it: those who can least afford to press a small claim are the ones who drop it. A further exercise turns the conflict between claimants back on the settlement. Where one person feels another's claim shrinking their own share, its decision focus holds that reading the dispute only as competition between individuals misses the problem of resources and standards.
The last exercise of the collapse year gives the clearest answer to the series' question:
The reasons for a judgment are left even for claims that are not accepted.
If only the outcome is conveyed, later review and the recovery of trust become difficult.
A refused claim is where a settlement is most likely to be contested, and the exercise asks that its reasons be kept. The reason given is not courtesy but what comes after: review, and the recovery of trust, both of which need to know why the decision went as it did. Colquitt (2001) found informational justice to be a dimension of its own. This paper reads the exercise as treating it as the hinge of revision: a settlement can be looked at again only if its reasons survive it. What would make the decision change, in this framing, is a claim reconsidered against reasons that were written down when it was refused.
The collapse year's exercises therefore frame fairness at the end as the work of hearing claims well. A claim or a share is named in the own words of 5 pairs, all of them in this situation, and a procedure, a standard, reasons, explanation, agreement or consent in 8 pairs, half of them here; no pair in either situation names equality or an equal share anywhere in its text.
1.3Starting over: the claims of those who came first
The eight exercises of the transition year are hypotheses set in an invented restart, and again they are read only for how they frame a decision. They frame the restart as a problem of time: a contribution made early asks to be paid in authority later, and the exercises answer by keeping the two accounts apart. The situation is set out in the same four sentences in all eight:
It is a situation in which the gap in authority and reward is widening between those who took part in the restart first and those who joined late. The early bearing of risk and the present contribution produce different standards. Early contribution should be recognised, but it must not be made a reason to monopolise every future opportunity. The return for having borne the risk can be settled separately, and the responsibilities and performance of present roles can be evaluated by published standards.
The frame's key word is "separately". Early risk is real and is to be recognised, and it is to be paid in its own account, not through a permanent claim on authority. Present roles are to be judged by standards that are published. The first exercise names what goes wrong when the accounts are merged:
Merit earned at the start must not become a right to leave new contributions unexamined.
Its question puts the danger in a phrase: having been there from the start can cut short the evaluation of the present. Seen through Adams (1965), this is an exchange stretched over time, in which one party's early input is offered as payment for every later comparison; that reading is this paper's. Where Deutsch (1975) expects a group to settle on equity, equality or need according to its conditions, the exercises do not settle. They accept that early risk deserves a return and that present work deserves evaluation, and they decline to let one stand in for the other.
The exercises then ask where the early account is kept:
If the settlement rests only on later memories, the narratives of contribution collide with one another.
Its question is whether there is agreement on who bore how much of the early risk. Here the record's absence works the other way from the collapse year. At the end, the missing record hurt those who had given without being asked. At the restart, it lets each person's memory of the early days grow to fit a present claim. Hellmann and Wasserman, The First Deal (2017), treat the division of founders' shares as a costly bargain in which an equal split is the simple choice and a differentiated one must be negotiated. This paper reads the exercise as identifying what makes that bargain hard after the fact: without an agreed account, there is nothing to negotiate from but competing stories.
One exercise carries the logic of the collapse year into the restart:
Past concessions are not repeated as present unpaid labour.
A burden already borne and a contribution newly requested need separate consent.
The pair is the restart's counterpart to the favour that could not be proved. Someone who conceded once is not thereby committed to conceding again. Consent given for the old burden does not carry over to the new request. This paper reads it, beside Daniels (1987), as a rule against the process by which an extra effort becomes an expected one and drops out of view; the exercise places the safeguard in asking again.
Two exercises apply the separation to people's positions:
An alternative is needed that separates respect for contribution from present fitness.
Economic rights and the everyday authority to decide may not rest on the same grounds.
The first concerns an early member whose role no longer fits; the second, the confusion of owning a share with running the work. In both, a single status is split in two. Respect is owed for what was done, and a role is judged on what is needed now. A right to a return is not a right to decide each day. This paper reads these as the same move as the frame's "settled separately", applied to persons rather than payments.
The remaining exercises of the restart look at those who arrived later and at who decides:
Someone who worked trusting only the published conditions may learn of other rules too late.
If decisions are made only among close founders, other experiences are hard to reflect.
The first describes a late joiner who learns that the conditions on which they were hired were never the whole of the rules. The second states the question the exercise asks about the restart's fairness, that it also depends on a procedure in which conflict can be spoken of. The exercise does not speak of having a say in the sense of Lind, Kanfer and Earley (1990). This paper reads it as placing a related concern at the level of the organisation's design: a restart decided among people who already trust one another leaves no point at which someone else's experience can enter. A further exercise asks that new members be able to see a path to authority, on the ground that a ranking present ability cannot change may lead people to leave rather than to cooperate.
The restart's exercises therefore frame fairness at the new start as keeping the past in its own account. Five pairs across the set name the early, the start, the past or what came first in their own words, and all five are in the restart, as are the 3 that name authority, rank, rights or ownership.
1.4The options set beside the questions
The options beside these sixteen questions are the studio's stock, and they do something the questions do not. The questions and decision focuses of 12 pairs name the settlement's own terms: contribution, proof, reward, a claim or share, authority, merit or risk. The option labels name them in 1. The option labels speak instead of review, checking or confirmation (7 pairs), of agreement or mediation (3), of roles (3), of recording (2), of sharing or explaining (2), and of workload, staff or timing (2).
Several fit their question well. The pair on reasons for refused claims is set beside a choice between recording the reasons for a judgment, and even the conditions for changing it, and recording only what execution needs, which is the question's own concern. The one pair whose labels speak in the settlement's terms is the first exercise of the restart. It sets rewarding contributions by kind against a single common standard, and the cost the studio gives the common standard returns to the collapse year's concern:
Special burdens and invisible contributions may be left out
Others fit loosely. The exercise on the personal burden of finding proof is set beside a frame for capacity:
Reduce the workload to secure room for recovery
Its alternative is to add staff and support to keep the scope. The question concerns a claimant's second burden; the options concern the organisation's workload. The exercise on explaining the standards to those who do not know them is set beside a frame for disclosure whose benefit, as the studio states it, is to secure time for the other party to prepare. Whether the options fit or not, a reader who weighs them is mostly weighing a decision about how the settlement is run, not about what anyone receives. This paper takes that conversion to come from the record set's construction, in which a stock of 29 frames serves all 496 exercises, rather than from anything the exercises argue.