Chapter 8. After the War: Names, Records and Return

8.1Return

The Odyssey is the oldest story in this study about going home. Homer's Odysseus returns home twenty years after leaving for Troy; his household has waited; and his return is a matter of recognition and of what has changed while he was away. The note gives no more detail than that, and this chapter will not supply more. What the note supplies is enough to name a problem this study reads across Meridian's aftermath settings: a return is not the same as an arrival. A person can arrive anywhere. A person can return only where something has kept going in the meantime — a house, a claim, a record, a household that will or will not recognise them — and that something decides what the return is worth.

The aftermath settings in Meridian's defence collection each have a committee, a deadline and a budget. In these settings somebody must decide what counts as the evidence of a right to return, and somebody must decide what happens to households that stay; a missing person's fate can remain unsettled while both happen. Those decisions are made under time pressure, with incomplete information, and on behalf of people who were not in the room when the decision was drafted. This study reads The Odyssey's gift to a book of questions as a shape rather than an answer: the one who returns meets a place whose account of the past has moved on without them.

This chapter reads five of Meridian's fifty defence questions on that shape. They run from 2026 to 2114. Unverified video disclosure (2026) asks how unverified material about relief facilities should be submitted for external verification. Missing-person record conflicts (2030) asks how a correction committee should handle two hundred and forty pairs of records that may concern the same person. Records to open (2033) asks what access regime should apply to thirty thousand handed-over postwar administrative records. Return or stay (2041) asks how a reconstruction budget should be allocated between households that wish to return and households that wish to remain. A restored voice (2114) asks what evidentiary status restored materials should have in descendant compensation review when no surviving witness remains.

Each question names a decider, a deadline, options with trade-offs, and what is missing; each carries lines that would reopen the decision; each is followed by thirteen AI-generated judgments that Meridian published with it. This chapter reads the five on the axes the whole book uses: who decides, what is listed as missing, what can be undone once done, and when the judgment was made. Its subject is what the records write down about the time between the question and the answer.

8.2Justice after war

The transitional-justice literature has five works in this study's bibliography, and each addresses a different part of the problem.

Ruti G. Teitel's Transitional Justice (2000) sets out the forms that justice takes after a political transition. The note gives five of them: criminal, historical, reparatory, administrative and constitutional. The point of listing them is that justice after a regime change is not one act but a family of acts — a trial, an official history, a payment, a dismissal from office, a rewritten constitution — and that they can be pursued, or not pursued, separately.

Martha Minow's Between Vengeance and Forgiveness (1998) takes the responses to mass violence — trials, truth commissions and reparations — as its subject, and places them between two poles named in its own title: vengeance and forgiveness. This study reads Minow as arguing that the ordinary instruments are neither of the extremes and that the choice among them is a choice about what a society is willing to remember and what it is willing to let go.

This chapter reads Hayner's subject as the institution that sits between the two extremes Minow names — a body that finds and publishes facts without convicting anyone — and the fact that a truth commission must decide, in its own rules, what it will say about individuals.

Gary Jonathan Bass's Stay the Hand of Vengeance: The Politics of War Crimes Tribunals (2000) is about the politics of the criminal form. This chapter reads Bass's question as one about the cost of enforcement as much as about law: what a state will risk, in its own soldiers, to bring an accused person to trial.

Elster's contribution to this chapter is the pairing of motive with constraint. A decision about who is punished or paid is made by people who have motives — the note gives three, emotions, interests and reasons — and who work inside constraints, and the same decision can look different once the constraint is named.

Read together, this chapter reads the five works as addressing the aftermath of political change, with some institution in a position to try, to record, to pay or to dismiss. That reading is this study's own; the notes do not attribute a shared premise to the five authors.

Meridian's five settings sit inside that same stretch of time rather than before it. A repository that has received thirty thousand documents has not yet decided who may see them, and its mandate forbids publishing statements in the records as established facts or as guilty findings. A correction committee facing two hundred and forty conflicting record pairs has no authority to establish death or to decide whether two records concern the same person. A compensation board in 2114 cannot accept a restored speaker as a personality or as actual testimony. A verification committee in 2026 cannot adjudicate criminal responsibility and holds three copies of a video without the original recording file. An authority in 2041 can set a budget allocation and cannot compel any household to return or to remain.

Where the records bear on the five works, this study reads them as bearing at the base. Hayner's choices about naming perpetrators and about access to records presuppose an access regime that somebody set in advance; Records to open (2033) is that advance moment. Elster's question about who is purged or compensated presupposes a process that can tell one record from another; Missing-person record conflicts (2030) is the process that has not yet decided which records belong together. Bass's limit — the reluctance to risk soldiers — has no counterpart here, because none of the five settings holds an enforcement apparatus at all: there is no army in these questions, only committees with mandates, deadlines and budgets. Where the records cannot bear, this chapter says so. These are synthetic settings in fictional polities; they contain no outcomes, no institutions that continued, and no testimony about what any of them accomplished. These questions describe what a decider was asked to write down before anyone could know what would happen.

8.3Records and names

Synthetic scene. Two of the four settings in this section are written as scenes; the other two are read from the records without a scene.

The Fictional Eden Truth Records Repository works in the weeks before the first viewing acceptance day set in the handover custodial agreement. Thirty thousand documents have arrived from a closed temporary administrative body, and sample inspection has verified that entries about missing-person movements and informant names sit in the same documents, with the truthfulness of individual statements unverified. The record's reopening lines name the roles the work ahead would need: a person in charge of access, who keeps the queue of family applications and publishes waiting times; a person in charge of protection, who verifies exposure reports after every viewing session; and an independent records auditor, who samples de-identified documents each quarter. Families of missing persons wait for a whereabouts verification that gets no easier with delay; informants wait to learn whether their names will appear; local researchers wait to learn whether structural patterns will ever be examined. The clock runs to 18:00 on 14 September 2033.

The question the repository must answer is written in the record:

The record lists what it does not know, and the first of the three items is the one that decides whether any regime can be calibrated:

Axis A2 asks what a question lists as missing. This one lists three unverified items: the range of identifying information needed to connect documents, the specific exposure risk the informants face, and the amount by which family applications will exceed detailed review capacity. These three name quantities — a range, a risk, a gap between demand and capacity — but the record does not fix the definitions or the access arrangements that would let anyone outside the room observe them.

Axis A4 asks what can be undone. The record's own option set is candid about the direction of that axis: one option can examine responsibility structures broadly, but the work of covering identifying information takes long and some families lose connecting clues. The published judgments converge on a family-priority regime and attach stop conditions; one of them states plainly what publication would cost:

This chapter reads that sentence as the hinge of the whole aftermath literature. A regime can halt further disclosure; it cannot recall a name that has already been read. The ledger codes the undoing of this decision as impossible, though the record does not say which words both coders read to find it. The reopening lines the record carries would make the decider reconsider the regime when delay, harm or lost connecting clues are verified. The asymmetry here is not a side issue. It is what makes the deadline matter.

The second scene is set four years earlier and nearer in time to a reader: the fictional Sirin Public Verification Committee, in an information crisis concerning relief facilities. A video circulating in the city asserts that a relief shelter unjustly excluded residents, and requests to cancel support have risen. The Committee holds three copies with different editing sequences and an operations log; it has not received the original recording file, and it has not established the authenticity or the circumstances of anything. Two teams of experts are available for review by tomorrow; masking the faces and voices still visible in the video requires separate personnel and time. The applicant wants rapid disclosure; relief shelter workers fear that incomplete materials will harden their reputation; the source is concerned about identity inference and reluctant to submit the original; residents who claim to appear demand that their explanation be reflected. Support agencies finalise next week's relief volumes immediately after the deadline, which falls at 20:00 on 11 September 2026.

Here A2 is at its most concrete: the missing thing is an object, and its arrival would change what verification can mean. The record's other two unknowns are the circumstances before and after the edited segments, and the production and transmission path of each copy. A4 bites differently in this setting than in the repository setting, because the thing that cannot be undone is not only a name but a story already in circulation. The record's own option trade-off says so:

A public summary, once issued, joins the copies already circulating. Masking can be done before publication and cannot be done afterwards to copies already taken. The ledger codes the undoing of this decision as unsettled — a cell where the two coders could not agree — and the record does not say why. This chapter reads the setting as one in which the choice of disclosure mode is partly a choice about which failures are recoverable.

Two further settings carry the same problem at different distances from a reader.

Missing-person record conflicts (2030) concerns the virtual Haini region, where postwar record integration has found two hundred and forty pairs of records that may concern the same person. Names and birth years overlap; some whereabouts information conflicts; automatic merging has stopped. The Committee may set correction procedures and how records are marked, and may not establish death or identity. Eighty in-person reviews are available for the next review period, and basic family support is maintained even in an unconfirmed state.

The unknown is a count, and the Committee's mandate bars it from establishing death or identity for itself; cross-checking may still turn up linking grounds. The record's other two unknowns — the existence of additional material held by families who cannot be reached, and the specific effect the discrepancies have had on each family's whereabouts investigation — are of the same kind: they are about people and events outside the institution's view. This chapter reads that as the structural difficulty of the missing-person problem. A correction procedure is designed by an institution that cannot, by its mandate, establish the fact the families want established, and the records that would settle the fact are held by parties the institution may not be able to reach. The published judgments converge on annotating conflicting records and cross-checking them sequentially, on the reasoning that preserving clues and keeping erroneous links reversible outweighs the speed of correction. A4 is therefore written into the design rather than discovered afterwards: the record's boundaries require preserved source data and modification histories, and the judgments make that traceability the condition of reversible correction. The ledger codes this decision's undoing as costly, and the reopening lines would have the decider reconsider when correction cancellations, misidentified links or unresolved family burdens are verified.

A restored voice (2114) pushes the same problem past the point where any witness is left. The Haerin Generational Records Compensation Board has received one hundred and twenty generative memory restored materials submitted by descendants. They are inference results made from letters and administrative records; their production history has been verified, and the historical truthfulness of their contents has not. Four specialist staff are available to cross-check originals, and six months run to the first review.

The unknown is the one that decides the evidentiary question, because it is about influence rather than about content. The record's other unknowns are the scope of consent an owner of a non-public family record would grant, and whether independent cross-checkable materials will newly be found where originals are rare. A4 here is not about disclosure but about belief. This study reads the record's own option trade-off as naming the risk directly:

A production history can be recorded and checked again; a content claim, once presented to a compensation applicant as context, may change how a family reads its own past, and the record lists that influence among the things it cannot verify. The published judgments converge on accepting restored materials only as investigative leads, on the reasoning that verified production history is not verified truthfulness, and on the condition that a defect in one material does not reject an entire compensation application. That restraint costs scarce staff time, delays compensation for other applicants, and can leave a descendant who paid for restored material without proof; the original-centred option would spare the verification load and leave descendants whose harms left no records carrying a heavier burden of proof. The ledger codes the undoing of this decision as costly, and the reopening lines would have the Board reconsider when common generated-to-original errors, withdrawals of consent or systematic review gaps are found.

Across the four settings this chapter notes only what the ledger codes for undoing: impossible for the access regime in 2033, costly for the settings in 2030 and 2114, and unsettled for the video setting in 2026. The records give no ranking of unknowns by verifiability and no test of dependence between A2 and A4.

8.4Return or stay

Synthetic scene.

The Barim Return and Resettlement Authority sets the allocation of a reconstruction budget in the weeks after a voluntary intention survey. Households wishing to return and households wishing to remain in the place of refuge are both numerous; households that did not respond to the survey have not been recorded as holding no preference, and property-rights verification is still running. The secured budget cannot complete both the restoration of the water-supply and treatment network in the return areas and the expansion of long-term housing at the place of refuge; each project carries start-up costs that are hard to recover if the plan changes midway. The record's reopening lines name the roles the work ahead would need: a facilities auditor to verify actual construction costs and service commencement schedules each month, a regional coordinator to verify service queues and access objections every other month, and a person in charge of residence choices to re-confirm voluntary intentions each quarter. Those wishing to return say that without infrastructure the right of choice is formalistic; those wishing to remain worry that return-centred investment leaves them as temporary residents. After 15:00 on 2 April 2041, construction and housing-acquisition contracts for the two areas will be signed.

The record offers three allocation methods, each with its own cost:

Several features of this setting deserve the reader's attention, and none of them recommends one of the three methods.

First, the decider is structurally outside the choice it funds. The Authority's mandate states that it sets only the budget allocation and cannot compel residents to return or to remain, and the boundaries add that support guarantees voluntary, fully informed choices and does not make any choice of residence a condition for waiving rights. The households decide; the Authority decides what is possible for them to decide between. The Odyssey's shape appears here in administrative form: the returning person meets a place whose conditions were set by others, and this study reads the household that stayed as bearing the conditions of investment made in someone else's direction.

Second, the option that keeps both choices alive pays in speed. All thirteen published judgments, including the judgment made alone, chose the method of household choice support with minimum investment in both areas, with confidence recorded as medium or low rather than high. This study notes that convergence as a fact about the records; it does not treat it as evidence that the method was right. What the judgments' own reasoning shows is the shape of the trade: the delay is accepted, recorded group by group, on the ground that a method which closes one group's choice base first cannot be justified by its speed.

Third, the record writes its own revision signals in the form of roles and cycles rather than in the form of outcomes: monthly cost and schedule checks, bimonthly reviews of service queues and access objections, quarterly re-confirmation of household intentions; the published judgments add a mid-point and end-point review at twelve and twenty-four months.

The ledger codes the horizon of this decision as years, which is one of the thirteen cases so coded among the fifty, and the undoing of it as costly: start-up costs in both areas would be spent before any switch could be made. This chapter reads that combination as the defining feature of an aftermath decision. The setting asks for a two-year allocation whose start-up costs are difficult to recover, made for households whose intentions the same record admits are unverified — how they will decide once concrete service and housing conditions are presented. The Odyssey's household waited twenty years, and the note makes the return a matter of recognition and of what has changed meanwhile. The households in this setting must decide where to wait on the strength of conditions nobody has yet built.

8.5The debate: truth, justice or repair first?

The question this chapter puts to its five settings is one this study asks rather than one the five works divide over: must the truth be established before justice can be done, must justice be done before repair can begin, or can all three run at once?

Teitel's five forms — criminal, historical, reparatory, administrative and constitutional — do not, on the note's account, come with a required order. The note gives the list and no ordering; whether the forms can be chosen separately is a question this study puts to the records rather than one the note answers. Meridian's questions take that form, each isolating one instrument and asking for its rules in advance: an access regime, a correction procedure, an evidence criterion, a disclosure mode, a budget method. None of the five settings is asked to sequence truth, justice and repair. Each is asked what one instrument may say and to whom.

Minow's study places trials, truth commissions and reparations between vengeance and forgiveness, and this chapter reads Minow as holding that the middle instruments are neither an evasion nor a compromise but a distinct form of response to mass violence. Hayner's truth commissions are the clearest case: an institution that establishes and acknowledges facts without convicting anyone, and that must decide in its own rules about naming perpetrators, reparations and access to its records. Here the records bear directly. Records to open (2033) is the access question a truth-commission analogue must answer before it can write anything: the repository's mandate forbids publishing statements in the records as established facts or as guilty findings, and its options trade family priority against structural breadth against confidential investigation. Missing-person record conflicts (2030) is the question of how an institution handles conflicting administrative records when it cannot settle which of them concern the same person — near Hayner's question about access to a truth commission's records, but not the same question. A restored voice (2114) is the extreme case: an institution that must decide what counts as evidence when no witness survives and the descendant has paid to produce restored material.

Bass's limit does not transfer easily, and this chapter says so. None of the five settings holds an enforcement apparatus, a contingent to deploy, or an accused to arrest. Unverified video disclosure (2026) concerns a committee that cannot adjudicate criminal responsibility and holds no original file; the boundary states that discipline of shelter operators and the support agencies' contract decisions are undetermined. Bass's limit — the reluctance to risk soldiers to bring an accused person to trial — is therefore only half-bearable here: these records show the cost of setting up a fact-finding process, not the cost of enforcing one. Where the analogy fails, this chapter notes the failure rather than stretching the work to fit.

Elster supplies the reading that best fits the five settings, and it does so at a distance. The judgments published with these five questions pair a stated priority with a stated constraint at an earlier stage. Each judgment names a priority in its own terms: the priority of families' irreplaceable time losses, the priority of preserving investigative opportunity, the priority of reducing the loss of irreversible identity exposure. This chapter reads those as reasons recorded in the judgments, not as evidence of the deciders' motives in Elster's sense. Each then names a constraint: a review capacity of six thousand documents, four specialist staff and six months, two expert teams and the time cost of masking faces and voices, an eighty-case in-person review ceiling, a secured budget that cannot complete two projects at once. This study reads those pairings as Elster's problem arriving early. The decision about who is compensated or punished is downstream of the decision about what a record may be taken to prove; the motives and constraints of the earlier decision set the range of the later one.

Where the questions bear and where they cannot must be stated plainly. They bear on the prior question — what an institution may say, to whom, on what evidence, and by when. They carry a deadline in every case; they list what is missing in every case; they name a decider in every case; and, by their form, they write lines that would reopen the decision. Those are properties of the questions, not findings about wars, and the form of Meridian's question is why they are present — Chapter 2 sets that out, and Chapter 9 reads the decider row across all fifty. The questions cannot bear on the literature's outcomes. No record here tells a reader what a truth commission established, what a tribunal convicted, what a reparations programme paid, or whether a family that received a whereabouts verification was satisfied. The settings are synthetic; the judgments are AI-generated, dated and revisable, each written in September 2026; and in each of these five cases all thirteen chose the same option. The questions hold judgments made before implementation and no account of what followed; this study groups the five works under its own heading of justice after political change. Neither substitutes for the other, and this study sets them beside one another rather than using either to settle the other.

The comparison this chapter can make is narrower than the literature it is set beside, and narrower than a reader might hope. Meridian's aftermath questions show what a decision record writes down when an institution must open a file, correct a record, restore a name's evidentiary status, disclose an unverified video or allocate a budget across two futures. They show what that record lists as missing, what it admits cannot be recalled, and when it says the judgment would change. They cannot show what any of that was worth.