Where does accountability end when a return is promised

20,000 people who departed from state control want to return, but the records that would separate simple departure from harm inflicted on residents are incomplete. Blanket pre-screening defers a large share of returns by four years. Starting from the premise that serious crimes cannot be granted immunity, how much liability for which acts should be reduced in legislation while preserving victims' demands for fact-finding and compensation?

Decision year: 2045 Deadline: 60 days until applications to participate in reintegration open Decides: Parliamentary Truth and Reintegration Legislative Committee World: Continuum Severity: Critical Decision

Mandate

Decision question

The three options

---

1Synthetic scene

Synthetic scene

The committee room on the fourth floor has the shape of a room that was never meant for this. A long table, four microphones that still hum, a window onto a street where nothing particular is happening. It is 2045. The committee is drafting the eligibility clause of the time-limited reintegration law, and the plenary session is waiting.

A staffer sets three files on the table, in the order they will be argued about.

The first is the departure roster. It is thick and it has gaps. Names are listed against dates and units, and then the list thins out where the records stop, and then resumes again as though nothing had happened. The staffer says the roster shows who left. It does not show who did what while leaving.

The second file is thinner. It is the joint statement issued by representatives of affected residents, printed and bound, with the signatures on the second page. The statement is short and it does not ask for punishment. It asks for three procedures to exist before anything is signed: fact-finding, compensation, and objection. One member reads the third of those out loud twice, as if the word itself were the dispute.

The third file is a single sheet with a bar on it. The bar is labelled 5,000. It is the number of people the review body can process in the first year. Beside it, someone has drawn, by hand, a much longer bar for 20,000, and beneath that a note reading four years, underlined twice.

A member picks up the roster, holds it at the edge so the pages fan, and asks the room the question that the whole statute turns on:

"If a name is on this roster, does that mean it caused harm?"

Nobody answers immediately, because the honest answer is that the roster cannot say. That silence is the problem. Part of the departing group does not want to disband and return to civilian life, and the records that would tell one story apart from another are incomplete:

The residents' representatives have already told the committee what they will and will not accept, and they said it in one sentence:

And the capacity sheet, the smallest file with the largest consequence, says this:

A second member turns the roster face down on the table, which changes nothing about it. "If we screen everyone first," they say, "we are telling everyone else to wait four years for a decision nobody has promised them." The first member answers, still looking at the residents' statement: "And if we don't, we are telling their neighbours that the roster is close enough to a finding."

The drafter puts a pen on the blank clause. There is a box where the immunity scope goes, and three ways to fill it, and the pen stays where it is.

1.1What the scene is showing

The scene is not an argument for any of the three options. It is a way of seeing the three constraints at once, because the constraints are what the decision has to be made inside of.

The roster is the record problem: incomplete, and incomplete in a direction that is not neutral. Gaps in a record do not say "no harm occurred." They say "nothing was written down." Any clause written into the statute will be read against those gaps for as long as the statute runs.

The residents' statement is the legitimacy problem. Affected residents have not asked for a veto and they have not asked for trials as a precondition of everything else. They have asked for three procedures to be real: fact-finding, compensation, objection. A reintegration law that grants immunity without those three procedures is not merely unpopular with them; it removes the mechanism by which they would ever find out what happened.

The capacity sheet is the timing problem. 5,000 in the first year against 20,000 applicants, with blanket pre-screening pushing a large share of returns back four years. That is not an administrative detail. If the period without a return pathway lengthens, the relationships and skills needed to return to civilian life may be lost. The delay itself is a harm, and it lands on people who have not been shown to have done anything.

The three files do not fit on the table at the same time without something giving. That is the shape of the decision.

---

2The question

2.1Who decides what, and by when

One body decides, and it decides three things in one act. The Parliamentary Truth and Reintegration Legislative Committee writes the eligibility, investigation, and compensation provisions of the time-limited reintegration law and sends them to the plenary session. The plenary session then votes. Applications to participate in reintegration open 60 days from the decision. That deadline is not a target; it is the point at which the abstract scope of immunity becomes a queue of named people.

The mandate is specific about what the committee is doing. It is submitting provisions, not granting immunity to anyone. Individual cases will be decided later by an independent review body that does not yet exist in statutory form. So the committee's act is a framing act: it sets the outer boundary of what anyone downstream can decide.

The mandate also states a limit that is not subject to negotiation:

That sentence fixes a floor before the argument starts. No option before the committee can grant immunity for serious crimes, because the mandate says the instrument cannot do it. This is why the question is not "immunity or no immunity." It is how far down the ladder of acts the reduction of liability goes.

The decision question, quoted in full in the head above, has two halves. The first half asks up to which acts conditional immunity will extend. The second half asks which of the three named options will be finalized. The two halves are not the same question: a committee could in principle write a clause that is narrower than any of the three, but the record presents the three as the choice to be made.

2.2Where the authority does not reach

Three boundaries sit outside the committee's reach, and they are worth stating plainly because they constrain all three options equally.

The first: immunity for serious crimes is outside the scope permitted by this time-limited bill. No option can reach it, and nothing said below should be read as though it could.

The second: harm cannot be established on group affiliation alone. A person's presence on the departure roster is not a finding. The converse also holds and is equally uncomfortable — the absence of a record of harm is not a finding either, and the statute has to operate in that double uncertainty.

The third: reconciliation or a face-to-face meeting cannot be forced on a victim. Whatever the law does about liability, it cannot make a victim sit in a room with an applicant as the price of anything. This has a practical consequence that runs through the options: testimony and disclosure cannot be obtained by pressuring the harmed party, so any incentive the law creates has to work on the applicant's side of the table.

There is a fourth limit, less formal but sharper. The committee can write conditions into the law. It cannot write capacity into existence. Whatever procedures the statute requires — independent review, victim notification, objection handling, compensation determination — must be staffed within 60 days and must absorb a first-year load of 5,000 against a backlog of 20,000. A clause that requires a procedure the state cannot staff does not protect anyone. It just converts a right into a queue.

2.3The two stakes

The record states why this cannot be deferred, and it states it as two risks pointing in opposite directions.

The risk of going too wide:

The risk of going too narrow, or of going slowly:

These two sentences are the whole decision in miniature. One describes an irreversible closing. The other describes an erosion that is also, once it has happened, very hard to reverse. The committee has to weigh a loss that is sudden and permanent against a loss that is gradual and cumulative, under conditions where it cannot yet measure either.

2.4The three options and their costs

The options are not ranked by the record, and neither is this piece.

Option O1 — "Conditional immunity for status and command defiance only." Immunity reaches the status violation of having departed from state control and the defiance of return orders. Individual acts that caused resident harm are excluded from the immunity scope and investigated separately. Serious crimes are always left in existing judicial procedures.

What O1 buys is a clean boundary and a revocable commitment. What it charges is that the boundary must be drawn case by case by an investigation, on records that are incomplete, for 20,000 people with 5,000 slots a year.

Option O2 — "Include individual minor acts in exchange for disclosure." Immunity extends to individual minor acts, conditional on the applicant disclosing facts and an independent review confirming them. Serious crimes are excluded regardless of disclosure or cooperation.

What O2 buys is breadth: more applicants have a reason to come forward, and more testimony enters the record. What it charges is that the price of that breadth is paid in a currency the state cannot currently verify, because whether harm with insufficient records can be independently confirmed is an open question.

Option O3 — "No immunity; adjust sentencing and return support." No immunity at all. Sentencing and return support are adjusted instead, and judicial procedure runs unchanged for every act.

What O3 buys is the total preservation of judicial access and the removal of the concealment risk that comes with any immunity bargain. What it charges is participation: without an incentive, fewer applicants engage, and the informal status of people who departed stays unresolved for a long time.

2.5What cannot be judged yet

Two unknowns are recorded as unverified, and they are the two variables every judgment in the record circles.

U1 is an evidentiary question. It asks how much of the harm can be established at all when the records are incomplete. U2 is a behavioural question. It asks whether the design of the law moves people — whether disclosure conditions, review speed, and victim procedures change what applicants actually do.

Neither can be answered before the decision, because both are answered by the law running. That is the structural difficulty. The committee must choose an option without knowing whether its central premise holds, and then watch for whether it holds. Everything in the next chapter — the conditions, the stop conditions, the milestones — is an attempt to convert these two unknowns into something observable rather than something assumed.

What cannot be judged now: whether O2 would actually increase participation; whether O1's narrower scope would actually reduce it; whether O3's preservation of procedure would actually cost returns; whether any of the three is superior in any measurable sense. The record is explicit that none of these are asserted. The present state of the question is pre-decisional.

---

3The prediction

3.1The shape of the thirteen

Thirteen AI-generated judgments were written against this record. They do not divide evenly and they do not divide by a large margin.

Seven choose O1. Six choose O2. None chooses O3. All thirteen are set at MEDIUM confidence.

Three features of that distribution matter more than the tally itself.

The first: nobody picks the option that would be easiest to defend on procedural grounds. O3 preserves every judicial proceeding and grants nothing. It is the choice that most obviously satisfies the residents' demand for fact-finding, and it receives zero votes. Every judgment rejects it, and most reject it for the same reason — the four-year delay that blanket pre-screening of all 20,000 would impose is itself a serious cost, and one that falls on people who have not been shown to have done wrong.

The second: the split is 7–6, which is a margin of one judgment. A record that divides this narrowly is not reporting a consensus. It is reporting a genuine disagreement between two defensible readings of the same three constraints.

The third: every judgment is at MEDIUM confidence, and the ground is the same across both camps, though each judgment states it in its own words — the two unknowns, U1 and U2, are unresolved. No judgment claims to know what the participation data will show. No judgment claims to know how much harm can be confirmed. Both camps are making a bet about which loss is more recoverable, and both say so.

3.2The seven who chose O1

The O1 camp is not one argument. It is one conclusion reached along at least three different routes: reversibility, boundary clarity, and burden of proof.

Multi-Perspective Final Judgment states the position in its cleanest form:

Its basis is the most important sentence in the whole record for understanding why the tally is 7–6 rather than unanimous:

That is an explicit statement that the record does not treat this as a counting exercise. The synthesis chose on the ground of which loss, once taken, cannot be undone. It rests on the incompleteness of records and on the residents' procedural demands, and it concludes there is insufficient basis at this stage to extend immunity to individual acts.

Review Coordination reaches O1 by a route of reversibility rather than certainty. Its decision is a limited and revocable commitment chosen over a comprehensive one that is hard to reverse, because the records cannot immediately distinguish simple departure from harm. Its accepted loss is specific about what the choice costs:

It goes further than most, acknowledging that fact-finding itself imposes a cost — statements, face-to-face meetings, re-experiencing — on both applicants and victims, and weighs that against the loss of permanently extinguishing redress for harm not yet surfaced. It treats the second as the greater loss.

Evidence and Sources Review argues that preemptively granting immunity up to acts under incomplete records would be difficult to reverse, whereas a return pathway limited to status violations can be opened now. It accepts a loss that the other judgments are less willing to name:

It also accepts delay in some returns to life. What it refuses to accept is the permanent loss of fact-finding concentrated on victims. It sets confidence in the boundaries of O1's rights and medium overall confidence in the participation effects, because those effects are unverified.

Domain Analysis argues from attribution: benefits and costs of the choice fall on different parties, so the harder-to-recover loss is avoided first. It notes that only 5,000 can be reviewed in the first year, and therefore accepts the risk that cases of individual acts will be delayed, judging that delay modifiable compared with permanently closing victims' judicial access and opportunities to give testimony.

Independent Challenge frames the trade as an asymmetry between two kinds of delay:

It is also the judgment most explicit about what would make it change its mind, and it points at O3 rather than O2.

Authority and Legal Review argues from the clarity of the authority boundary. Where records for distinguishing actors are incomplete, widening the immunity scope can close off even harm not yet confirmed, so the scope with the clearest boundary of authority is chosen. It separates the effect on personal responsibility from the effect of support for return to daily life, and it accepts the risk that this discourages applications relative to O2.

Outcome Review argues that the benefits of immunity and the losses of extinguished responsibility are attributed to different parties, so the irreversible lapse of victim rights is avoided first. It is careful to say that reduced participation is not an observed fact but an unverified variable, and it sets a specific precondition: if the procedures and the first-year operating plan for 5,000 are not ready, the O1 bill's introduction is deferred rather than pushed through.

Across all seven, the accepted loss is roughly the same package: investigative burden, delayed returns, possible lower participation, a 5,000-person bottleneck, and the administrative cost of not resolving contentious cases quickly. What none of them accepts is permanently closing the possibility of redress for harm that has not yet been confirmed.

3.3The six who chose O2

The O2 camp reaches its conclusion through the same three constraints, but reads the delay constraint as binding rather than secondary.

Operations Planning states the choice most directly:

It adds the two guards that distinguish O2 from blanket amnesty: status and command defiance are reviewed quickly against standardized criteria, and other minor acts receive conditional immunity only if they pass the applicant's disclosure and an independent review. Serious crimes are never granted immunity, regardless of any disclosure or cooperation.

Its accepted loss is the mirror image of the O1 camp's: the risk of incomplete disclosure and victim distrust, plus the administrative burden of individual reviews. In exchange, fact-finding, compensation, and objection procedures become essential components of immunity rather than optional add-ons, so that the loss is not shifted unilaterally to victims. It also explicitly avoids the structure in which blanket pre-screening delays return to daily life for a long period.

Judgment Synthesis gives the most careful statement of what O2 accepts:

It then bounds that acceptance: those losses are limited by revocable conditional immunity and a victim objection procedure, and it explicitly frames O2 as not a blanket amnesty that places expanded participation above victims' rights to fact-finding, compensation, and objection.

Decision Quality Review argues that the trade-off is only justified if compensation and objection procedures and independent review are placed alongside immunity, so that risk is not transferred to victims alone. It accepts that incomplete disclosure or concealment may mix into immunity, that victims will distrust the procedures, and that individual review is burdensome — and in return it takes the reduction in delayed returns.

Decision-Maker's View grounds itself in the distributive question: participation benefits and the losses of concealment and distrust are attributed to different parties, and which loss is actually larger has not been confirmed. It prioritizes revocable conditional immunity with victim objection procedures over a position of no immunity, which risks long-term delay of the return pathway. It also names an option the others leave implicit: detailed criteria for minor and serious are not in the record and must be separately codified in the law.

Improvement Design frames O2 as the choice that keeps both the reintegration pathway and victims' opportunity for fact-finding intact within incomplete records. It accepts that some minor acts may be wrongly exempted and that victim distrust will follow, and it places revocation and compensation alongside so those losses are not transferred to victims unilaterally. It adds a condition that neither camp disputes: if disclosure is incomplete, some minor acts will be wrongly exempted, and the law must be able to refuse or withdraw immunity for that individual.

Solo Judgment argues for limited O2 excluding serious crimes and attaching disclosure, independent review, objection, and compensation. It names its losses plainly, and notes where they land:

Operations Planning's uncertainty statement is the clearest admission in the camp that the choice is a bet:

Across all six, the accepted loss is: concealment risk, incomplete disclosure surfacing late, victim distrust, individual review burden, misclassification risk, and some slow processing. What none of them accepts is the long-term closure of the return pathway that blanket pre-screening or no incentive structure would produce.

3.4What neither camp accepts

It is worth marking the agreement as sharply as the disagreement, because the agreement is larger.

No judgment chooses O3. No judgment accepts blanket pre-screening of all 20,000 as a precondition for return support. No judgment accepts immunity for serious crimes. No judgment accepts that harm be established on group affiliation alone. No judgment accepts forcing reconciliation or a face-to-face meeting on a victim. No judgment treats return support during the wait as a finding of no wrongdoing or as immunity for harmful acts. Most judgments keep disclosure, independent review, victim notification and objection, and compensation as gates before immunity takes effect; two omit compensation from that list. Most make immunity revocable when a material omission in disclosure is confirmed; two provide only for suspending or refusing finalization. Several separate the question of return support from the question of immunity.

This shared ground is larger than the disputed ground. The disagreement is not about whether victims keep their rights; all thirteen say they do. The disagreement is about how wide the conditional immunity clause should be before those rights mechanisms are proven to work.

The two camps also agree on what they do not know. All thirteen state that the scope within which harm with insufficient records can be independently confirmed is unresolved, and that how much the conditions of the bill change voluntary return and participation in testimony is unverified. Neither camp asserts superiority on participation rate or on recovery of harm. Both set confidence at medium for the same reason.

3.5How each side sees the other

The camps characterize each other accurately, which is a sign that the record is not caricaturing the dispute.

The O1 camp's description of O2's strongest argument:

The O1 camp's reason for not accepting it: because the capacity to confirm harm is unclear and review capacity is limited, the risk of hardening incomplete disclosure into final immunity is not accepted at this time.

The O2 camp's description of O1's strongest argument:

The O2 camp's reason for not accepting it: O1 also retains the investigation burden of distinguishing acts, and O3 may prolong return delay long-term, so a limited O2 is selected rather than a narrow one.

And the position both camps reject but both take seriously, the O3 argument:

Independent Challenge's version of this is the most dangerous to its own position: in a state of incomplete records, immunity for status and command defiance can itself become a pathway for concealing acts of resident harm, and there is no guarantee that revocation of conditional immunity actually works. That is an argument that O1's revocability may be formal rather than real.

3.6The three stop conditions

Written across thirteen judgments, the stop conditions collapse into the same three signals. They appear in almost identical language in both camps, which means they are the parts of the design that do not depend on which option is chosen.

Signal one — a material omission in disclosure. When independent review confirms an applicant materially omitted or falsified facts, that applicant's immunity procedure stops; the harm-related case is routed to ordinary or individual investigation; and any immunity effect already granted goes to revocation review. Domain Analysis:

Signal two — the victim objection procedure backlogged for 90 days or more. When that happens, new immunity finalizations are suspended while staffing, notification, and processing procedures are redesigned. Critically, non-immunity return support for safety and livelihood is not automatically revoked by the stoppage. Authority and Legal Review:

Signal three — the gap between applications and review capacity exceeding twice the approved first-year cap. With a cap of 5,000, twice that cap is the threshold named in the record. When applications exceed capacity by that much, automatic approval is refused, expansion of intake is stopped, and the bill is re-deliberated instead of widening the target. Decision Quality Review:

A fourth rule appears across the judgments as a narrowing rule: if serious crimes or resident-harm acts are repeatedly reclassified as status violations or minor acts, the classification method is suspended and the scope of the law is narrowed. A companion rule works the other way, on omission rather than classification, and Improvement Design states it plainly:

These signals are not predictions that any of these things will happen. They are control rules to be written into the law now, while writing is cheap, before a case arises in which stopping is expensive.

3.7One timeline

The judgments' conditions, stop conditions, and milestones can be laid into a single sequence running from the decision to the first full re-judgment. Times run from the decision, and where a milestone follows the opening of applications, that is stated.

Day 0the decision

The eligibility, investigation, and compensation provisions of the time-limited reintegration law are submitted to the plenary session. The plenary debates the three options and finalizes one. The immunity scope, the closed list of eligible acts, the criteria separating status violations from resident-harm acts, the independent review body, victim notification and objection procedures, and the compensation route are written into the statutory text. If these cannot be established within 60 days, the implementation of the immunity clause is held in abeyance rather than passed without them.

Day 60the deadline, and the implementation readiness review

Applications to participate in reintegration open. Before they open, one last confirmation is made. Operations Planning:

Outcome Review frames the same checkpoint as a legal test rather than an administrative one:

The application window, the independent review, and the victim notification and objection procedures launch together on this day. Support for housing, employment, and status restoration can be provided while awaiting review, kept separate so that it does not mean immunity for harmful acts or a determination of no wrongdoing. At intake, a baseline is recorded: number of applications, number of fast-tracked status violation reviews, number of conversions to individual investigation, number of victim objections.

90 days after applications openfirst observation

The first scheduled reading of the machine. What is checked: application count, completed reviews, confirmed material disclosure omissions, cases referred to individual investigation, the length of the victim objection backlog, and the referral of excluded cases. Also recorded: the rate of requests to supplement fact disclosures. This is the first point at which the disclosure gate can be seen working or failing, and the first point at which any of the three stop signals could appear.

6 months after applications openparticipation check

Differences in return applications and testimony participation by condition are examined for the first time. This is the earliest reading of U2, and both camps watch it: the O2 camp to see whether wider scope produced wider participation, the O1 camp to see whether narrower scope cost participation it had already accepted as a loss.

12 months after applications open, at the first-year 5,000 review pointre-judgment

The largest checkpoint. What is compared: actual throughput against applications, change in return and testimony participation, classification reversals between status violations and separate acts, return delay, and the gap between the 5,000 review capacity and actual case classification. At this point the committee decides whether to maintain, narrow, or expand the scope that was chosen on day 0. If any stop signal appeared earlier, this re-judgment happens before the scheduled point.

Any pointearly reopening

The three signals do not wait for a schedule. A confirmed material omission, a 90-day objection backlog, or a gap exceeding twice the approved first-year cap of 5,000 triggers re-deliberation immediately. The record is explicit that these are observation signals to be collected in the future, not results confirmed now.

---

4The branches

Two event branches are recorded against this decision. They diverge from the same initial state. They are separate possibilities, not a sequence, and neither is an observation schedule or a hint about a correct answer.

4.1Branch B1 — the victims' joint material arrives

This branch puts two new things on the table at the same time: allegations of harm that were not previously documented, and acts that do not appear on any existing list. It is the harder of the two branches, because it stresses the statute at exactly the point where the record is weakest — classification.

Under O1, this branch lands first on the classification machinery. Harm suspicions convert into individual investigation, which is where O1 said they would go. The acts that are not on the existing list are the problem: O1's closed list of eligible acts has no entry for them, and a closed list that encounters uncategorised conduct has only two responses — hold them for individual review, or admit them by analogy. Several judgments anticipate this and require that acts whose facts are unclear are not treated as status violations but held for individual review. Authority and Legal Review goes further and states the failure mode as a stop rule: if repeated cases are identified in which a serious crime or a separate harmful act is wrongly classified as a status violation, the application of the law is narrowed and the decision is reopened.

Under O2, the same branch lands on the disclosure gate. Minor acts not on the list are precisely what O2's exchange was designed to capture — disclosure in return for conditional immunity. But capturing them requires that the minor nature of each act be established, and the branch supplies allegations rather than findings. Judgment Synthesis's stop condition fires here:

Improvement Design's version is the stricter one: a serious omission of fact found in independent review stops the immunity procedure for that applicant immediately and triggers a review of whether an already confirmed immunity is to be revoked. Decision Quality Review adds that if cases are found in which serious crimes are classified as minor acts, the application of that classification is suspended at once.

The branch also puts pressure on the 90-day objection clock. If the new material produces a wave of victim objections, the objection procedure absorbs them or it does not. If it does not, new immunity finalization stops in both camps, because the objection procedure is the mechanism by which the residents' stated condition for accepting reintegration is actually met.

There is a further possibility, and it is the one that would end the argument rather than settle it. Independent Challenge holds that if a future observation emerges that independent confirmation is nearly impossible or that major omissions recur, its own position is abandoned and the switch is made to O3:

In other words, branch B1 has a failure state in which one judgment, Independent Challenge, would abandon its own position and switch to O3. Several other judgments name the same endpoint from the other direction: if material omissions or misclassifications recur and status immunity also obstructs accountability determination, a switch to O3 is made. The branch does not automatically lead there. It is the branch in which that destination becomes discussable.

What B1 would tell the committee is narrower than it first appears. It would not tell them which option was right. It would tell them whether the classification line — the line between a status violation and a harm act, between a minor act and a serious crime — can be drawn at all on this record. That is U1, arriving as an operational problem rather than a theoretical one.

4.2Branch B2 — the participating group signs a pledge

This branch is quieter and, for the O2 camp, more uncomfortable. It removes pressure rather than adding it. A group that was expected to require an incentive to cooperate signs a pledge to take part in independent statements and compensation procedures, and it does so without demanding or accepting any forced meeting with victims — which is exactly the boundary the record set.

The pledge respects the boundary already written into the law. Domain Analysis:

The question branch B2 raises is whether the bargain O2 proposed — disclosure in exchange for conditional immunity — was ever the thing securing cooperation. If cooperation arrives without the exchange, then the exchange's main justification, the widening of participation and testimony, has less to rest on, while its main cost, concealment and incomplete disclosure, remains. That is the condition under which the exchange's stated benefit — the widening of participation and testimony — becomes the point both camps must test rather than the point either camp can assume.

The record supplies the exact reconsideration conditions for this. Evidence and Sources Review:

Read against branch B2, the first half of that sentence is now in play from the opposite direction: if participation arrives without O2's exchange, then the participation effect of the exchange is unproven in either direction, which is the unverified variable U2 that both camps set as a reconsideration condition. Multi-Perspective Final Judgment sets the same test from the O1 side, requiring that future signals confirm independent verification operating reliably, victim objections processed within 90 days, and return and testimony participation substantively improved over O1 before a limited expansion of O2 is reconsidered. Branch B2 would weaken the third of those three conditions rather than strengthen it.

For the O1 camp, branch B2 tests something different: whether the pledge holds when the first material omission surfaces, or when the first objection backlog grows, or when the first applicant is asked to disclose acts that will not fit the closed list. A pledge is not a procedure. The residents' stated condition was fact-finding, compensation, and objection procedures — mechanisms with staff, deadlines, and consequences — and a pledge supplies intent but not machinery. Evidence and Sources Review's conditions before applications open require statutory criteria, an independent review body, a victim objection procedure, and a path to compensation to exist in law before the window opens. Branch B2 does not substitute for any of them.

Both branches, read together, show the same underlying tension in different directions. B1 supplies too much unclassifiable material for the machinery. B2 supplies too little incentive for the exchange. One tests whether the line can be drawn. The other tests whether the bargain was needed.

Neither branch is a sequence. The record states that the two event cards are alternative branches diverging from the same initial state, that they do not occur in sequence automatically, and that they do not indicate an observation schedule or a correct answer.

---

5The check

Status: Before the deadline; awaiting verification.

The observations below are fixed now, before the decision is made and before the law runs, so that the outcome can be checked against them rather than against whatever happens to be most visible at the time. No scores are assigned. No option is marked correct. Each row names what will be observed, who checks it and when, which judgments it would bear out if observed, and what it would reopen.

Two features of this table deserve to be said out loud.

The first is that several rows cut both ways. A confirmed material omission would be read by the O1 camp as evidence that disclosure cannot be trusted at scale, and by the O2 camp as evidence that the revocation machinery works as designed. A participation increase would be read by the O2 camp as vindication and by the O1 camp as a cost they had already priced in. The observation alone does not settle which reading is right; what it settles is that a reading is now owed.

The second is that the row for the 90-day backlog bears out no camp in particular. That signal is designed to be neutral between the options, because it tests capacity rather than scope. If it fires, the question of which immunity clause was chosen becomes less urgent than the question of whether the objection procedure can function at all — which is the procedure affected residents named as a condition of accepting reintegration in the first place.

The state of the check is unchanged: before the deadline; awaiting the check.

---

6The questions that remain

The record closes with three watch questions. They are handed back here without answers, because they are the questions a reader is more likely to face than the specific statute above.

6.1Who caused, and who could have prevented

In this matter the committee has to name that actor class without a complete roster, and the statute's answer will be a classification rule rather than a list of names. In your own organisation or community: when something has already gone wrong and the records are partial, who do you write down as responsible for having caused it, and — the harder half — who do you write down as responsible for having been able to prevent it? The second list is usually shorter in practice and longer in principle, and most accountability frameworks only document the first.

6.2Evidence and interpretive authority

Here the separation is meant to be structural: applicants disclose, an independent review body classifies, victims are notified and may object, compensation is determined by its own route. Each step is a different holder of authority, so that no single party both produces the evidence and decides what it means. In your own setting: does the body that gathers the facts also decide what the facts signify? If it does, then every condition you have written is a condition on one actor's self-review, whatever the policy language says.

6.3Obligations from new rights

Recognizing a right to return creates a review obligation and a queue. Recognizing a right to object creates a notification obligation and a threshold beyond which the procedure counts as backlogged — here, 90 days or more, at which point new immunity finalization stops. Recognizing a right to compensation creates a funding obligation and a determination procedure. In your own organisation: when you next grant a right, write down the three obligations that follow before you announce the right, because the obligations are what will be tested first and the right is what will be blamed when they fail.

6.4The frame the decision sits inside

This is the subject the decision-maker argues over, and the record's creator states that they have no conclusion about it. Neither does this piece. The O1 camp accepts delay and lower participation to keep redress open. The O2 camp accepts concealment and distrust to keep the pathway open. Both accept real losses, both name who bears them, and neither claims the loss they chose is smaller — only that it is more recoverable.

And the symbolic frame the record chose for exploring the procedure itself:

This is an exploration perspective chosen by the creator. It makes no claim about supernatural truth or the superiority of any particular option. Read as a practical question, it asks what the fact-finding procedure itself costs: statements to be written, meetings to be attended, accounts to be re-lived, a waiting period during which neither side knows the outcome. Several judgments acknowledge that cost directly — that fact-finding can impose the burden of statements, face-to-face meetings, and re-experiencing on both applicants and victims. A procedure that recovers the truth can still damage the people it is run for, and the design question is whether the damage is minimized by running fewer procedures, more of them, or different ones.

---

7Sources and notices

7.1Background documents

OECD · Strategic Foresight Toolkit (2025). Scope: "A method of exposing assumptions and constructing alternative futures to check policy. It does not guarantee the probability of occurrence or the technological realization of the matter below." Checked 2026-09-07. Used as background to the topic and the institutional question; it does not prove the future events, numbers, or technological realization of this matter.

UNESCO · Ethics of Neurotechnology (2025). Scope: "An ethical principle for neural technology and mental privacy, autonomy, and identity. It is not evidence that memory duplication or consciousness upload is possible." Checked 2026-09-07. Used as background to the topic and the institutional question; it does not prove the future events, numbers, or technological realization of this matter.

UN · Declaration on Future Generations (2024). Scope: "A declaration reflecting the interests of future generations and long-term impacts in institutions. It cannot determine the preferences of future generations." Checked 2026-09-07. Used as background to the topic and the institutional question; it does not prove the future events, numbers, or technological realization of this matter.

7.2The source text of the dilemma

[Scenario 3.1.2.1] Informal negotiations with armed factions that departed from an AGI command system: conditions, patterns of development, and strategic management options — a synthetic scenario from a user-provided Notion source, created and viewed 2026-09-07. Scope: "A synthetic scenario from a user-provided Notion. It is the source text of the decision dilemma and the exploration perspective, and does not prove future events, probabilities, or technological realization." It is the root of the problem discovery; its future events and formulas are not treated as verified facts.

The record notes how it was reworked: "Quote from source: induce disarmament and social reintegration over the long term" — reprocessed by converting the source's long-term reintegration goal into parliamentary choices on immunity scope, victim rights, and review capacity rather than operational tactics.

7.3Symbolic frame and value frame

The story of Solomon's judgment — an exploration perspective chosen by the creator: "What losses does the very procedure of revealing the truth inflict on the parties?" It makes no claim about supernatural truth or the superiority of any particular option.

Value or norm — "The benefits and costs of each option are attributed to different parties. Which loss to accept remains undecided." This is the subject the decision-maker argues over; the material's creator has no conclusion.

7.4Notices

Every person, institution, event, and figure in this piece is a synthetic setting. The Synthetic state, the Parliamentary Truth and Reintegration Legislative Committee, the departure roster, the residents' joint statement, the 5,000-person review capacity, the 20,000 applicants, the 60-day deadline, the four-year delay, and both event branches are part of a constructed scenario. The background documents are sources for topic discovery and are not evidence of future facts or probabilities. No figure here is a forecast, and nothing here should be read as describing a real state, a real negotiation, or a real population.

The thirteen judgments quoted throughout are AI-generated. They were written 2026-09-07 against version 1.2.0 of the record, and their confidence levels are as recorded above — MEDIUM across all thirteen. They are not the views of any person, institution, or body, and they are not the views of the decision-maker.

This piece is an AI-written expansion of the record. It adds structure, scene, and connective analysis to material that already exists; it does not alter the record. Where a judgment, an option, a premise, an unknown, an event branch, a watch question, or a material use is quoted, the wording is reproduced from the record exactly. Where a scene detail, a line of dialogue, or an interpretive passage appears, it was added for this piece and is not part of the record.

The record's two event cards are alternative branches diverging from the same initial state. They do not occur in sequence automatically, and they do not indicate an observation schedule or a correct answer.

The background document premise is stated in the record as follows: "All figures, institutions, events, numbers, and future technologies in this matter are a synthetic setting. The background document is a source for topic discovery and is not evidence of future facts or probabilities."

Translation note: this is an English translation of the Korean record, which is the original.

Input hash: 07967b3a2c1aa4a0dc07779002aebd8d74b11d043875cbdd6bf796319507207b