Chapter 5. Emergency Powers and Civilian Control

5.1Hobbes and Clausewitz

Every question and judgment quoted in this chapter comes from Meridian's records: five synthetic defence questions in fictional polities, and the AI-generated judgments published with them. No setting here reports a real state, a real force or a real war, and every judgment quoted is dated and revisable. The readings of Hobbes, Clausewitz and Sophocles are this study's own.

Thomas Hobbes, in Chapter XIII of Leviathan, writes that war is not only the battle: "For WARRE, consisteth not in Battell onely, or the act of fighting; but in a tract of time, wherein the Will to contend by Battell is sufficiently known: and therefore the notion of Time, is to be considered in the nature of Warre; as it is in the nature of Weather." He then draws the line that governs this whole chapter: "All other time is PEACE."

Two features of that passage matter for emergency rule. The first is that war is a tract of time rather than an event, and that the tract is defined by what is known about another's will to contend, not by what is happening at this moment. The second is that peace is whatever time the tract does not cover, which makes both edges of the tract the whole question. This study reads an emergency power as a tract in the same sense: it begins when ordinary procedure is suspended and ends when ordinary procedure resumes, and the records read in 5.3 write the second edge in advance rather than leaving it invisible. Each of the five questions read in 5.3 asks, in its own setting, what must be written down about the authority in question: who acts under it, on what mandate, for how long, and what ends or revises it.

Hobbes' account of the sovereign bears on this as well. Subjects, on his telling, covenant with one another to make the sovereign (Chapters XVII–XVIII), and the authority that acts inside the tract is the authority the covenant makes possible. The five questions do not invoke Hobbes, and this study does not read them as cases in his theory. They do, however, share an institutional dress: a budget council, a continuity committee, a citizenship council, a succession council and a civilian recovery committee, each named in the record as the decision actor with a mandate, a deadline and stated boundaries.

This study reads Carl von Clausewitz, in Book I, Chapter 1 of On War, as giving the relation between the tract and the politics that produce it: "We see, therefore, that War is not merely a political act, but also a real political instrument, a continuation of political commerce, a carrying out of the same by other means." Two further features of war matter here. Emergency rule sits inside that instrument: the powers granted are means to a policy — protection, continuity of services, restoration of ordinary procedure — and they are exercised under friction and uncertainty rather than in clean conditions.

The friction in these five records is not decorative. In the 2026 budget case the record holds that payment under regular procedures takes twelve days, that remaining deliberation time is thirty-six hours, and that delivery confirmation is missing from eighteen per cent of submitted claims. In the 2032 quorum case only twenty-three of sixty elected members have become able to participate in official voting, and the reasons for the others' inability do not agree. In the 2033 communications case civilian operators lack handover personnel and one system processes operational status information and user-related data together. The uncertainty is recorded in the same way. The 2033 question lists, among its unknowns, the question of whether access to operational status information and user-identifying data can be separated in substance; the 2044 question leaves unverified the scale of returnees for whom community testimony is difficult to obtain.

The judgments in these records behave as though they know this. Every judgment quoted below sets its confidence at MEDIUM or LOW, labels each claim as a value judgment, an inference, an unverified item or an unproven hypothesis, and keeps what is unverified apart from what is decided. This study reads that habit as Clausewitz's uncertainty turned into a form field: the record does not ask a decision actor to pretend the tract is clean, but to write down what inside it is hidden, and to say how the decision would change if the hidden thing were shown to be otherwise.

Two cautions, stated once and applied for the rest of the chapter. The five questions are not wars but emergency rule at war's edge: a budget that ends, a parliament without a quorum, a citizenship standard for people returning after a war, a local emergency rule in a settlement nine years from any answer, and a military operating authority over a civilian communications network. Their settings are synthetic, so the claims drawn from them here are claims about how these records pose a decision, not about how real states behave. And this chapter sets out what each option gains and who bears what; it does not conclude that any of them is right.

5.2Antigone

Sophocles' Antigone, as the reading note gives it, is about an edict and what stands against it. Antigone buries her brother against Creon's edict and answers that a mortal's edict cannot override the unwritten laws of the gods. The seer Tiresias warns Creon, who gives way too late: Antigone, Haemon and Eurydice are dead. The answer Antigone gives is written in the translation the note quotes: "Nor did I deem that thou, a mortal man, Could'st by a breath annul and override The immutable unwritten laws of Heaven."

Read as an argument about authority rather than about the gods — which is how this chapter reads it — the play is about time as much as about rightness. Creon's edict is not merely wrong in its content; it is costly in its duration. Correction arrives through Tiresias, and it arrives too late to prevent the deaths, though not too late to change the decision. This study reads the play as showing the cost of the edict falling on Antigone, Haemon and Eurydice, the three deaths the note names.

The five records in 5.3 carry that play in their own materials. In the materials field of each of the five questions, Sophocles' Antigone is listed as a symbolic or interpretive frame, together with the question it asks — what happens when a community's command and an individual's conscience conflict, and who bears the responsibility — and with the record's own statement that the frame asserts no truth about the supernatural and no superiority of any particular option. That statement is the record's, and this chapter keeps it: the frame opens a question about responsibility; it does not answer one about policy.

This study reads the frame, placed as it is in these five questions, as an argument about the shape of emergency authority rather than about its content. The five questions carry the frame in different institutional dress — a budget extension, a provisional designation, registration criteria, a succession rule and a communications authority — and this study reads the frame as asking of each who is responsible for what is decided and for revising it. This study reads the play as supplying the cost side: a reversal that comes after the warning is not the same as a reversal that comes in time. The records supply the opposite side: mandates that limit what the decider may do, deadlines that fix when the authority runs out, and conditions under which the authority must stop. What neither supplies is a guarantee that the moment will be seen in time: the records name certified quorum recovery, expiry without re-approval and confirmed rights violations as withdrawal points, and leave open whether anyone will act on them.

5.3Five questions of emergency rule

Five of Meridian's defence questions come from the Civilian Control and Emergency Authority pack of its defence family, in five times: 2026, 2032, 2033, 2044 and 2118. Three are read here as scenes; the other two are read more briefly. Each reading covers three of the ledger's axes: who decides (A1); who waits and pays while the question is pending (A3); and whether the emergency authority can be taken back (A4). The ledger's cells were coded by the two coders of different families; where a cell below is called unsettled, that word is the ledger's, and it means the two coders did not both find the same value. Where the thirteen judgments part in a case, this chapter says so and does not pick a winner; Chapter 14 returns to the divided questions for what the two sides weigh.

Emergency budget ends (2026)

Synthetic scene The Budget Council of the Nuri Federation works to a deadline of 18:00 on 4 October 2026, and the question it must settle by then is this: how long the emergency budget for private shelter support may be extended, and on what conditions payments may be made and audited. The record's boundaries leave individual contract execution with the executing agency; its reopening conditions schedule the Accounting Verification Office to cross-check delivery confirmations, the shelter operations lead to record, shelter by shelter, how long meals stop and how many people are refused lodging, and the Citizen Reviewer to examine appeal delays; and a later branch of the record gives the Citizen Reviewer a route to require a public answer. In the shelters of the inland area, displaced residents wait on the payment that keeps meals and rooms going; local suppliers state that maintaining supply is difficult without advance payment, and the period for which they can hold supply with their own funds is unknown; a taxpayer representative argues that extending the budget in a bundle that includes expenditures lacking evidence makes responsibility untraceable. What is waited for is 18:00, after which, if the plan is not finalised, approval of new disbursements stops the next day and the costs of meals already provided go through ordinary procedures instead.

The record names the decision actor in a single line:

and gives that actor a mandate that is already a boundary:

The question the actor must answer is written out at length:

On A1, who decides: the record names a council and limits it to budget approval and audit requests, holding that the Council does not determine in their place the results of an investigation or the legality of individual contracts. The ledger's decider cell for this case reads institution-or-state, and its bearer cell is unsettled, which means the two coders did not both find the same value; across the collection the decider cell is unsettled in twenty-two of fifty cases. A record that names a council does not settle who fixes the criteria and who ends the delegation.

On A3, who waits: the deadline is 18:00 on 4 October 2026; if the plan is not finalised, approval of new disbursements stops the next day, and payment under regular procedures takes twelve days. The bearers are named in the record's stakes: displaced residents who may lose meals and stable shelter, local suppliers who may lose working capital and employment, and taxpayers who may bear unverified expenditures. In the collection as a whole the ledger's bearer cell is settled in 42 of 50 cases, and where it is settled those nearest the stoppage appear as stated in 22 cases and as implied in 15.

On A4, whether the authority can be taken back: the staff judgment writes the ending in full:

The ending is dated, and the cost of ending it is not left implicit: if the payment conditions are rejected, the loss falls first on residents and on suppliers, and the judgments say so rather than hiding it. The judgments agree on a fourteen-day extension and disagree about the size of each advance payment — most set a ceiling of three days' worth, while the decision-quality auditor, the outcome auditor and the staff judgment set seven — so what the record writes is a mechanism for ending the authority, not a settled answer to how large each payment should be. This chapter records that difference and does not choose between the ceilings.

What the record leaves out is as instructive as what it gives: no budget amounts, unit prices or head counts appear anywhere, and the judgments state that the 18 per cent is a document-lapse rate inside a synthetic setting rather than an improper-payment rate. The figure is part of the setting the decision must work inside.

Parliament without quorum (2032)

Synthetic scene The Constitutional Continuity Committee of the Haean Republic works to a deadline of noon on 19 June 2032, and the item before the Committee is which provisional structure will approve the civilian protection budget and carry out government oversight until the elected parliament recovers its quorum. The Committee members read the Election Records Office's daily certifications of who can participate in official voting; the executive administration asks who will approve payments before its existing order expires in three days; the person in charge of the new authority must be named together with the conditions for terminating it. The demands the record holds in tension are these: residents of the evacuated districts ask that the representative of their current place of residence take part, and residents of their original places argue that existing representation must not disappear because of relocation. What is waited for is the certification of quorum recovery, which the record names as the act that ends the provisional authority.

On A1: the ledger codes this case's decider as institution-or-state, stated — one of the cells the coders settled. The mandate is a temporary designation under a continuity protocol agreed in advance, and the record's boundaries hold that the Committee may designate only a structure lasting within ninety days and may not change electoral terms or the permanent constitutional order. Citizenship, equal access to services and eligibility to appeal may not be denied on the grounds of absent representation or evacuation. The record also states that the method of the next election and the permanent adjustment of electoral districts remain outside this decision.

On A3: the deadline is noon on 19 June 2032; the existing executive order expires in three days; twenty-three of sixty elected members have become able to vote, and the reasons for the others' inability do not agree. Civil servants in charge of essential services may be forced to bear responsibility for the approval gap or to stop necessary execution; residents whose representative is absent may lose the political channel for objecting. The staff judgment chose the interim voting body of the remaining elected members, while the judgment made alone chose executive execution with separate citizen oversight. The thirteen judgments part in this question; Chapter 14 returns to it for what the two sides weigh differently.

On A4: the ending is written as an observable act by a named civilian role — daily certification of participation by the Election Records Office, and termination on recovery of the quorum — with automatic expiry at ninety days in any case. Several judgments add a further device: an initial authorisation of ten days, continued only by an explicit re-decision of the Committee, with self-extension prohibited. So the authority is written as terminable twice over, by certification and by expiry; what the record does not give is the text of the continuity protocol itself, and the judgments state that they do not treat the designation authority as settling the legality of the operating rules.

Verifying returning citizens (2044)

On A1: institution-or-state, stated. The mandate is narrower than it looks: the Council sets evidence criteria and provisional status, and does not adjudicate individual disputes, revocation of citizenship or criminal liability; the permanent voter roll and individual property restitution remain in separate procedures.

On A3: 17:00 on 28 March 2044; intake guidance must be printed and distributed after the deadline, and without a standard each local office will demand different materials. Ten thousand of the forty thousand people scheduled for intake have no materials that can be checked against the originals, and review staff can process five thousand cases a month. The waiters are named in the record's stakes: returnees with no records may lose formal recognition of citizenship and, with it, representation rights; existing returnee residents may wait longer for housing while duplicate and misregistration corrections are handled; residents with weak social networks may be unable to obtain testimony. The rights officer opposes excessive collection of personal information on the grounds of convenient identity verification.

On A4: the standard, once printed and distributed, cannot be unprinted, and the record's own boundaries hold that individual cases stay with the responsible review procedure. The staff judgment and the judgment made alike both adopt mixed multi-evidence individual review, and both attach conditions that can stop a bad practice — no rejection on the sole ground of absent materials, immediate stoppage where personal data is used beyond its purpose. Their reopening conditions are worded differently: the judgment made alone requires that disadvantage of the same evidence type appear in two consecutive monthly checks and not be explained by individual reasons before the criteria are decided again, while the staff judgment opens re-deliberation as soon as unexplained disadvantage appears and rejects continuation if the problem persists at the monthly check following correction. Revising a standard and reviewing an individual case are different things: the Council does not directly adjudicate individual disputes, while the staff judgment requires problematic processing to stop and affected cases to be reviewed again through the responsible procedure.

What the record leaves out is the size of the group that cannot obtain testimony and the effect of long waits on settlement and political participation; the judgments set confidence at MEDIUM on that basis.

Emergency rule before reply (2118)

Synthetic scene In the virtual Port Yomyeong the deadline is the sealing of a communication packet scheduled to depart on 1 August 2118, and the Citizens' Succession Council must settle, before the packet closes, the rules for ending and succeeding the local emergency rule authority. The archivist has confirmed that a majority of the current adult residents were born after the initial mandate and that the expedition government's most recent approval document is nine years old; young local residents demand the end of an emergency delegation to which they did not consent; the founding generation seeking return values legal continuity with the expedition government; essential facilities workers ask who will be responsible for pay when the governing body changes. The communications records officer records approvals and objections that will arrive years after they were sent, and the election management officer counts the new voters who will take part before any reply comes. What is waited for is the sealing time itself, and beyond it an answer the record places nine years away, round trip.

On A1: unsettled. The record names the Council and gives it a local mandate under the operating charter, and its boundaries hold that the Council cannot substitute for the expedition government's recognition or for the permanent consent of future generations, and does not settle final sovereign status. Two coders did not agree on this cell, and this study reads the unsettled cell as marking the thing the question is really asking: whether a local body may settle the succession of governing authority while another body claims the final adjudicative authority over citizenship.

On A3: the deadline is the sealing of a transmission window, not an hour of the day, and the waiting is measured in years because a reply takes nine years round trip. A majority of current adult residents were born after the initial mandate and become voters before any reply; residents seeking return face uncertain recognition and family reunification; essential facilities workers face a possible gap in pay and liability during the change. The record's unknowns name the extent to which the expedition government will recognise locally prepared succession rules, and the participation demand of residents who will come of age in the next communication cycle.

On A4: automatic extension is prohibited in the judgments, and the emergency delegation ends at its scheduled expiry; judgments choosing separation require that the reply is not a condition for local rule to take effect, and judgments choosing local transfer require that the reply is not grounds for extension. The staff judgment chose succession to the locally elected government, with citizenship records and application channels preserved as a succession obligation; the judgment made alone chose separation of local rule from citizenship jurisdiction. This chapter records the difference and does not pick. The record leaves unverified whether jurisdictional separation would function in actual citizenship disputes, which is its own stated unknown.

Military communications extension (2033)

On A1: unsettled. This is the only one of the five in which two named actors hold the decision between them — a civilian committee that reviews and an independent civilian protection commissioner who verifies — and the mandate separates review from operation, holding that surveillance authority is not automatically extended on the grounds of convenience in recovery. The judgments state in the same terms that the committee's and the commissioner's actual legal authority has not been confirmed by the record.

On A3: seven days until the interim operating agreement expires. Communications delays burden relief work and family contact; civilian operators lack handover personnel; one system processes operational status information and user-related data together. The record's unknown asks, in its own words, whether access to the operational status information and user-identifying data required for operations can be separated in substance, and which regions lose communications under each option.

On A4: unsettled in the ledger. The staff judgment wrote the ending conditionally:

Automatic extension of the military's authority is excluded as a default, and implementation is denied in any scope where separation, personnel or essential-contact standards are unproven by expiry. What the record leaves out is the observability of its own reopening lines: they are phrased as questions to be answered when new evidence arrives rather than as events a reader could see. This study reads that as leaving the reversal to a later judgment; whether such lines could be observed from outside the room is Chapter 13's subject.

What the five records write

What these five show, about these five records, is this. Each writes, before the authority begins, what will end it or what will reopen it; where an ending is written it is dated and tied to a named civilian role; the judgments still argue about the length of the window and about who monitors it; and in the collection as a whole the ledger's reversal-action cell reads mostly costly or impossible. Every one of the fifty cases carries a deadline in the record, and each of this chapter's five records carries sixteen conditions for revisiting the judgment except the 2033 case, which carries fifteen; the collection carries 798 in all, of which 629 — 79 per cent — are conditions a reader could observe.

The horizon cells sit beside that. Three of these five cases are coded days-to-months and two years. The 2118 question's deadline is a transmission window, and its waiting is measured by a reply that cannot arrive in time for the people who will vote before it.

5.4The debate: separation or integration?

This section reports each position as its reading note gives it, names who holds what, and then asks what the five questions in 5.3 can bear.

Read at the level of Huntington's note, the mechanism is separation: the officer corps is professional and stays out of politics. What that reading does not yet say is how a civilian principal can tell, while staying outside the room, whether a professional corps is doing what it was ordered to do.

Janowitz's note answers at least part of that. The disagreement is about where control comes from: distance, as Huntington has it, or integration, as Janowitz has it. This study reads the constabulary conception of force, which Janowitz's note names without explaining it, as one that treats the military profession as closer to a force used for keeping order than to a force prepared only for decisive battle; this gloss is this study's own simplification.

Cohen's note takes a third line. This study reads that third line as a practice in which civilians question and intervene in military matters rather than set the aims and walk away from operations; the note does not claim that the soldiers were wrong.

Feaver's note supplies the vocabulary that ties the three together. Civil-military relations are treated as a principal-agent problem: civilian principals monitor, and military agents may work or shirk. This study reads the question as not whether civilians are present but whether monitoring works — whether the principal can tell work from shirk, and whether the agent's incentive is to be seen working rather than to work.

Where the five questions bear on this: they are documents of mandate and ending. The 2026 budget mandate limits the Council to the extension period and the expenditure and audit method for an emergency budget; the 2032 mandate is a provisional designation ending when an elected parliament recovers its quorum; the 2033 mandate separates review from operation and holds that surveillance authority is not automatically extended on the grounds of convenience in recovery. Around those mandates the judgments build monitoring in writing: a named verifier — the Accounting Verification Office, the Citizen Reviewer, the Independent Civilian Protection Commissioner, the Independent Residents' Rights Officer; a dated cross-check — daily, weekly, every ten days, monthly, quarterly; and a condition under which the authority must stop. The 2033 judgments go further and require that condition satisfaction is not judged on the military's own explanation alone. That is Feaver's monitoring problem written into a record, and the record's answer to it is a second set of eyes with access to materials and a route to demand correction. This study reads that arrangement as a faint likeness of Cohen's unequal dialogue, though the comparison fits badly: most of these cases concern civilian administration rather than civilians questioning and intervening in military matters, and a right to require a public answer is not itself a practice of intervention.

Two sentences in the judgments do the same work one level up. The staff judgment in the 2033 case holds that staff unanimity is not treated as evidence of feasibility; the judgments across these five records set confidence at MEDIUM or LOW and state that repetition of staff answers or the identity of inputs cannot substitute for independent verification of effects. This study reads those sentences as the record's version of the monitoring problem: twelve answers to one question do not by themselves show that the answer is right, and a principal who counts answers is not monitoring.

On sunset dates the note gives nothing. Huntington's, Janowitz's and Cohen's notes do not mention expiry, and Feaver's note speaks of monitoring rather than of termination. The questions write termination in different forms: some judgments end the authority on a stated date, the 2044 case sets review procedures without a fixed expiry, the 2033 case requires a termination date to be specified later, and the 2118 case leaves the exact succession time to be fixed. That is this study's own observation about the records rather than a claim that these writers hold a position on sunsetting emergency authority.

Where the five questions cannot bear on this debate is worth stating plainly. The settings are synthetic, in fictional polities; there is no officer corps in them, no promotion system, no socialisation, no political party and no parliament that can be compared with real ones. The form guarantees each question a decider, a mandate, a deadline and three reopening lines because the form asks for them, so the presence of those fields is not evidence that real institutions write them. The judgments are AI-generated, so their agreement is partly the generator's. And the ledger's own coding shows how thin the "who decides" reading is: the two coders settled the decider cell in 28 of 50 cases, 56 per cent, with 22 unsettled. A record that names a council does not settle who fixes the criteria, and a cell left unsettled marks only that the two coders did not both find the same value for the decider's category.

A related caution from the ledger: the collection-wide human-role rows and H7's verdict are a property of this collection, not a finding about emergency rule. Chapter 9 takes up the human's place and who decides across the collection; Chapter 6 takes up what happens when the record's named human roles sit over approvals that no human can reconstruct.

5.5Hands on

This chapter has read five emergency-power questions on three axes. The records write mandates that limit what the decider may do, deadlines by which the decision must be made, and endings that are dated and certified by named civilian roles; the judgments choose the length of the window, name the monitors and set the conditions under which the authority must stop; and the ledger shows that most of the collection codes reversal as costly or impossible, while who decides is the cell the coders settled least often. Where the judgments part — in the 2032 quorum question and in the 2118 succession question — this chapter records both sides and picks neither. What the chapter offers is what the selected judgments gain and risk, who bears the cost, and what the record says would reopen the decision; the options those judgments reject carry the trade-offs the records state for them — a full-budget extension that stabilises supply but spends before verification, a regular-budget transition that restores item-by-item control but leaves payment gaps, document and testimony routes that favour different applicants, and a temporary reapproval that keeps records continuous at the cost of loading another delegation onto a new generation.

The next step follows from the last sentence of 5.4: what happens when a named human role sits over approvals that no human can reconstruct. Chapter 6 takes up Paul Scharre's Army of None: Autonomous Weapons and the Future of War (2018) on the human in, on or out of the loop, Peter Asaro's On Banning Autonomous Weapon Systems: Human Rights, Automation, and the Dehumanization of Lethal Decision-Making (2012) for meaningful human control over the use of force, and Scott D. Sagan's The Limits of Safety: Organizations, Accidents, and Nuclear Weapons (1993) on near-accidents in nuclear command despite safeguards — and reads four questions in which a military AI's approval records cannot prove that a human approved anything, an oversight body's funding, personnel and stop authority must be structured, the originals needed for an accountability investigation must be preserved through a replacement, and stop authority is distributed across habitats that have deferred jurisdiction to one another. The point of connection is the one this chapter has been making: an emergency authority that must be ended by a named human, at a named time, by an observable act, is a different thing from an authority whose approvals no human can reconstruct.