The directive did not sound brutal: no organized evacuation of Reedtown; residents free to leave on their own; public transport to run as usual.

This is the fictional United City-States of Ends. The neighboring army is shelling Reedtown and will soon cross the river. The written reason for refusing evacuation is that it would signal abandonment of the West Flats and cause panic farther inland. That day the First Executive promises to stand with the town 'to the last person.' The defense subgroup approves the directive two votes to one. The Executive for External Affairs dissents.

The regional government orders buses. The Office of Defense requires military passes, issued three days later. Meanwhile, a separate evacuation on foot takes place at night. Enemy troops enter on the ninth day and hold Reedtown for five weeks. About 61 percent of residents have left. The old, the sick, people unwilling to leave home, and staff keeping the hospital and water station running are among those who remain. One hundred forty civilians die during the fighting and occupation. The division of causes and responsibilities is still being investigated in Year 12.

We cannot charge every death to one directive. Enemy shelling, combat, and delayed passes require separate investigation. But something is already visible in the document: what the country must signal comes before how its people will survive. The order of reasons is wrong.

Saving the country—for whom?

If the state itself is at stake, what cannot be suspended? Few questions sound more compelling in an emergency. In this case, the assessment is 110,000 troops assembled across the Boundary River, capable of crossing within seventy-two hours. A government may have to act before ordinary deliberation is possible. But no time to wait does not mean unlimited authority thereafter.

A state matters because of the lives, rights, and possibilities of cooperation it protects. Its survival borrows its importance from them. If preserving it authorizes silencing those people, blocking their complaints, or demanding that the entire community die with its institutions, the means has consumed its purpose. Losing territory, a capital, or a political formation can be a devastating defeat. It is not a reason to declare that people's lives must end with it.

Reedtown's residents were not forbidden to leave. Yet permission to walk away is not an evacuation route for a sick person who needs a bus. Organizing evacuation need not mean forcing everyone out. It means making departure possible for those who want it but cannot manage alone. Those who choose to stay do not forfeit protection and supplies. Their safety cannot be exchanged for reassurance elsewhere, or made into a flag displayed to the enemy.

These are proposed limits on state power, not a report of existing military law. Nor do they demand omniscience. They demand explanations: why were the passes delayed? What caused each death? An aggressor's responsibility does not erase a defending administration's decisions. Its decisions do not erase the aggressor's responsibility either.

One chain of command, more than one source of authority

Checks on military power need not mean rival armies or several officials issuing incompatible battlefield orders. One force needs a coherent command. A different question is whether its commander also gets sole authority to declare war status, use troops against residents, suspend ordinary constraints, and decide when all this is over.

At two Iron Valley arms factories, a strike brought an army regiment to the gates for six weeks. It set up checkpoints and assisted arrests. The local district had not requested deployment, its assembly opposed it, and the national Joint Assembly had not been asked to confirm it. The executive said it was protecting military supply facilities, not deploying against residents. But workers were passing through those checkpoints. A label cannot decide what an operation is doing.

Domestic use of the army requires confirmation not monopolized by its commanders, specifying scope, place, and duration. A lawful court ruling requiring enforcement can supply that confirmation, limited to its enforcement. This is a functional requirement, not a claim that only one institutional design can work. Confirmation is necessary, not a complete justification: authority, necessity, and the means used still require scrutiny. And separate questions stay separate. The union eventually reached arbitration; the district sought court review; prosecuted workers had access to courts. We cannot therefore declare that every avenue of refusal was sealed. Why arbitration took six weeks, and how a worker suffered a broken bone, remain questions. None of those avenues retrospectively supplies the missing deployment confirmation.

Conversely, moving troops between existing bases without deploying them against residents is not automatically domestic military intervention. Yet two battalions moved into South Bay's naval base a week before its secession referendum invite an explanation. Timing deserves examination; it is not already proof of intimidation.

Peacetime and wartime command arrangements must also be distinct. Concentrated wartime authority draws its reason from the war; peace cannot borrow that reason indefinitely. Moving between the arrangements needs a time-limited, challengeable decision not monopolized by one office. Otherwise special command lasts merely because nobody empowered to do so declares it over.

Emergency powers have the same distinction. Originally, the emergency needed confirmation within seven days and renewal by the Joint Assembly every sixty days. After amendment, the Executive Council could renew it itself, while termination required three-fifths of the Assembly. A later vote favored ending it, 176 to 142, but failed the threshold. Temporary authority becomes durable not only when nobody wants it to end, but when ending requires conditions its holders can evade.

Party affiliation alone does not establish that review is fictitious. Reviewers need information, the ability to judge independently, and power to stop a measure. Different institutional names alone prove little either. If continued war status extends legislators' own terms, their independence on that particular question needs another look.

What if the assembly has been bombed and cannot meet? Alternative confirmation arrangements must be prepared beforehand, not improvised by the leader choosing someone willing to approve him. If both ordinary and substitute bodies are disabled, provisional measures are confined to the shortest pre-specified term and cannot renew themselves. This is no promise that procedures will survive every catastrophe. It denies that their temporary destruction wipes out the obligation to answer afterward. New provisional measures do not permit the executive to shelve an already effective final judgment.

The difference between receiving a complaint and changing someone's position is stark in detention. Six hundred forty people are held under a clause allowing thirty-day terms, renewable by the detaining side; the longest detention has reached 150 days. The Inspectorate can inspect files and demand explanations but cannot release anyone. Constitutional review of the clause is not an independent review of each detention. On the arrangements described, that missing release route is the violation. If investigation establishes an effective individual review not recorded here, those cases must be reassessed on their facts.

Temporary power cannot wait for its own consent to leave

A twenty-year-old from Levee Street hears a familiar promise: after the war. He objects to bearing arms on grounds of conscience, without religious affiliation, and applies for alternative service. His application is received but review is suspended. Refusal to report carries two to five years in prison. He serves for two years. When review resumes, the board answers: service completed; no further review.

Precisely because those years cannot be returned, his application still needs an answer. The question is not whether he must serve next year, but whether he should have been forced to bear arms then. Applications must be available during service as well as before it, without a religious test. Pending review, there must be no compulsory induction or prosecution for the refusal at issue; rejection must be open to court review. Review checks whether the application concerns bearing arms itself and is consistent, not whether an official certifies the person's inner conscience as true. Alternative service can impose burdens, but differences need public reasons and cannot be retaliation. Objection to one war or one kind of mission remains an unresolved question in the paper; this route is not a personal veto over every order.

Emergency power can stay in more courteous forms. A wartime surcharge becomes a reconstruction tax. Surveillance ends when a peace treaty is signed. Requisitioned farmland will be reconsidered 'after the war.' Names change while control remains. Peace treaties may never arrive.

The relevant endpoint is therefore not simply the government's announcement. It is the removal of the attack that made defensive war necessary. In this fictional record, the invading forces have been driven back across the river and hostilities cease the following spring. The original defensive necessity ends there, although formal war status continues for more than another year. Staying on the other bank is a new action needing its own justification, not an extension silently supplied by the original defense.

Powers grounded in that war lose their basis at that endpoint. Their holders must dismantle them, rather than wait for opponents to prove them unnecessary. A peacetime defensive line or public function can be established anew: why now, within what limits, under whose review, and until when? Arrangements can be made in advance so there need be no operational gap. They cannot run on the old war's authority. Ordinary defense expenditure is not all a temporary wartime power; a budget percentage alone cannot tell us what must go. Each arrangement needs its own account of origin and present basis.

Where hostilities recur and no clear endpoint can be identified, authorization still needs an expiry date and independent confirmation. Renewal is a fresh decision. Powers with no term, or only a promise to end at a future peace treaty, must be re-established as time-limited decisions within a pre-specified short period or lapse. An untidy conflict is not an invitation to permanent emergency.

What needs saving is the possibility of people going on with their lives. Reedtown's territory matters; political institutions matter; the young man's two years matter too. The instruments that protect a life cannot require that life to prove why they should ever be relinquished.