The first four essays were all locating work: what war is, where it sits, how it relates to the structures around it, what divisions run inside it. Only once that is done can standards be set.

This essay is the hardest part of the series — four theorems constraining war. Together they form cultivation's complete protective mechanism inside a wartime situation.

1. The Shared Grammar

Here are all four side by side.

Theorem One · Cannot not start. Beginning a war must satisfy a cannot-not: every route through negotiation and maneuver has been exhausted, subjectivity has been forced to its limit, and without fighting, cultivation will be entirely broken in this field.

Theorem Two · Cannot not expand. Once a war has begun, every expansion must pass a fresh test at the level of Theorem One. "We are already fighting" may not be used to smuggle anything through.

Theorem Three · Cannot not be directed toward ending. Every action in a war must point toward its ending — the moment the original cannot-not is dissolved and cultivation regains the conditions to continue.

Theorem Four · Cannot not be open to questioning. A warring party holds no privilege of going unquestioned. The questioning may come from its own interior, from inside the other side, from third parties, or from future history.

All four begin with the same words: cannot not.

This is not stylistic symmetry. "Cannot not" performs a specific structural function — it strips the subject of its sense of authority.

The typical grammar for handling war is "I decide," "I permit," "I judge," "I have the right." That grammar puts the warring party in the position of sovereign: he chooses how to fight, when to stop, whether to accept questioning.

"Cannot not" cancels that position. It says: the party is not choosing to go to war, it has been forced into a corner by structure and cannot not go. It is not permitting questioning; it has no position available from which to go unquestioned.

This grammatical discipline is not a preference. It is the shell the principle lives in. Grant the warring party the grammatical standing of a permitter, and you have structurally opened the door to the Kingdom of Means.

2. Theorem One · Cannot Not Start

The beginning of a war is a threshold. Cross it, and cultivation undergoes extreme-state handling. So the threshold has to be set very high.

The test has two conditions, and both must hold at once.

External condition: routes through negotiation and maneuver are exhausted. Contests of judgment, bargaining over interests, diplomacy, third-party mediation, the making and repairing of agreements — every non-violent route has been tried and has failed. Not abandoned after one or two attempts, but genuinely exhausted: under the existing structure there is no non-violent path left.

Internal condition: subjectivity has been forced to its limit. One's own position as a subject has reached the point where it will be lost without resistance — the carrier of cultivation itself is about to be eliminated. Note what this does not mean: not injured interests, not wounded standing, not possible future insecurity. It means that existence as a subject can no longer be structurally sustained.

Neither condition suffices alone. Meet only the first, and cultivation can still continue by other means — endurance, concession, trading space for time. Meet only the second, and there are still non-war routes untried.

Under this test, the following common reasons for starting a war all fail.

Preventive war (they may threaten us later, so strike first) — a future threat has not caused present loss of one's own cultivation.

Demonstrative war (fight one to warn the others) — this uses the other side's soldiers as carriers of a signal, which structurally instrumentalizes them already.

War for gain (resources, territory, trade rights) — acquiring gains is not defending cultivation. Applying the vocabulary of protecting cultivation to a contest over spoils is the most typical slide of all.

War for honor (an insult, an affront) — damaged honor does not touch cultivation's continued existence.

Pre-emption — this requires strict separation. If the other side's attack is genuinely underway, that is not pre-emption, it is resistance. If it is merely about to happen, we are back at preventive war.

3. Seeing Is Not Substituting

Theorem One contains a structure that is easy to misread, and it needs unpacking on its own.

"Defending subjectivity" has two faces. The first is defending one's own, and that is the trunk. The second is recognizing suppressed subjectivity inside the other side — sometimes the situation is not only that we are being pressed, but that people inside the other side have been pressed to the limit too.

Can the second face alone justify starting a war?

No. It can only be something seen alongside a first face that already holds.

And four extremely strict gates apply.

Gate one: it cannot start anything by itself. A subject may not cross a border to make war because "another country oppresses some of its people." Doing so treats war as an instrument for solving someone else's problem on their behalf, and that substitution is itself another cancellation of the suppressed party's subjectivity. Time after time in history, wars of liberation and humanitarian interventions have begun their slide at exactly this gate.

Gate two: seeing is not substituting. Even where the first face holds, recognition of the suppressed inside the other side must remain a posture of seeing. I can acknowledge your predicament and that you should be released as a subject; I do not decide what you are to become, choose your leaders, or plan your future.

Gate three: the rising must be real. The suppressed party's rising must be a genuine self-declaration from inside, not an appearance manufactured from outside, not a resistance that exists only because outside support keeps it alive, not something constructed to supply legitimacy for someone else's war. Outsiders may recognize and may support. They may not manufacture.

Gate four: an intervening party should have stakes of its own. This is the most counterintuitive: a completely "selfless" cross-border intervention is structurally suspect — a selfless intervention has no cannot-not of its own, and therefore no position at which it can be held accountable. A party with a genuinely legitimate position to intervene usually has stakes in the collision itself. Stakes are not a stain. They are a healthy anchor.

But "stakes" must be defined precisely. Not expansionist interest — territory, resources, markets — but the intervening party's own cultivation already being caught up in the situation. And a harder limit applies: that involvement must be structural erosion happening right now, not a probable future risk derived from a forecast.

This boundary matters enormously, because "future risk" is the seedbed of every preventive-war rhetoric: they will threaten us eventually; if we do not act, we will be affected too; once they are strong it will be too late. That rhetoric can package any expansion as defense.

The line falls here: a party genuinely protecting its own cultivation can point at what is already happening. A party pretending to protect its cultivation can only point at a prediction.

4. Theorem Two · Cannot Not Expand

A war has begun. That is only the first structural judgment. War is not static — it shifts, spreads, escalates. And no shift automatically inherits the legitimacy of what came before it.

Expansion is not one-dimensional. Expansion of scope (from a local theater to a wider geography), of intensity (from limited engagement to total war), of participants (bringing in allies, drawing in third parties), of objective, of duration, of means — an expansion along any one of these axes must pass its own separate test.

The test has this form: without expanding, the position defended under Theorem One will be lost.

An adequate expansion has to satisfy three things at once. The position originally defended is in fact being lost. The direction of expansion still points directly at dissolving the original cannot-not. And the search for non-war routes has not stopped just because fighting has started — before expanding, one must confirm those routes remain closed.

There is a particular trap to watch for here: the drift of a war's objective while it runs.

Three drifts are most common. From defending to destroying — what began as keeping subjectivity from being lost becomes destroying the other as a subject. From removing a threat to remaking the other — what began as stopping suppression becomes turning the other into the "correct" shape. From resolving the present conflict to settling it once and for all — what began as handling this cannot-not becomes eliminating every possible future threat.

None of these constitutes a legitimate expansion. They are the objective going out of bounds — even when they look more thorough, more efficient, more decisive in military terms.

The Napoleonic wars are the long case study in objective drift. The starting point was structurally defensible: a young republic facing coordinated intervention by several dynasties is a real defensive situation. But across the following fifteen years the objective slid steadily outward — from repelling intervention, to installing sister republics in neighboring states, to redrawing the map of Europe, to distributing relatives as kings, and finally to marching into a great power in 1812 to force compliance with a trade blockade.

Each step grew "naturally" out of the last. And not one step turned back to ask: is the cannot-not that justified starting still there?

Something else moved in step with it. Conscription levels, censorship, and the police apparatus inside France deepened with every expansion. External expansion and internal suppression were indeed two faces of one thing.

A war that began as defense spent fifteen years, without a single expansion test ever being applied, becoming the thing it had originally opposed.

5. Theorem Three · Cannot Not Be Directed Toward Ending

For a war legitimately started and legitimately expanded, the next structural question is whether it is running toward its own end.

"Ending" has a precise sense here. It does not mean the other side surrendering, being destroyed, or accepting our terms. Those states may accompany an ending, but they are not what the ending is.

The ending is the moment the original cannot-not is dissolved — the moment cultivation regains the conditions to continue.

This definition anchors the end of a war outside the war, rather than in any internal state of winning or losing. That anchoring is the whole theorem.

From it follows a constraint that hangs directly beneath Theorem Three:

Any means that forecloses the possibility of that ending in advance, or that structurally makes the other side's post-war cultivation impossible, is inadequate in principle.

Not because such means are cruel or inhumane — those are ethical judgments, which this framework does not make. Because they structurally destroy the direction of cultivation's recovery. At the moment they are used, Theorem Three has already been violated.

Which means? Large-scale indiscriminate destruction, where it removes at a stroke the basis on which the other side could recover as a carrier of cultivation. Precision strikes aimed at the capacity for reconstitution — eliminating a cultural core, an educational core, a spiritual core in one blow. Even where casualty figures look modest, if the effect is that the other side cannot reorganize as a carrier of cultivation, the constraint is violated. And irreversible structural alteration — rendering a place uninhabitable long-term, destroying what cannot be rebuilt.

Carthage in 146 BCE is the standard case on the wrong side of this line. When the war ended, the city was systematically demolished, the surviving population sold into slavery, and the land reorganized as a province. What was done went past removing an opponent's military capacity. It removed the minimum conditions under which a civilization could stand up again at all.

The result: that war did not end at the moment cultivation could continue again. It ended at the moment the other side ceased to exist as a carrier of cultivation. Those are two entirely different ending forms.

But the constraint needs a precise boundary, or it will be read too broadly.

It does not forbid every action that weakens the other side. Striking the other side's present capacity to suppress, its present war-making capacity, its present military organization — these may well be adequate, if they serve to dissolve the cannot-not.

The boundary falls here: present capacity to suppress may be struck; the minimum conditions for post-war reconstitution may not.

This boundary also imposes a retroactive requirement on the decision to start. At the moment a subject decides whether to begin, it must already be able to answer: if this war is fought, how does it end? How does cultivation continue on both sides afterward? What condition is the other side in when it is over?

An inability to answer means the sense of direction has not been established. And a war without a sense of direction will almost certainly lose direction while running — and what it breaks in the end is not only the other side's cultivation but its own.

6. Theorem Four · Cannot Not Be Open to Questioning

The first three theorems constrain what a war does. The fourth constrains what the warring party is.

It is also the deepest of the four, because it protects the warring party's own cultivation. The first three mainly prevent a war from breaking the other side's cultivation. The fourth prevents the warring party from breaking its own — through suppressed dissent, eliminated questioning, deified decision-makers.

A warring party that permits no questioning has already broken its own internal cultivation. Even winning militarily, it has lost the internal conditions for cultivation to continue afterward.

Theorem Four also holds a special position: it is simultaneously a norm and a diagnostic.

Judgments under the first three theorems can be self-declared. A party can say it has exhausted every route, that its expansion was necessary, that it is oriented toward ending. Those declarations are very hard to verify from outside, because outsiders cannot see the internal process of judgment.

Theorem Four is different. Whether questioning is permitted is directly visible from outside.

But seeing it accurately takes four layers of diagnosis.

Layer one: does questioning exist? Is anyone, any institution, any channel actually questioning? If everyone who dissents is removed and every channel is closed, Theorem Four fails immediately.

Layer two: can questioning reach the decision core? Questioning that circulates only at the periphery does not count. It has to reach the highest decision-makers and the core mechanism of decision.

Layer three: are questioners protected? Are they suppressed, persecuted, or sidelined for questioning? If the act itself carries a cost, then even where questioning still occurs, it is not sustainable.

Layer four: does questioning actually affect decisions? Even with the first three satisfied, if the results of questioning carry no weight in the decision, the questioning is theater.

All four layers must hold. One breach is failure.

The Athenian debate of 415 BCE over the Sicilian expedition is the best teaching case for these four layers — because the first three passed and the fourth failed.

Questioning existed: the assembly debated in the open, and both those urging the expedition and those opposing it spoke. It reached the decision core: the assembly was the decision, and nothing is more central than that. Questioners were protected: Nicias, who opposed the expedition publicly, was not punished for it — he was appointed one of its commanders.

Then the fourth layer broke. Hoping to talk the assembly out of the idea, Nicias described the forces and supplies such an expedition would require in the most extravagant terms, expecting the sheer scale to frighten everyone off. The assembly's response was to approve everything he had named and send a larger expedition.

The questioning was heard, recorded, even "adopted." It did not change the direction of the decision. It only changed its magnitude.

That is exactly why the case is valuable: having a debate is not the same as passing Theorem Four. The four layers have to be checked one by one.

One more point. Faking openness to questioning can succeed over short timescales and must fail over long ones.

In the short run, a warring party can manufacture the appearance of passing Theorem Four through information control, performative mechanisms, and quiet persecution of questioners. But genuine openness to questioning is an expression of a society's subjective health, and subjective health shows up directly, over long stretches, in the rate of development. A society that permits no real questioning has its internal subjects systematically suppressed; they contribute no genuine creativity or judgment; and its development rate necessarily falls behind — with the gap compounding across decades and generations.

A short-run disguise can pass. Over long timescales, the bill is settled at the level of development.

7. The Four Are One Thing

Though all four share a grammar, "cannot not" constrains a different object in each, forming a sequence of increasing penetration.

Theorem One constrains the initiation of an action. Theorem Two constrains its continuation and escalation. Theorem Three constrains its direction. Theorem Four constrains the warring party's own position — it does not restrain any particular act; it restrains the form in which the party exists as a subject.

Initiation, continuation, direction, position. Theorem Four penetrates deepest: before it, the warring party loses the identity of permitter. He is not the one who permits questioning. He is the one who undergoes it.

So the four are not a flat checklist of norms. They are concentric constraints — each one protecting cultivation from being swallowed by war at a deeper level than the last.

One further observation: the four reinforce each other. A subject that starts adequately under Theorem One usually also tests expansions strictly, orients toward ending, and permits questioning — because all of these come from the same source. The reverse holds too: a subject violating one usually violates the others. Such violations are not independent events. They are multiple symptoms of a single structural tendency.

Which means that assessing a war rarely requires a complete judgment on all four. And Theorem Four, as the externally observable diagnostic, is especially useful — look at it, and you can usually infer the approximate state of the other three.

The four theorems are not norms for war. They are cultivation's protective mechanism inside a war.

Next

Wars end. And there are only three ending forms: victory, defeat, stalemate.

Each carries its own temptation. And the one that most needs to be refused outright is, in a good many civilizations, treated as the most dignified choice available.