Introduction

Two people who live together have been stuck on one thing for a long time.

One of them holds that household matters should be talked through before they are done, and that acting without talking first amounts to writing the other person out. The other holds that what one can finish alone need not be put to a meeting, and that having to clear everything in advance amounts to distrusting the other's competence to get things done.

Neither of these is a preference. For clarity the case is set here in pure type: both ends are already laws of their own. Actual entry into handling is not like that — there I can confirm only my own end, and sections two and three say why. Over a long stretch the first person learned that whoever is routed around ends up treated as not there. Over a stretch just as long the second learned that whoever must always ask permission ends up losing the standing to act at all. Neither is in a temper. Both have tried yielding, and after the yielding the thing came back unchanged.

They have tried reasoning, and found at the end of it that the other had understood every word and still did not agree. They have tried splitting the difference, and the compromise fell apart on its second use, because it violated both laws at once. They have tried not raising it, and after that the matter became the background hum under everything else in the house.

At this point a person asks: what else is there.

And before that question can be asked well, something more important has not yet been seen. These two are not two opinions that can be weighed against each other. They are two laws. Once they are two laws, weighing does not fail because their weights are too close; it fails because ranking them would require a measure standing outside both bodies of law at once and converting two first-person positions into a single quantity. Neither of those can be had.

The ten sections that follow put the object, the boundaries, the layers and the routes in place. Types of handling are established among them, but no procedure is supplied that could simply be copied out. Once the position is right, method has somewhere to stand.

---

Upper Volume · Locating

This volume first places the object, the boundaries, the layers and the routes. It establishes types of handling, but supplies no procedure that can simply be copied out.

---

1 · Which collisions fall here

A collision enters here through two gates.

The first is set out in section two and asks after its structure: is this two already-formed laws requiring incompatible action in the same situation.

The second is set here and asks what its handling touches: would the way this collision is handled put at risk my capacity to go on making first-person revisions and to go on being open to questioning.

A collision that touches that position under no handling at all may be trouble, may be loss, may be a thing that hurts badly, but it is not conflict in the sense used here. Some accounts can be settled. Some things dissolve once said out loud. In some, one side gives a step and nothing is taken from the one who gives it. All of these are fine, and none of them needs what comes later.

Both gates must be passed, and from this a counterintuitive consequence follows. The size of a collision does not fix the scope. A very small matter, handled by suppression, may touch that position; what those two people are stuck on might be who pays the electricity. A matter that looks very large, where both sides have room, may never touch it once.

That the gate is anchored on my side has a reason.

Leaving room for the other runs through the whole treatise. It bears on every step of method and on when to stop. It does not bear on the door that screens cases. Once the other's side is written into the entry condition, the whole thing contracts to a single sentence: I first judge whether you are worth it, then decide whether to take you on. And that is precisely the structure this treatise exists to dismantle.

If the other's side is to enter this gate at all, it can only enter in this form: might the handling of this case put me in the position of constituting on the other's behalf. That question is still about me.

Several labels look as though they could fix the scope, and cannot.

Difference of opinion, clash of interests, misunderstanding, showdown. Not one of these is enough on its own to rule a collision in, and not one is enough on its own to rule it out. Whether interests are commensurable is not presupposed here; some convert and some do not, and this has no bearing on whether the case is the object under discussion. What dissolves once the misunderstanding is cleared never passed the first gate to begin with — while a conflict can stand entirely intact after both sides have understood every word.

Showdown needs a sentence of its own. It is one of the actions by which conflict is handled, not one of the kinds of object that conflict is. A conflict may reach showdown and may run its whole course without one. Listing showdown as a kind of conflict mistakes an action for an object.

Accordingly, what admits of diagnosis is this operation on this occasion, not this person. One paper has established: there is no finally-judged evil person, but there are diagnosable evil operations. The direct consequence here is that no test is given anywhere in this treatise for determining which layer the other is running on.

One last cell is drawn off. A person may have two things of their own that cannot be set down, and they may require incompatible action in a given situation; that is another matter and is not here. Its clearest kind — conflict between two already-claimed value contents under one and the same bearer — is treated in Value Theory Paper 5, On the Conflict of Value. Where two requirements cannot both be reduced to already-claimed value contents, the residue is expressly held over here and passed to Action Theory.

One further boundary should be marked. Value Theory Paper 6, On Value and the Other, sets three limits on the value side — unreadable to exhaustion, not comparable, not merged — and expressly declines to supply a procedure. What is handled here is the law side: how two already-formed laws go when they require incompatible action in the same situation. Those three limits are not re-established here and are not borrowed as premises of derivation. Each side stands on its own; neither substitutes for the other.

---

2 · The object gate and the operation gate

The content of the first gate is this:

Two independently formed laws, each its bearer's own and each already formed, require actions that cannot be jointly realized in the same situation, and three things hold at once: they cannot certify each other's arrival, neither may legislate for the other, and no single party's withdrawal can complete the settlement.

Left standing on its own, that sentence locks.

Where it locks: if I must first confirm that the other end really is a law of its own before I have standing to begin, then I never have standing, because certifying the other's arrival is forbidden in the next section. And if I may begin without confirming, then the two laws in the sentence are not an operative gate at all but only an ontological description.

The way out is to separate the layers.

Object layer. What is under discussion here is, ontologically, two independently formed laws, each its bearer's own, requiring actions that cannot be jointly realized in the same situation.

Operation layer. I can confirm only my own end. The other end enters as a requirement that the other bears in the first person and by which the other goes on organizing their action. I do not certify whether it has attained the standing of a law of its own, and I do not cancel it on the other's behalf for want of such certification.

With the layers apart, both sentences hold and neither trespasses on the other. Handling begins on the principle of non-substitution. If it is later confirmed that the object type never obtained, one exits — one does not turn and judge the other. Exiting says: this case does not belong here. Turning to judge says: the other end does not qualify. The first is about scope; the second is about a person, and the second is not within the powers exercised here.

Now the three narrowing conditions.

They cannot certify each other's arrival. Two first-person positions each running at 15DD in the relational domain in question may each have derived the same law, and this is not consensus, not sharing; it is separate arrival. I cannot use my arrival to certify the other's, nor the other's to support mine. This has been established. Its consequence here: between the two laws there is no position one could occupy to say that this one is truer than that. It is not that the position has yet to be found; it is that such a position's existence contradicts the first-person origin of each law.

There is a courteous-sounding move: don't say whose matters more, say both matter equally. It does not hold either. Equally-mattering is still two marks on one scale; it has not removed the scale, only placed both readings at one point.

A general boundary is needed here or the sentence will be read far too wide. What cannot be compared is the first-person origin standing of the two sides, not every fact about action. Safety risk, harm, rights boundaries, legal authority, procedural priority and resource conditions remain comparable and remain adjudicable. Such adjudications arrange action; they cannot write back which law is truer in the first person.

Neither may legislate for the other. Law can say you may not crush the other. Law cannot say you should yield. The second is legislating on the other's behalf, and that is colonization. This too has been established.

But it must be followed at once by a distinction, or it will seal off what should be kept. I may state reasons. I may make requests. I may set out boundaries. I may propose an action plan. What is actually forbidden is something else: my law cannot, on its own strength alone, acquire the authority to settle for you that you must yield. Taking my reason as already established on your behalf — that is legislating for you. Without this distinction, argument, request, consensus and external arbitration are all sealed off in advance, which is not what the rule forbids.

No single party's withdrawal can complete the settlement. This one must be stated precisely. It does not say one cannot physically leave. It says: no party can, by withdrawing unilaterally, close out this incompatibility as regards that party's first-person responsibility, the shared relation, third-party responsibility, or institutional consequence.

Written this way it holds intimate relations, shared child-rearing, organizational role and public responsibility, and it does not conflict with the Final Volume's allowance that both sides may each depart. Departure can happen; departure does not by itself close those accounts. Merely disliking departure, sunk cost, or the inertia of a relation is not enough on its own to open this gate.

---

3 · Whether this end is my law

The previous section wrote two laws into the object. This section examines only my end.

Because one thing is possible: the end I take to be mine may not be mine.

It may have been installed. It may be the shape left by some relation. It may be a requirement that recurs and was never claimed by me. One paper has treated this structure and given three cues.

First: with the external conditions stripped away, does it still hold. When nobody is watching, when nobody will praise or blame me for it, is it still there.

Second: can it coexist with my other important ends without eating them all. The genuine one tends to stand alongside other things; the installed one often demands that everything else be cleared.

Third: when it is briefly interrupted, what appears — regret and a pull to return, or a sense of the whole person going down. The former supports its being an important thing of mine; the latter supports my having been propped up by it, which is another matter.

The three can only support or weaken a judgment; they are not necessary and sufficient conditions. They do not guarantee getting it right.

And all three require a first-person internal reading. Whether the thing is still there once external conditions are stripped can only be read from inside the position. That a subject holds an irreplaceable first-person reading position does not mean the subject necessarily reads accurately, nor that the subject already knows. Taking these three to test whether the other's requirement is genuine is legislating for the other.

Three registers must be kept apart here, because the same shape recurs later. There are cues; there is no closure-guaranteeing final criterion; and there are traces visible from outside.

The first register is the three cues above. The second says: giving no criterion does not mean there are no cues; what is denied is only the closure-guaranteeing kind. The third exists for this question too: in the materials available, does the requirement appear only when a particular person is present, can be seen, or will be evaluated; does it vanish with the end of some relation; is there evidence of its being carried out when nobody was watching. These are inferential traces; they support or weaken a judgment and do not adjudicate genuineness. They read the conditional correlations in the available material and cannot see everything nobody knows. The full shape of the three registers is used once more in section twenty-six.

Adjudicating among cues, relational facts and one's own retrospection is the work of judgment. How judgment forms and how it is exercised is passed to Action Theory. A subject can also get it wrong; nothing here guarantees otherwise.

If the confirmation of this end does not hold, the manner of handling matters. One does not declare the requirement false. One does not start the four constraints. One registers a single thing: the object type of this case has not been obtained. Then one moves to the diagnosis of colonization and introjection, which is other work.

This is not some subset of the theorems falling through. Nothing here has started at all.

---

4 · Boundaries and two ledgers

Those two people, months into the impasse, one day heard one of them say something heavy. After that sentence there was more than one thing in the room.

Now there are two different questions.

The first: are the action requirements of those two laws still incompatible. The second: has what was just said already caused harm that must be acknowledged.

The first is heard here. The second is not; it has its own adjudicating venue, treated in Moral Law Paper 4, Procedure of the Moral Court.

What matters is this: these two are not an either-or but two ledgers kept in parallel. One case may perfectly well have both. A conflict may also have no harm in it at all — just two laws colliding, with nobody saying anything heavy.

Two ledgers in parallel closes off two roads.

An apology does not dissolve the incompatibility of two laws. The heavy sentence is acknowledged, repaired, the two make up, and the question of whether household matters should be discussed first has not moved a word. Taking the settling of harm for the settling of conflict is covering an unsettled matter with a settled one.

Nor may it be run backwards. One may not, on the ground that the large matter is unresolved, leave what was said hanging unacknowledged. The two settle separately.

There is a further stretch that neither adjudicating machine settles: harm newly arising while a conflict is under way, which neither side has yet self-judged. What has occurred and been self-judged has its venue; what has not been self-judged cannot be taken there either.

That stretch is expressly held over here, without pretending it has been placed.

Holding over is not nothing. It does not block fact-checking, does not block protective withdrawal, does not block the prior operation of the corresponding rights or safety mechanisms. Nor do those acts complete, on the party's behalf, the self-judgment the other venue requires. In particular: continuing along the unilateral path does not constitute handling of that stretch.

And the harm ledger keeps its own authority. Apology, acknowledgment of fact, repair, compensation, safety guarantees — none of these is established here as a universal duty in all cases. But the harm ledger, the rights and safety mechanisms, existing procedure, or a subject's own law may perfectly well list one of them as a necessary condition in this case. That ledger boundary is not crossed here.

One more cell is drawn off. Two laws meeting at 14DD is treated in Legal Series Paper 1. That is an attribution of where the object comes from; it does not mean the subject position drops to 14DD while handling proceeds here.

---

5 · Graded handover

There is a situation whose handling does not belong to this treatise.

Suppose that of those two people one begins doing this: taking control of the other's income, cutting the other's contact with friends one thread at a time, setting the topic and the time of every discussion, making it steadily harder for the other to leave at all. What is happening in that room is no longer merely two laws being incompatible.

Once asymmetric operations begin to compress my capacity to interrupt, to exit, to speak, or my basic conditions, this treatise no longer hears the case alone. At that point it runs in parallel with, or hands over to, the Power Series, rights mechanisms, law and safety mechanisms. Protection may come before any reciprocal handling.

One grade further on there is another threshold. The War Treatise states of itself that it handles the expansion or preparation for expansion of a 14DD party's systemic purpose, and that falling back to the extreme state is conditioned on the whole spectrum of handling outside war having been exhausted. Only where the object further satisfies these two, and where the conditions of cultivation on my side have already been substantively drawn in, does the case pass to the War Treatise.

The two grades may not be jumped. Control, resource asymmetry, agenda manipulation, manufactured dependency and the compression of exit in personal relations usually stop at the first grade and do not automatically reach the threshold of the second. Handing them straight to the treatise on the extreme state amounts to making a relation that should have protection first wait on a threshold it will never reach.

Only the domain boundaries and handover conditions of Power Series Paper 1, On the Origin of Power, and Rights Series Paper 5, On the Limitation and Violation of Basic Rights, are used here; their internal machinery is not unfolded.

---

6 · Conflict cannot be exempted

Two laws are each internally generated, each separately arrived at, and are not produced by one and the same procedure. That being so, there is no reason whatever to guarantee that their action requirements are compatible in every concrete situation.

And the residual remainder after mutual chiseling is not necessarily smaller than before. One paper has said this with the fractal coastline: the finer the tracing, the more edge may show. Two people talking more deeply do not necessarily leave fewer surfaces exposed to collision.

From this one can say what avoidance is. Avoidance is not a mode of settlement by which a conflict has been handled. The incompatibility has not been closed out by avoidance. It remains in the shared field. It may be deferred, may change form, may be temporarily deactivated.

Nothing here predicts that the consequences are necessarily worse. It says only this: avoidance did not settle it.

That it cannot be exempted does not mean it should not be avoided. Better no conflict where none is needed. Changing an arrangement so that the step need not occur is a good thing.

Here the accounts must be kept by tense, because section eight uses the same action again. Changing conditions while the object has not yet obtained is prevention or avoidance. Changing conditions after the object has obtained, so that the current incompatibility is dissolved or deactivated, is an action arrangement. The act may be identical; the theoretical identity is not. The first lies before conflict; the second is one of the kinds of conflict-handling.

Everything here begins after avoidance has failed on this particular matter.

---

7 · Two relational conditions, and the two paths that divide from them

The division of paths goes by the current relational condition, not by persons.

The reciprocal condition is these two things holding at once: in this relational domain, both parties can currently acknowledge that the requirement the other bears in the first person is not mine to legislate for the other and not mine to cancel for the other; and that my own judgment is open to the other's questioning.

What is confirmed here is a non-substitutable relational condition, not a mutual certification of arrival. This has the same logical form as the operation-layer statement of the previous section.

The path taken while the reciprocal condition is currently available is called the reciprocal path. The path taken while it is currently unavailable is called the unilateral path. Unilateral refers to the constraint running on my side alone; it does not mean that I control the other unilaterally.

Both definitions speak only of conditions, not of persons. Conditions are current, revisable, and shown by what has already happened in this relation.

The mechanism of selection must be stated precisely, because this is where one most easily slides back into judging persons.

Making no prior determination means only: not judging layer, not judging inner state. It does not mean not examining safety, not examining exit, not examining resource asymmetry and procedural manipulation. Once the protective gate of the previous section and the basic power-condition check have been passed, one starts on the reciprocal path.

After starting, joint deepening either happens or does not. If the other does not enter mutual chiseling, if each time they fall back to treating my law as something that can be crushed, then in this relational domain the reciprocal condition is in fact unavailable, and one turns to the unilateral path.

The ground of turning is the fact that the current reciprocal process is unavailable. Joint deepening may fail to happen because of unwillingness, because of a present incapacity, because the timing is wrong, because procedure is not trusted, and also because the other does indeed see my law as something crushable. The turn requires only their common conclusion; it does not require an added account of the other's inner state or layer.

Not-safe divides in two. If it is merely unsuitable to enter joint deepening at this moment, and none of the protective conditions on subject position, exit, safety and rights has been triggered, one may turn to the unilateral path. If not-safe has already triggered the first-grade handover of section five, or the stop gates set later, then one suspends, protects, or hands over first, and does not enter unilateral intervention.

There is a further reason to write it this way. Someone who has not yet come that far and someone who could and chooses not to may present the same behavioral signature in a single or local interaction. Long-run relational fact may shift the weight of my judgment; it cannot complete a certification. And the turn of path depends in any case only on whether the current reciprocal process is available.

A structural explanation is available for why the reciprocal condition often fails: a position still running at 14DD has no structural reason to acknowledge that another law carries the same first-person absoluteness as its own. That explains why such failure is structurally possible. It does not establish that in some current case the reciprocal condition is unavailable, and it carries none of the load of the division of paths.

---

8 · Action arrangement and law are not the same layer

Back to those two people.

One day they stopped arguing about whether things should be discussed first, and instead sorted household matters into three kinds: some finished first and mentioned afterwards, some mentioned before doing, some settled only by both together. After that the thing stopped happening daily.

What happened there must be stated exactly.

Neither law changed at all. The first still holds that whoever is routed around ends up treated as not there; the second still holds that whoever must always ask permission loses the standing to act.

What changed is something else. Time, resources, roles, procedure and the conditions of action were rearranged, so that the current incompatibility of action was dissolved or deactivated. And the arrangement itself neither generates a third law of one's own nor, on its own strength, rewrites either party's law.

Without this layer, what happened that day is misread as one of four things: some law was revised; a third shared law appeared; one party legislated for the other; or the problem is simply being stalled. All four are wrong, and they arise because there is no position that can hold "neither law changed and the problem is solved."

Within this layer there is one special kind: between the two parties' respective non-negotiables, a solution can be found that equals neither party's initial position and requires neither to give up its non-negotiable. Economics Series Paper 1, Four Layers of Economic Rationality, calls this C-production and has established its conditions: both parties have non-negotiables; the solution is not identical with either initial position; the search for it depends on genuine inquiry into both parties' bottom lines; and where no such solution exists, the failure to form one can be registered as a valid result of the process rather than automatically judged a failure. It is a production, not a compromise, because it is a new solution neither party had before entering the process. The three-way sort is such a solution.

But C-production is one kind within this layer, not the name of the layer. One-sided adjustment on my own side, protective arrangement, adjudication by external procedure — all belong to the layer of action arrangement, and none is that joint production.

On the unilateral path this layer is still present, only without joint inquiry. What one side can do is change time, resources and conditions of action on its own side, and propose plans. It cannot unilaterally declare a C-production to be in force.

How arrangements are formed, chosen, executed and reviewed is passed to Action Theory. Only the division of types is done here.

---

9 · Position as means, and the objective function

Handling conflict is a means. The direction is to keep the openness of both subject positions at once: outward, not legislating for the other; inward, not letting the conflict rewrite me. The first is called cultivation, the second self-cultivation.

Both sides are written out together in order to block a particular reading: that in order to cultivate the other, I must therefore work on them.

The objective function is accordingly not winning. The true losing is this:

In conflict my law may be rewritten by me; it may not be rewritten for me by the conflict.

Two glosses are needed or the sentence reads as one of two extremes.

The first is that it does not forbid change. A conflict may of course give me reasons, materials, pressure and new facts, and thereby trigger my revision. What is forbidden is not that conflict has causal effect on me; what is forbidden is its occupying the final position of first-person confirmation. Conflict may trigger, promote and alter my available materials; it cannot complete for me the constitutive confirmation that this rewriting is indeed mine.

The reciprocal path indeed requires that I re-examine my own construction before the other's frame of reference, and this may lead me to revise my law. That is the path running normally, not a breach.

The second is that it gives no criterion. Whether this is a revision I endorse or a colonization of me is judged by the subject, through judgment. No closure-guaranteeing criterion is supplied here, and the subject may get it wrong.

The position in which I go on making first-person revisions and go on being open to questioning is written as self-cultivation. Its opposite is self-colonization. Self-cultivation here does not mean a programme of self-improvement or moral perfection; it names the preservation of my first-person position of revision and of openness to questioning. It shares a name with outward cultivation and runs in the opposite direction; wherever it appears it is marked as such.

One further thing is not among my options.

A person will sometimes think: suppose I simply act as though I do not see that the other also has a position of their own, and fight this as a negotiation of interests — would that not be far simpler.

That step is not among the options. Not because it is immoral, but because I have already confirmed in this case that seeing the other's subject position belongs to my own law. I may therefore suspend, withdraw, hand over — and I cannot write acting-as-though-I-do-not-see into the same position as one of its handling strategies. Two further things serve as side evidence: once a remainder is produced it cannot be cleared to zero; and downshifting loses depth of discrimination about others' intentions. Both are side evidence and carry none of the constitutive modality of the impossibility of downshifting.

---

10 · Three points of difference from the War Treatise

By this point the difference from the SAE War Treatise can be stated. That treatise handles the extreme state under the oppression of systemic purpose; this one handles two laws meeting at the personal scale. Three differences bear on how the later volumes are written.

First, first-person responsibility cannot be borne by proxy. In personal conflict the one who decides, the one who executes and the one who bears may not be one and the same life, and harm may fall on a third person. What can be had here is only this sentence: I cannot have my own action's first-person responsibility borne by someone else.

Second, there is no uniform ritualizable wartime boundary. Unlike that treatise, there is no clean switch between wartime and peacetime. But while a conflict is not activated, procedures can still be built, exit modes agreed, third-party mechanisms set.

Third, there is no ritual of termination. Conflicts often merely stop being raised. Stalling without closure very easily becomes the default path. When the third constraint is set later, this will weigh far more here than it does in the War Treatise.

Two further items must be placed. There is a state in which the matter never forms a situation requiring handling at all; that is non-activation, belongs to the stretch before conflict, and is not one of the positions on the unilateral path later. And drawing on the asymmetry within a relation belongs here to the question of means and power raised under the second constraint; its domain handover is the first grade of section five.

---

Middle Volume · Four Constraints

11 · Which kind of cannot-not these four are

All four below open with cannot-not. The form of that opening must be settled first, or the four sections after it will all be read wrong.

These four are not four norms. They are not four rules one ought to observe, nor four virtues. They issue no normative force.

Nor are they causally automatic. It is not that once a conflict appears, these four acts obtain by themselves.

They are conditional-constitutive:

So long as I go on handling, from the first-person position, a conflict that has passed the gates, and go on holding that my own law may not be written for me by the conflict, then facing, deepening, terminus-directedness and openness to questioning are conditions this handling position cannot itself cancel.

Accordingly the shared grammar is not "I have the right to face" or "I have the right to question." It is not that I have acquired four sovereign powers, but that my own law makes four exemptions cease to be modes of handling I can acknowledge.

One consequence of writing it this way must be stated at once, because it is commonly read backwards.

I may stop this operation. I may suspend, withdraw, hand over to procedure. Stopping does not prove the four wrong; it means only that this handling is no longer running. What is forbidden is one thing alone: I cannot, while still claiming to occupy the same position, cancel any one of them.

What obtains after stopping must also be written precisely. Once handling stops, the first three no longer run in the present tense, and the fourth generates no requirement of continued contact or continued response to the original other party. But the decision to stop, remove, arbitrate or suspend is itself still not to be placed in principle beyond examination of reasons. What continues there is an effect further upstream, not the original conflict-handling still running.

The four start only when several things already hold: the collision has passed both gates, the confirmation of my own end holds, and the protective gate of section five has been passed. Ordinary disagreement, exitable relations, disputes settleable straight off by existing rules — none of these is here.

All four are my first-person constraints. They give me no power to judge the other, and they cannot make the other do likewise. This is where they part company with every technique of conflict handling.

Facing and deepening do not require maintaining contact and do not cancel safe withdrawal. The harm ledger settles separately in parallel; none of the four settles harm.

The four stand together and their penetration deepens in order: entry, continuation, direction, one's own position. They are not four items on a list but four concentric circles.

---

12 · Cannot-not face it

Once a conflict has occurred, I cannot not face it. Facing, at the operation layer, is acknowledging: the law of my own that I have confirmed at my end, and the requirement the other bears in the first person and by which the other goes on organizing their action, cannot be jointly realized in the current situation. I do not reduce this incompatibility wholly to misunderstanding, timing, mood or one-sided fault, and I do not thereby certify that the other has attained the standing of a law of their own.

Why this is a condition rather than a piece of advice.

If I use misunderstanding, timing, or one party's fault to wholly reduce or replace a two-subject structure that has already passed the gates, I am not solving it; I am rewriting the object. After the rewriting, what I handle is no longer the thing that was there. Facing is the minimum condition for preserving the identity of the object.

The scope of this must be set precisely, because too narrow a reading seals off what should not be sealed.

What is forbidden is wholesale reduction, not discussion. Misunderstanding, mood, timing and fault may perfectly well coexist with a genuine incompatibility. Those two did have occasions where something was not said clearly, did hit one very bad moment, did have someone say something heavy. Discussing these does not cancel the two-subject structure. The harm ledger in particular may well have to judge fault. What is forbidden is only using any one of them to replace an incompatibility that already obtains.

Rewriting has five shapes.

Into misunderstanding: we just did not say it clearly. Into timing: later. Into mood: they are having a bad day. Into the other's fault. Into my fault.

The last two need separate treatment, because they look opposite and act the same.

Both convert a two-subject structural problem into a one-subject problem of fault. Blame hands it to the other; self-blame gathers it to myself. Self-blame looks like taking responsibility, and may even look more responsible than blame, yet it cancels the incompatibility just as thoroughly: if the matter is my failing, then the structure in which two laws require incompatible action need no longer be faced.

Avoidance is drawn to exactly this. Fault has a solution; incompatibility has no solution of that kind. A person would rather admit being in the wrong than admit that there stands a thing nobody did wrong which still does not come apart.

Three things do not follow from facing.

Facing is not speaking. Facing is acknowledgment on my side. Whether to speak, when, and how are matters of action, not given here.

Facing is not handling at once.

Facing does not require presence. This point must be stated explicitly. Covering a conflict with another name and then leaving is avoidance; withdrawing from danger first, stopping contact, or protecting a boundary does not thereby constitute avoidance. Facing is first-person acknowledgment; it can happen after withdrawal and can happen a long way off.

From this, leaving can be spoken of properly. Leaving is an external act and takes no technical identity by itself. Sorted by the state the handling is in at the moment of leaving, there are at least four:

Leaving under another name without having faced it is avoidance and breaches this constraint. Leaving first for safety, boundary or fact-checking is protective withdrawal and does not. Terminating the running of the handling within the original relation without having obtained a bilateral terminus is removal; where it is only a temporary halt not yet handed over, it is suspension. Each departing after both have obtained a terminus is action after the terminus.

The division looks not only at whether facing occurred but at which state the handling is in at the moment of leaving. And where shared resources, third-party responsibility or procedural obligation remain, leaving does not close them out.

---

13 · Cannot-not deepen the understanding

Having faced it, I cannot not deepen my understanding of this conflict.

The reason is direct: without clarifying which set of action requirements the incompatibility actually lies in, no adjustment can know whether what it changes is the conflict or something else. If those two never work out whether the dispute is about discussing-first or about being-written-out, they may change what they do a hundred times and only shift posture in place.

The direction of understanding must be stated precisely, or it slips into one of two deviations.

One is spending the effort outward. There are six shapes, and their natures are not the same, so they must be told apart.

Expanding this one matter into "you are always like this," and drifting from this matter to "whether this relation is worth it," rather clearly change the object. What was to be clarified was a specific set of action requirements; what is now to be clarified has become a person's character or a relation's whole account.

Bringing in third parties and dredging up old accounts require room. Bringing in a third party is sometimes obtaining evidence, or identifying a recurrent structure; dredging up old accounts is sometimes judging whether this instance belongs to the same pattern. They can be drift and can fail to be.

Escalating tone and means, and drawing on the asymmetry within a relation, are chiefly not questions of object but of means and power. Their domain handover is the first grade of section five.

The other deviation is making understanding a precondition and daring to do nothing. This too must be blocked. Protective withdrawal, temporary suspension, resource separation, holding the status quo — these reversible arrangements may go first. They do not pass themselves off as final handling. Where a necessary protective measure is irreversible, its authority is given by the rights, legal or safety ledger, not authorized by this constraint, and it does not pass itself off as this conflict's terminus. If handling continues or restarts, clarification of the object is still owed; if the matter has been responsibly removed, no further requirement of continued clarification is issued here.

The object of understanding is the shape of the incompatibility, not the other as a person.

And deepening has a stopping boundary. Deep enough for the incompatibility to be seen clearly, not deep enough to understand everything. Remainder is not cleared to zero by understanding, and talking more deeply does not necessarily leave fewer surfaces. If understanding-everything is made the goal, this constraint becomes impossible to satisfy and the directedness required by the next one falls through with it.

Deepening also does not require continued contact. This constraint is mine in the first person; it generates no duty on the other to keep talking, keep listening, keep explaining, and no requirement from me that they do.

A last sentence points back to section three: the examination of my end can still only be done by me. Both parties entering the same reciprocal process does not make the examination at my end a joint examination, and the three cues still may not be taken to test the other.

---

14 · Cannot-not be directed toward a terminus

So long as conflict-handling is still running, I cannot elevate the maintenance of the conflict into an end.

Directedness-toward-terminus is the name of this anti-self-maintenance condition. It does not mean every path can obtain a terminus in the technical sense. On the reciprocal path it takes effect as directedness toward consensus or toward acknowledged incommensurability. On the unilateral path, while the original object still obtains and handling is still running, the anti-self-maintenance requirement takes effect as: making the current incompatibility genuinely deactivate through a change of action arrangement or of activating conditions, or removing the handling responsibly. If a genuine first-person revision later occurs at my end and dissolves the original object, that is not a third result this constraint requires me to aim at; it is this constraint losing its object from that point on, and it is registered separately as a scope exit. Stopping point, removal and object dissolution are not renamed terminus on that account.

Crushing the other is not a terminus but the substitution of suppression for conflict; stalling without closure is not a terminus but the substitution of a standing condition for conflict.

The reason was given in section nine: handling conflict is a means, and therefore the maintenance of conflict cannot be elevated into an end. That is the shared core, and what follows on the two paths is its unfolding, not an exception to it.

There are only two termini, and they are obtained only on the reciprocal path. Consensus, or acknowledged incommensurability. Both termini require that both parties enter the same reciprocal process; neither requires physical co-presence or continued contact, and neither is something I can obtain alone.

That the unilateral path has no bilateral terminus should be conceded outright rather than blurred. And precisely because it has no terminus, that path must have stopping conditions. What they are is left to the Final Volume.

The unilateral path has one further thing which this constraint does not require and which will make this constraint cease to apply: a genuine first-person revision at my end that dissolves the original object.

Several points must be stated together here, and stated explicitly. This constraint does not forbid genuine revision. This constraint also does not require me to revise myself in order to end a conflict. Once a revision genuinely occurs and the object disappears, the four cease to run in the present tense of conflict-handling: the first three lose the original conflict object, and the fourth generates no requirement of continued contact or response; while that revision and the decision to stop remain under review of reasons further upstream. No third terminus has been obtained here. It is not a handling strategy, not a terminus, and not derived from directedness-toward-terminus. Its identity is left to the Final Volume.

A core sentence belongs here, because it is what the passage above most easily overturns.

A relation may be terminated, a channel may be closed; one may not on that account pass final judgment on this person.

Actual relations and channels of contact may be terminated permanently. This constraint does not require keeping contact details, does not require remaining reachable, does not require permitting restart at any time. What it forbids is something else: writing the other's present unavailability for reciprocity into a final verdict on them as a subject.

Two outcomes are especially liable to be mistaken for a terminus and must be excluded here. A stopping point, removal and suspension are also not termini, but each has its own technical identity in the Final Volume.

Crushing. It will indeed stop the matter being raised, and it substitutes suppression for conflict.

Stalling. Its shapes are: not raising it, going around it, becoming the background hum under everything else in the house, covering it with busyness. Personal conflict has no instrument of surrender and no armistice; often there is simply a day on which it stops being said. Stalling without closure therefore very easily becomes the default path. This constraint weighs far more at the personal scale than it does in the War Treatise, precisely because a ritual marking the end is missing here.

Two more things directedness does not include.

Directedness is not haste. Acknowledging a consensus I do not in fact acknowledge, in order to have the thing end, uses this constraint to breach the previous one. The temptation of speed is another shape of avoidance.

Directedness is not my bringing the terminus about. A terminus takes both parties. What I can do is have every step on my side directed toward it. Requiring the other's agreement returns to legislating for the other. This is the constraint most easily broken of the four, because directedness and bringing-about are very hard to keep apart in feel.

Finally, removal must be given an exact position. Removal terminates the continued running of this handling within the original relation; it is not called directedness toward closure. This constraint is not breached by removal, and removal is not renamed terminus either.

---

15 · Cannot-not remain open to questioning

I cannot place my own action in a conflict, in principle, beyond examination of reasons.

The point to fix first is that it is I who must remain open to questioning. This constraint gives me no standing to question the other. Of the four, only this one can be read backwards in its grammar, and so it must stand first.

Why this is a condition. In two steps.

Step one: revisability requires that reasons can in principle reach me. A position placed in principle beyond all examination of reasons cannot coexist with "my law may be rewritten by me": an exemption-in-principle deprives revision of its point of entry. If no reason can reach me, my law no longer has an occasion to be revised by me; it is only held by me.

Step two: preventing me from monopolizing the selection, interpretation and adjudication of reasons requires at least one route of external review not wholly under my control. Step one yields only that reasons must be able to reach; it does not yield that the route must lie outside me. And a channel in which I decide entirely which reasons count is no channel at all.

This proof covers only exemption in principle. It does not cover any particular instance of not responding.

An upstream theorem in Moral Law Paper 1 has the same form: cannot-not let one's own non-colonization structure be exposed to questioning, so as to prevent self-colonization. Moral Law Paper 1 also supplies a two-layer precision there, taken over directly here.

The first layer is the screening of standing and validity. I cannot be my own final adjudicator and declare in advance that some class of questioner is permanently without standing. This layer I cannot exempt. Once I have that power I have become my own court of last instance.

The second layer is the routing of response and safety. When to respond, through whom, whether to suspend, how to protect privacy and safety — this layer I must decide and only I can.

Their relation is this: where an institution exists that is fit for purpose, available, and not controlled by one party to the dispute, the principal screening may be borne by the institution, because a communal procedure can distinguish genuine questioning, noise, manipulation and harm. In a structural vacuum, where no such institution is present, a subject may defer, refer on, set boundaries — but may not upgrade such arrangements into an exemption-in-principle that is permanent.

Of these, "fit for purpose, available, and not controlled by one party" is not in the upstream text. It is added here as a procedural-safety limitation for the conflict field, so that whatever bears the screening is not captured by power.

This constraint should not be confused with four claims, and each must be ruled out. It is not that anyone may question me at any time. It is not that I must answer immediately. It is not that private information must be surrendered. It is not that the other thereby acquires standing to adjudicate my inner state. What it requires is not placing myself beyond examination, not being on call.

Keeping the channel open is also not my having to go and ask them. An open channel says that I do not close it; it generates no duty on me to put questions to the other, still less any right to question them.

Closing the channel has five shapes: you would not understand; you have no standing to ask; wait until this has passed; I have already done more than enough; I am doing this for your own good. The fourth appears especially readily during a suspension on the unilateral path; the fifth is at the same time a marker of colonization. But these five are shapes of closing only when used to terminate all review; they are not automatic criteria in themselves. Someone who once said "wait until this has passed" may only have been unable to go on that day.

Questioning has four sources under two identities. The other party and third parties are an external avenue of questioning. The me of that time and the me of later are supplementary sources. Self-questioning cannot substitute for an external avenue of questioning, or this constraint slides from being questioned back into closed self-reflection, which is precisely what it exists to prevent.

What is needed is not necessarily the channel running from the original other party directly to me. Actual relations and direct contact may be closed permanently; the external avenue of questioning may be borne by a fit third party, by procedure, or by later public review. And only an institution fit for purpose, available and not controlled by a party to the dispute has standing to bear the principal screening.

A final qualification must be exact and not overstated. Of the four, this is the one most likely to leave a public trace, and it is an irreplaceable anti-closure condition. But it is not sufficient to certify whether the first three occurred. A person may perfectly well keep the channel open while not facing, not deepening, and directed at no terminus at all.

---

16 · Among the four

The four penetrate to different depths.

The first three each constrain one juncture: entry, continuation, direction. The fourth differs: what it constrains is not some act but my form of existence as a subject in this conflict. It is therefore the deepest.

One dichotomy is to be avoided: saying that the first three face outward and the fourth inward is inexact. All four are my first-person constraints; that is where they occur, and it is here in every case. The difference lies in their principal object: the first three face chiefly the object of the conflict and my action, the fourth chiefly my own position of exemption.

The four therefore also protect my side together. A person who places himself in principle beyond examination has already interrupted his own self-cultivation, however restrained he appeared in that conflict.

And that all four are first-person settles that no technique of mediation can be given here. None of the four can make the other do likewise. If someone wants to use them to require the other to face, to deepen, to be directed at a terminus, to remain open to questioning, that is no longer these four; it is the extrapolation of a first-person constraint into an ethics over persons, and that step has already been ruled a structurally illegitimate overreach further upstream.

What these four can do is one thing only: keep me, while handling this conflict, from rewriting the object and from letting the conflict rewrite my own law for me. They do not make conflicts fewer, and they do not guarantee that conflicts have solutions.

---

Lower Volume · Two Paths

17 · How the two paths divide

The four constraints are one set. They fall differently under two relational conditions.

The division goes by current relational condition, not by persons. That was established in section seven; three of its consequences are set out here, because the next two sections are where one most easily slides back into judging persons.

First, the choice of path is a provisional reading. It is revised with the state of the relation. It is not the product of a single determination, and it need not hold once for all.

Second, paths can be changed. Changing path does not require walking through facing again, because a path was in any case a reading taken in the course of things, not a classification made at entry. If the reciprocal condition is restored later, one turns back; if it lapses later, one turns across.

Third, the name of a path exempts nobody. Harm can occur on both paths, and being on either constitutes no exemption from it. The unilateral path in particular supplies no duty to endure oppression: the protective gate of section five holds on both paths, and neither can override it.

One thing should be said first. Economics Series Paper 2, The Other End of the Bridge, has unfolded in an economic register a quadrant in which I can see the other's position and the other cannot see mine. That paper handles asymmetric perception, cultivation, the acknowledgment of non-cultivability, and exit cost and institutional substitution. The unilateral path spoken of below is the same quadrant falling on the side of conflict-handling. Its conclusions are used directly here and not re-established.

One's own law may be revised by oneself. What the gates set later guard is who does the rewriting, not that the content must not change.

---

18 · The reciprocal path: joint deepening

On the reciprocal path both parties enter the same reciprocal process. This does not mean physical co-presence and does not mean continued contact.

Deepening on this path is joint. It is not each speaking past the other but re-examining one's own construction before the other's frame of reference. If those two genuinely reach this, what may happen is not that one of them is persuaded but that each sees where their own line had never been made clear.

A kind of yield can form here, and it must be written as a possibility. Mutual chiseling can make both parties' legislating more active; one paper has established this. What it may bring is not a division of goods but that the tracing itself activates each side's construction.

But the same thing may bring other things too: exhaustion, harm, and newly exposed incompatibilities. After a genuine joint deepening two people may be clearer than before about exactly where they do not fit.

Three qualifications follow.

First, the yield does not mean harm has disappeared. Harm still occurs on this path, and section four's parallel ledgers hold here as well. One may not use "ours is positive-sum" to deny that what was just said did in fact wound.

Second, a larger remainder neither automatically indicates failure nor automatically indicates greater precision. That two people see more difference after talking than before shows by itself neither that the talk went badly nor that it went well. What must be looked at is whether the newly added differences can be located, can be restated, can be questioned, and whether the object has held identical. If so, it may be registered as a possible gain in precision, and that registration remains correctable later; if not, it is only one more patch of blur.

Third, the examination at my end can still only be done by me. Both parties entering the same reciprocal process does not make the examination at my end a joint examination, and the three cues still may not be taken to test the other.

And the C-production of section eight has its conditions of production only on this path, because it requires genuine inquiry into both parties' bottom lines, and inquiry requires both to enter the same reciprocal process.

---

19 · The reciprocal path: directedness toward a terminus

There are two kinds of terminus on this path: consensus, or acknowledged incommensurability. Consensus has two histories of occurrence within it: arrangement consensus and revision consensus.

Two further possibilities are not termini. One is discovering after deepening that the object type never obtained — the after-the-fact occurrence of section two's exit, which cancels entry. The other is that the object did obtain and was later dissolved by one party's genuine revision without a shared reading forming. Both are scope exits and are handled in the Final Volume separately from the two histories of consensus.

Only this is said here: two laws unchanged and the problem solved — that kind is a terminus, not stalling.

Termini are not ranked against each other. Neither is written as better, nor are they written as equally good, because the latter is still two marks on one scale and has not removed the scale.

This path may also fail to reach a terminus. Both parties are within the reciprocal condition, the thing was worked through, and neither consensus nor acknowledgment was obtained. This must be kept apart from turning of path:

Failing to reach a terminus is the reciprocal condition still being available and no terminus obtained. Turning of path is the reciprocal condition being currently unavailable, whether it never started, started and broke off, or was available for a time and lapsed.

The former stays on this path; the latter moves to the next.

---

20 · Cultivation on the unilateral path

Change of scene.

A person has done a job in one place for many years and has a long-unresolved disagreement with someone above them. He holds that a thing should be done a certain way, for reasons of its own; the person above reads every disagreement as a problem of attitude. He has explained, and after hearing him out the other said "I know it's not easy for you." He has given way, and the giving way was recorded as "he has finally come round." He has tried saying it hard, and after that the matter turned into an assessment of him as a person.

What is happening here is not that he expresses himself badly. It is that in this relation whatever he says gets translated into a different register. The reciprocal condition is currently unavailable in this relational domain. And he cannot leave — or rather, leaving would not close the matter out.

The unilateral path is not a synonym for cultivation. It also includes arrangements on my own side, protection of boundaries, ceasing to apply, suspension and removal, and these do not all take changing the other's side as their object.

Only where I still make an active intervention on the other's side does that intervention, if it is to remain within conflict-handling, have cultivation as its direction: providing latent conditions for the other to discover and develop a law of their own. Cultivation constitutes no duty to intervene and does not rewrite the unilateral path into a project for another person's growth.

Four things must follow immediately, because this word is the one most easily read as its own opposite.

First, cultivation is not instruction, not remaking, not turning the other into something. The path does not carry the word cultivation in its name precisely because once it is called the path of cultivation it will be read as "I judge you lower, so I shall cultivate you." The name governs entry, the content governs practice, and the two are kept apart.

Second, the first step is not getting them to see me. One paper has said it: a more unfolded position can only grow out of a stable one. The first step is therefore helping them stand more firmly on the things they themselves cannot set down. Before that, efforts to have them consider my position carry a high risk of being translated as a strategic signal to make them yield. Every time that man explained, what the other heard was "he is contesting this."

Third, there is a structural condition ex ante. This one is checkable in the first person and requires no judgment of the other's layer or inner state:

An intervention may not compress the other's authority, channels or practical possibility of refusing, interrupting, revising, exiting and restating themselves.

An intervention that compresses the other's authority, channels or practical possibility of saying no does not belong to cultivative intervention as spoken of here, however much it looks like help. Offering reasons may change whether they are willing to refuse; it cannot lower whether they are able to refuse. This distinction matters: without it, the act of setting out one's grounds would itself be judged colonization.

Fourth, there is a set of ex post indications, and they are only supportive. That the other becomes better able to account for themselves, better able to refuse me — these can support a judgment that cultivation occurred. They are not necessary and sufficient criteria, still less my warrant for certifying their growth.

From this, dependency can be stated properly. The kind of dependency manufactured and locked in by the intervention, such that the other cannot say no to the intervener, is the gentle form of colonization. But not all dependency is colonization. People depend on each other in any case: a child on care, a patient on treatment, an apprentice on a master. Judged colonization is that kind alone: it was made, and it locked the channel of refusal.

Two last sentences: cultivation does not presuppose success, and it does not predict that the other will change. The reason for doing these things does not lie in a prediction.

---

21 · Four positions facing the other on the unilateral path

One paper has established five names as the spectrum of collision handling outside war, without unfolding them. The last of the five says that the matter never forms a situation requiring handling at all; that belongs to the stretch before conflict and was placed in section ten. The remaining four are unfolded here at the personal scale. That paper set the five as an ascending sequence; only the names are taken here, not the identity as a sequence, for the reason given below.

Fa-mou (伐谋, working through the strategic situation): change the situation so that the incompatible step need not occur. That man can turn what must be argued face to face into writing, turn a two-person matter into one with a third present, turn what is decided ad hoc each time into what is settled in advance.

Fa-bian (伐辩, working through argumentative exchange): set out the grounds of my line, without requiring that they accept the conclusion.

Fa-si (伐思, working through thought): ask only that the question stand for them.

Bu-fa (不伐, withholding intervention): no longer applying anything, only maintaining conditions.

These four are different positions, not a sequence running low to high. There was a reading on which the further along, the less force applied; no common dimension supports it. Changing the situation is often far more forceful than setting out reasons, because it deploys resources, space, time and the structures of third parties. Only their difference is claimed here, not that they decrease. The first three are operations; the fourth is a posture. The four are not an exhaustive classification of everything done on the unilateral path: arrangements on my own side, protection of boundaries, suspension and removal are none of them among the four.

The line between fa-bian and fa-si must be written on fa-bian, not only on fa-si. Bian sets out the grounds; si asks only that the question stand. Neither requires that they accept the conclusion. Written only on fa-si, fa-bian becomes homomorphic with technique of persuasion, and technique of persuasion is exactly what this whole path exists to avoid.

Fa-si carries a further limitation, because it degrades most easily. The question must remain refusable, revisable, interruptible. A question the other cannot refuse is not fa-si; it is fa-bian in disguise, or has already passed into pressure and the operation of power. Planting a question in someone's head so that they cannot shake it off is not making a question stand; it is something else.

Fa-mou carries one too. Changing the situation changes the conditions under which the incompatible step occurs; not the other's schedule, relations or choices. To be exact: one cannot, merely on the strength of "I am handling a conflict" or "I am cultivating you," acquire authority to arrange another's life unilaterally. Other authorities of role, of law, or of procedure settle separately within their own domains and have nothing to do with this path.

A last sentence concerns where force goes. Once the current object has deactivated, further force applied no longer belongs to the handling of this conflict. If it nonetheless occurs, it passes to the ledger of power, of harm, or of other action. This is not forbearance; it is that the additional force is no longer aimed at anything under handling.

---

22 · The limit of one-sided understanding on the unilateral path

There is no joint deepening on this path sufficient to form a shared reading. The other may still take part at certain points, and that does not constitute a shared reading.

Hence the strongest sealing sentence on this path:

That I have understood why they are like this does not mean they have agreed to my understanding, still less that the matter is thereby settled.

Taking a one-sided understanding for a shared consensus is legislating for the other. And this is the version most easily let through as consideration: how could being so thoughtful about them be colonization. If that man works out in his own head a full account of the other's behavior and then arranges everything afterwards according to that account, this is what he is doing.

A one-sided understanding also may not serve as my warrant for having done enough. "I have understood him, what more do you want" is, in shape, the fourth way of closing the channel of questioning. Understanding produces no exemption.

---

Final Volume · Terminus, Stopping Point, Removal

23 · Three technical destinations and two scope exits

Two layers first.

Only where the object still obtains do registrable technical destinations divide into terminus, stopping point and removal. Two scope exits fall outside the three.

Object correction: it is later confirmed that the entry conditions had never been met. What is cancelled is the prior classification.

Object dissolution: the object type had obtained, and thereafter a genuine first-person revision, or a relinquishment of the requirement at my end, made the original incompatibility of action cease to hold. Where no shared reading between the parties has formed, it is not a terminus; the original object simply exits scope from that point. If the two later jointly confirm a new compatibility, only then is it registered as revision consensus.

The operation layer carries one limitation. I may confirm a genuine revision at my end; where the change has occurred only at the other's end while the reciprocal process remains unavailable, I cannot certify on their behalf that the origin layer has changed, and can only judge afresh, from the visible relation of action, whether the object is still activated.

Back now to the three technical destinations. These three are not three outcomes; their identities differ.

Consensus and acknowledged incommensurability are termini, obtained by both parties. A local stopping point is not a terminus. Removal is not a terminus either.

Five words below each have their own place and may not be exchanged. Everyday words — ending, wrapping up, letting it go, no longer minding — are not enough to designate any one of them.

Identity	Content
Terminus	Consensus or acknowledged incommensurability, obtained by both parties on the reciprocal path
Stopping point	An articulable arrangement or condition has been established that genuinely deactivates the current incompatibility of action; it does not pass itself off as a terminus; it may reopen when activating conditions recur
Removal	The original relation no longer carries the handling, or the handling is carried by a fit procedure
Suspension	The unilateral path halts temporarily on a stop gate being triggered; the route may be judged afresh when conditions permit
Stalling	No technical closure has formed at all; silence, avoidance or backgrounding stands in for handling. It is not a legitimate fourth terminus

Suspension is not a form of removal; the two are listed separately. External arbitration is one form of removal.

Where the object still obtains and a registrable technical destination is obtained, it divides into terminus, stopping point and removal. Handling may also rest for a time in suspension, or never obtain any technical closure and stall; object correction and object dissolution are scope exits.

One sentence belongs at the head of this volume.

A conflict has an identifiable boundary of object and may reach a terminus. This person has no terminus.

What is closed here is this one incompatibility, not this relation and still less this person.

Several further sentences carry over and hold across the whole volume. The three technical destinations and object dissolution neither ratify the past nor settle the harm ledger; object correction only cancels the entry classification here and does not write back the actions, relations and harms that in fact occurred during a mistaken entry. Acknowledging the other's origin position issues no permission for action. Arbitration takes only disputes about action, not first-person constitution. And one terminus obtained is no guarantee against conflict later.

---

24 · The two histories of consensus

Before this point two things may occur, and neither is a technical destination.

It emerges in the talking that there was no genuine incompatibility of action there: one took the other to be demanding clearance on everything, the other took the first to be demanding a meeting for every act, and what each actually cared about had never collided at all.

This is the after-the-fact occurrence of section two's exit: the case does not belong here. What it cancels is entry, not closure. The case should never have been classified as falling within scope. Counted as a terminus, it would leave the impression that deepening can always dissolve incompatibility, which is not what deepening can do.

The other is object dissolution. The thing did obtain, and afterwards one party genuinely completed a revision by their own law, and the revised requirement no longer stands in an incompatible relation to the other's. If the two jointly confirm a new compatibility, that is the second kind below and is a terminus; where no shared reading has formed, it is only the original object exiting scope, not a terminus.

The two below are the histories of consensus, and both consensuses are termini.

The first is arrangement consensus. Neither law changed; the two jointly formed a solution under which the current action requirements can be jointly realized. Sorting household matters into three kinds is a case of this.

This is a terminus. And this kind must have a place here, or the layer newly set in section eight has no outlet: neither law changed and the problem is solved would have nowhere to sit and could only be misread as stalling.

The second is revision consensus. One party or both separately completed first-person revisions, and the new action requirements attained compatibility. It must be stated as one party or both, not that both must change. One re-examines their own construction before the other's frame of reference and changes; the other does not; and the two now fit. That is still a terminus.

Neither is completed by an act of persuasion. Persuasion may supply reasons and may causally promote one party's revision; the persuader's success does not itself constitute the consensus. What constitutes consensus is the first-person acknowledgment each party completes, or an action arrangement jointly formed and separately acknowledged.

Two qualifications follow.

Consensus does not automatically generate a third first-person legislator. The two may perfectly well have a standing rule from then on, sorting household matters into three kinds. But that rule's binding force must pass separately through each party's own law, or come from a procedure both have acknowledged. It is not a third law above the two. That each holds independently constitutes no mutual legislation.

Consensus does not ratify the past. It closes out this one incompatibility now; it does not rewrite what either party did before.

Revision consensus is obtained from the moment joint confirmation forms; it does not retroactively write an earlier one-sided object dissolution into a shared terminus that existed at the time.

---

25 · Acknowledged incommensurability

The content of the second terminus must be set out precisely, because it is the one most easily reduced to an empty phrase.

What is jointly confirmed in mutually acknowledged incommensurability — hereafter acknowledged incommensurability — is not that each has certified the other's arrival. That is the act forbidden in section two; if the terminus depended on it, the whole treatise would contradict itself.

What acknowledgment confirms is these two:

The two action requirements, each borne by its party in the first person, cannot be jointly realized at present; and neither party has authority to cancel the other's undertaking on their behalf.

This has the same logical form as the operation-layer statement. The object layer speaks of two laws; the operation layer says I confirm only my end; and what acknowledgment acknowledges is undertaking and non-cancellability, not standing. I do not need to certify that they have arrived in order to acknowledge that I cannot cancel on their behalf.

Four sentences follow.

Only the origin position is acknowledged; no permission for action is issued. Acknowledgment does not mean their action may therefore be allowed, does not mean I must bear its consequences, does not mean public rules may not limit it, and does not cancel the harm or rights ledgers. That man's acknowledging that his superior also has things they care about does not make those translations legitimate.

Acknowledging incommensurability is not acknowledging that the other is right, and not that I am wrong. This must be stated explicitly, or the whole second terminus is read as splitting the difference.

Incommensurability is not a failure of understanding. It is the reading taken after the shape of the incompatibility of action has been seen clearly within the boundary of object set by deepening. What separates it from "we never got it clear" is exactly whether the second constraint was carried through.

After acknowledgment each retains their own law; the relation may continue or the parties may go separate ways. They may walk away after this terminus, but leaving does not itself constitute the terminus. Where shared resources, third-party responsibility or procedural obligation remain, departure does not close those accounts and can only pass to arrangement or arbitration.

---

26 · The local stopping point

The unilateral path cannot obtain either of those two termini. The one technical closure it can obtain within the original relation is a stopping point. Beyond that it can also suspend or remove, but the latter two are neither a terminus obtained by that path nor a stopping point.

A local stopping point is that an articulable arrangement or condition has been established which genuinely deactivates the current incompatibility of action. That man may have turned everything requiring face-to-face argument into writing, so that the matter now fails to flare once a year.

It does not pass itself off as a terminus. The relation of action requirements that constitutes this conflict has not been rewritten at the origin layer; should the activating conditions recur, the two sides may again require actions that cannot be jointly realized, and the original object may re-enter handling.

A stopping point changes an action arrangement or an activating condition, not the origin standing of the two laws. And a stopping point may come from an arrangement on one side that is not a C-production — he alone changed how he does things, and the other took no part. Once joint inquiry is restored and a C-production forms, one should turn back to the reciprocal path and register it as arrangement consensus.

Now the hardest point in the treatise, which must be stated outright.

There is no closure-guaranteeing external criterion distinguishing a stopping point from stalling.

Stalling is that no arrangement addressing the incompatibility itself has been established, and not-raising-it stands in for handling. A stopping point is that an articulable arrangement or condition has been established. From the third person the two may look exactly alike: at this moment no incompatible action requirement is activated, neither is being raised, both are quiet.

On the surface evidence of silence or of not raising it alone, the third person cannot be guaranteed to tell them apart. Outside, supportive evidence can be had — whether those arrangements are public, whether they are stable, whether they can be repeatedly observed. But there is no closure-guaranteeing final criterion.

One thing is visible from outside, and its weight must be stated exactly.

Declaring a stopping point while closing every channel of review is a strong risk signal that the stopping point may only be stalling. It is enough to lower substantially the public credibility of the declaration; it is not enough to complete the internal classification for the subject.

Three registers must be kept apart here, and the same shape is used at three places in the treatise. There are cues — the three self-examinations. There is no closure-guaranteeing criterion — established in section three. And there are traces visible from outside — the risk signal just described. None may pass for another: cues are not criteria, and traces are not verdicts.

---

27 · Removal

The third technical destination is moving the conflict out of this relation.

Removal is the removal of the handling from its relational venue: the original relation no longer carries this handling, or a fit external procedure carries it. It concerns the handling, never the removal of either person. It is not called a terminus. External arbitration is one form of removal.

A suspension on the unilateral path is not a removal. Suspension is that path halting on a stop gate being triggered, not yet handed over, and not yet terminating the original relation as the field that carries the handling. The two may not be exchanged.

External arbitration needs a limitation drawn first, because this is where reverse exploitation is easiest.

Ceding the power of adjudication to something external is not a failure of subjecthood but an awareness of its boundary. But this holds only for a certain class of external body: a procedure both parties have acknowledged, a mechanism with lawful authority, or an institution with legitimate authority for the protection of basic rights.

Handing the power of adjudication to any external party whatever does not automatically embody boundary awareness. A third party selected unilaterally by the stronger side may be only an extension of that side's power. If that man is required to hand the disagreement to an arbitrator designated by his superior, that is not removal; that is the same thing continuing under another name.

What is ceded must also be specified precisely. What is ceded is the procedural power of adjudication over disputes about action, not the constitutive power over one's own law. Arbitration may decide who uses which resource when, which action may be carried out, how procedure is arranged. It cannot decide which law is truer in the first person, and it cannot decide whose subject standing is higher.

The position of institutions follows. Economics Series Paper 2 has established: institutions do not create the "we"; institutions lower the cost of exit. Institutions can change information, incentives, the structure of contact and the conditions of safety, and thereby may help acknowledgment to occur; but they cannot complete acknowledgment for a subject and cannot guarantee it. Law saying "you may not treat her as a means" is the externally enforced version of the ontological imperative: it marks off a forbidden zone, and violation has consequences. It is the substitute and the scaffolding for when the "we" is not present; it is not the "we" itself.

---

28 · Three stop gates, and restart

Can the unilateral path be followed indefinitely? No. There are three gates, and the triggering of any one suspends that path.

Gate one: my subject position or my conditions of exit are being actively compressed. This is a safety gate. Its triggering also triggers the graded handover of section five at the same time, because that is no longer only a matter of conflict-handling.

Gate two: my first-person position of rewriting is being taken by pressure, dependency or fear. This too is a safety gate. Note that what it guards is not whether my law has changed but who completes this rewriting.

Gate three: non-cultivability has been judged in the current relational domain. One paper has written this as a relational diagnosis rather than a moral judgment and not a judgment of capacity: under current conditions, the expected marginal benefit of further investment falls below the expected marginal cost. This is a subject's local judgment about the current relation, not a common cross-subject measure, and it does not require quantification.

The three must stand together, because each set alone is read askew. Gate one alone reads as: so long as the other has not compressed me openly, I must keep going. Gate two alone reads as: I may stop whenever I am uncomfortable. Gate three alone reads as: only do it if the numbers work.

Gate one carries a further sentence. Lacking the reciprocal condition does not itself constitute evil. That someone does not run that far in this relational domain is not evil. Actively cancelling another subject's position, or compressing their conditions of exit — that is what enters the diagnosis of evil operations.

What follows a suspension.

A suspension is only a suspension. People change; conditions change. But this sentence guarantees no later opportunity; it says only that suspension is not written as a terminus. What is suspended is this path, not a final judgment on this person.

The conditions of restart must keep the two ledgers apart, or a boundary is crossed.

Apology and forgiveness are not established here as universal conditions of restart in all cases. That belongs to another adjudication and is not set by conflict-handling.

Nor are they ruled out. The harm ledger, rights and safety mechanisms, existing procedure, or a subject's own law may well list some acknowledgment and repair as a necessary condition in this case.

Conflict-handling judges one thing afresh, namely the route. If the reciprocal condition is restored, one turns back to the reciprocal path. If the reciprocal condition remains unavailable and none of the three gates is triggered any longer, the unilateral path may restart. And where necessary conditions listed on other ledgers have not been met, no restart is declared here across those ledgers.

A last matter, to be sealed immediately after it is said.

How Is the Inter-Ontological Possible §3.3 has established: the turn from single to plural usually requires a rupture, and one key function of cultivation is to reserve enough blank space for a subject before the rupture has occurred. It did not say that withdrawal creates the blank space.

A possibility is set here separately: in some relations, continuously occupying the channel may fill that blank space, and withdrawal may take part in forming newly usable room. That sentence is set here; it is not the upstream conclusion.

And three things must be kept apart. The three gates give the triggering conditions under which the unilateral path must suspend; they do not exhaust every reason for removal. A subject may also choose removal by their own law, by the conditions under which the relation can carry the handling, or by a fit procedure. Avoiding filling the space may take part in explaining why one withdraws in this way at this time, but may not be written as the cause or the guarantee of the other's later growth. One cannot write the other's later growth into one's own work, cannot claim to know how this withdrawal will complete their rupture, and cannot on this basis predict that they will change. A possibility is a possibility, not a judgment.

And during a suspension, the decision to suspend remains open to review. This is not the original conflict-handling still running, and not the channel from the original other party to me still standing open.

---

29 · Drawing together

Across the three technical destinations, the scope exits and the state of suspension, the four constraints do not stand in the same position.

After a terminus of consensus or acknowledgment, the first three are completed or satisfied and no longer require, in the present tense, continued deepening. The decision to stop remains reviewable, and generates no requirement of continued contact.

Resting at a local stopping point, the first three are suspended in the background and resume when the conflict is reactivated. The decision to declare a stopping point is not to be exempted in principle.

After removal or handover to arbitration, the first three terminate as this first-person handling conducted by me within the original relation. What procedure carries is the dispute about action; not the first three, not one's own law, and not first-person confirmation. The decision to remove is itself not to be exempted in principle.

During a suspension on the unilateral path, the first three are suspended in the background. The decision to suspend is not to be exempted in principle, and restart follows the previous section.

After object dissolution, the first three no longer have an object to run on; and that revision itself remains reviewable: was it completed by me, or completed for me by pressure, dependency or fear.

After object correction, the four are registered neither as completed nor as failed; only that this case did not belong within scope. The prior mistaken entry judgment, the acts of handling and the decision to exit remain reviewable. Discovering that it was not a conflict after all does not mean that what was said, what was done and what harm occurred are written back into not having happened.

What the second column in all of these says is one and the same thing, and it is not that "the four are still running." What continues is the reviewability of the decision to stop, an effect further upstream, not the original conflict-handling still under way.

None of the three technical destinations guarantees that no conflict of the same type will occur again. The same two laws may still require incompatible action in another situation. That incompatibility cannot be excluded in advance is said once more here at the close.

What, then, does all of this do.

It does not guarantee that the number of conflicts in a life falls. It does not guarantee that I do not change. What it constrains is the direction in which each conflict is handled, and whether, if my law changed within it, that rewriting was still completed by me.

Some things have been passed out. How judgment forms and how it is exercised. How action arrangements are formed, chosen, executed and reviewed. How this whole order runs on an actual day. These pass to Action Theory. This is a forward-pointing debt, recorded here, to be publicly honored, reassigned or withdrawn.

---

30 · What this treatise has not done

First, what it does hold.

What is held here is not that two laws never change, nor that every conflict at last reaches consensus, but that several things cannot stand in for one another: behavior within a relation cannot judge for me whether this law of mine is mine; a one-sided understanding cannot pass for a shared reading; an action arrangement cannot pass for either party's first-person revision; and neither party can certify or cancel the other's undertaking on their behalf.

Accordingly, no four virtues of how conflict ought to be handled are proposed here. What is done is to confine conflict-handling to a first-person constitutive operation: the object cannot be rewritten, one's own law cannot be written for one, the other's first-person undertaking cannot be certified or cancelled on their behalf, an arrangement cannot pass for a revision, and a one-sided understanding cannot pass for a shared reading.

The following are expressly not done, listed out rather than left to be inferred.

No procedure is given for determining which layer the other runs on. The other's end is not certified as having attained the standing of a law of its own, nor cancelled on their behalf for want of such certification. No closure-guaranteeing criterion is given for determining whether my own line is genuine. No method is given for reaching consensus. No norms of apology, forgiveness and reconciliation are given, and other ledgers are not prevented from listing them as conditions in a given case. Harm that has occurred and been self-judged is not handled; that has its own venue. Harm in progress and not yet self-judged is not handled; it is expressly held over. Conflict between two cannot-nots within one person is not handled; its clearest kind is treated in Value Theory Paper 5. No closure-guaranteeing external final criterion is given to distinguish a stopping point from stalling; external evidence, risk signals and possibly accurate supportive judgments still stand. Whether the other's origin layer has changed is not certified on their behalf, even where their action has plainly changed. The internal machinery of the Power Series and the Rights Series is not unfolded; only their domain boundaries are used. Cultivation, non-cultivability and institutional substitution, already treated in Economics Series Paper 2, are not re-established. No normative force is drawn on anywhere.

Back, at the last, to those two people.

They may reach the three-way sort, with neither law changed and the thing no longer happening daily. They may reach an acknowledgment that these two do not fit, each keeping their own, rearranging the household or going separate ways. They may also never reach anything at all, and simply stop raising it from some day on.

From outside, the third may not be distinguishable from the first two. For each bearer, the division between a stopping point and stalling can be made on their own handling ledger; a third party can also take a supportive judgment from whether an arrangement exists that is public, stable, restatable and genuinely deactivating of the object. The two readings may differ, and the external judgment may be wrong. What is denied here is that any party or any procedure holds a closure-guaranteeing final criterion, not that external evidence and accurate judgment are possible.

One thing is visible from outside: whether they have closed the channel of review on this matter. That is not a criterion; it is a trace.

And wherever they arrive, one sentence holds. This conflict has its boundary, and may reach a terminus, and may fail to. These two people have no terminus.

---

Acknowledgments

This treatise is a standalone paper in the supplementary layer of the SAE (Self-as-an-End) philosophical system, positioned alongside the Legal Series and the War Treatise. Its philosophical argument was completed independently by the author. All argument, all choices and the final text were adjudicated by the author, who bears sole responsibility. A four-way cross-review method was used in composition: internal precision auditing, ontological-limit and contradiction detection, boundary testing, and structural gatekeeping with pre-publication sign-off.

Drafting assistance and the four reviews above were all conducted with the aid of large language models. Drafting and internal precision auditing were undertaken by Claude (Anthropic), ontological-limit and contradiction detection by Gemini (Google DeepMind), boundary testing by Grok (xAI), and structural gatekeeping with pre-publication sign-off by ChatGPT (OpenAI). Thanks are due to these systems and to the research and engineering teams behind them. Their comments were adjudicated one by one by the author, with both adoptions and rejections recorded.

Several load-bearing points were settled through review. The scope was changed from a single gate to two, and an operation layer was newly set: the object remains, ontologically, two independently formed laws each its bearer's own, while actual entry confirms only one's own end, does not certify whether the other has attained standing, and does not cancel on the other's behalf for want of such certification. The two paths were renamed from subject-layer terms to current-relational-condition terms, and the ground of turning between them was changed from a determination about the other to whether the current reciprocal process is available. The modality of the four constraints was changed from an external-force form to a conditional-constitutive form, with it written explicitly that stopping the handling does not prove the constraints wrong. The unilateral path's closure was renamed from terminus to local stopping point, and the general statement of the third constraint was accordingly re-established on a shared core, with the two paths' realizations as its unfolding. Action arrangement was separated from law as its own layer, its jointly generated form on the reciprocal path taken from Economics Series Paper 1, and the Final Volume accordingly gained arrangement consensus as a kind. The stop gates went from two to three, and the judgment of route on restart was separated from conditions listed on other ledgers. The Final Volume further gained two scope exits, object correction and object dissolution, so that a case in which one party genuinely completes a first-person revision without a shared reading forming has an address; and the third constraint was written to state that genuine revision is not commanded by directedness-toward-terminus, and that once it occurs and dissolves the object, the constraint loses its object rather than obtaining a third terminus.

An upstream-paraphrase audit is a standing product here, comparing this treatise's paraphrases with upstream originals line by line. A sentence-level citation verification table is issued separately before publication, together with an archive manifest of the source snapshots on which it was checked; the paper is not published without that table's sign-off. The four-way review markedly improved this treatise's precision and consistency, and thanks are due for it.

---

References

All works below are archived on Zenodo, with Chinese and English versions in the same record. The DOIs cited are concept DOIs, which always resolve to the latest version; version numbers are therefore not noted here. The version DOIs and snapshot hashes on which sentence-level verification was performed are given separately in the release package's citation verification table and are not mixed into this list. All works are single-authored (Han Qin). Drafting and review assistance is described in the Acknowledgments. Works are ordered by upstream series position, not by first appearance in the text.

Qin, H. (2026). One's Own Law: The SAE Critique of Ethics and Morality [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.19037566

Qin, H. (2026). How Is the Inter-Ontological Possible [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.19297847

Qin, H. (2026). One's Law Meets One's Law: The Structural Origin of Law from 14DD Showdown (SAE Legal Series Paper 1) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.19548237

Qin, H. (2026). SAE War Treatise: How Cultivation Is Not Swallowed by War Itself [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.19656617

Qin, H. (2026). The Four Foundational Theorems of the Moral Law (SAE Moral Law Series Paper 1) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.20011018

Qin, H. (2026). Procedure of the Moral Court (SAE Moral Law Series Paper 4) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.20066987

Qin, H. (2026). Positive-Sum: The Ontological Ground of 15DD Mutual Chiseling (SAE Moral Law Series Paper 5) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.20081078

Qin, H. (2026). Four Layers of Economic Rationality: A Diagnostic Extension of the Rational-Agent Assumption from Dimensional Sequence Theory (SAE Economics Series Paper 1) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.19358010

Qin, H. (2026). The Other End of the Bridge: Economic Rationality When 15DD Faces a Party Who Does Not Enter 15DD (SAE Economics Series Paper 2) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.19381666

Qin, H. (2026). On the Origin of Power (SAE Power Series Paper 1) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.20370224

Qin, H. (2026). On the Limitation and Violation of Basic Rights (SAE Rights Series Paper 5) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.21212595

Qin, H. (2026). On the Conflict of Value (SAE Value Theory Series Paper 5) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.22122819

Qin, H. (2026). On Value and the Other (SAE Value Theory Series Paper 6) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.22191071

引言

一对住在一起的人，为一件事僵住了很久。

一个人认为，家里的事该说清楚再做，没说清楚就动手，等于把另一个人排除在外。另一个人认为，能自己做完的事不必先开会，事事都要商量，等于不信任对方能把事情办好。

这两条各自都不是偏好。为了把问题看清，这里先按纯型设定：两端都已经是各自的法。实际进入处理时并非如此，那时我只能确认自己这一端，这一点第二节与第三节会说。第一个人在一段很长的时间里学会：被绕过去的人最后会被当成不存在。第二个人在同样长的一段时间里学会：什么都要请示的人最后会失去做事的资格。两个人都不是在耍脾气。两个人都试过让步，让完之后那件事又原样回来。

他们试过讲道理，讲到最后发现对方全都听懂了，而且听懂之后仍然不同意。他们试过折中，折中方案执行两次就散了，因为它同时违反两条。他们试过不提，不提之后那件事变成了家里所有别的事的底噪。

一个人到这里会问：还有什么办法。

而在问这句之前，有一件更要紧的事没有被看清。这两条不是可以拿来比大小的两个意见。它们是两条法。 一旦是两条法，比大小这个动作本身就不成立，不是因为它们分量太近，而是因为要给它们排出高下，须有一把尺同时站在两套法之外，并把两个第一人称的位置换成同一种量。这两件都办不成。

接下来的十节先把对象，边界，层次与路线放对。其中会立起处理的类型，但不给一套可以照搬的操作流程。位置放对之后，办法才有落脚的地方。

---

上卷 · 定位

一 · 哪一类碰撞落在这里

一场碰撞落在这里，须过两道门。

第一道在第二节说，问的是这场碰撞的结构：是不是两条已成形的法在同一情境要求不相容的行动。

第二道在这里，问的是这场碰撞的处理会牵动什么：它的处理方式，是否会把我能否继续做第一人称的修订，能否继续受问的那个位置，置于风险中。

一场碰撞若怎么处理都不触及那个位置，它可以是麻烦，可以是损失，可以是一件很伤心的事，但它不是这里所说的冲突。有些账算得清，有些话说开就散了，有些事一方让一步而让的人心里没有被拿走什么。这些都好，也都不必用后面那套东西。

两道门须同时过，由此可以推出一件与直觉相反的事。碰撞的大小不定射程。 一件很小的事，若以压制的方式处理，也可能触及那个位置；上面那对人僵住的可能只是谁去交电费。一件看上去很大的事，若双方各有余地，可能从头到尾一次也没有触及。

这道门锚在我这一侧，是有理由的。

为他者留出条件这件事贯穿全篇。它落在每一步的做法上，也落在什么时候该停手上。但它不落在筛选案件的那道门里。一旦把他者一侧写进入场条件，整篇就会缩成这样一句话：我先判断你值不值得，再决定管不管你。而那正是全篇要拆掉的东西。

他者一侧若要以某种方式进入这道门，只能以这一形式出现：本案的处理是否可能使我占据替他构成的位置。这一问问的仍是我。

还有几个名目，看起来能定射程，其实不能。

意见不同，利益相争，误会，对决。这四个里没有一个足以单独把一场碰撞判进来，也没有一个足以单独把它判出去。利益能不能计量，这里不作预设；有些利益换算得了，有些换算不了，这与它是不是这里的对象无关。误会澄清了就散的，本来就过不了第一道门；而一场冲突可以在双方全都听懂之后仍然完整地在那里。

对决要单说一句：它是处理冲突的一种动作，不是冲突的一种对象。 一场冲突可以走到对决，也可以从头到尾没有对决。把对决列为冲突的一类，是把动作错认成了对象。

也因此，可以诊断的是这一次的这个操作，不是这个人。《自己的法》立过：没有终审的恶人，但有可诊断的恶的操作。这句话在这里的直接后果是，全篇不给出任何判定对方运行在哪一层的检验。

最后划出一格。一个人自己有两件都放不下的事，而它们在一个具体情境里要求不相容的行动，这是另一件事，不在这里。其中最清楚的一类，即同一承担者名下两项已经认领的价值内容之间的冲突，已论于价值论 P5《论价值的冲突》。若两条要求不能都还原为已认领的价值内容，剩下那一部分这里明文留置，交由行动论。

另有一处界要标明。价值论 P6《论价值与他者》在价值那一侧立过三条限界，读不尽，比不了，合不拢，并明拒给出程序。这里处理的是法这一侧：两条已经成形的法在同一情境里要求不相容的行动时怎么走。那三条限界这里不重立，也不借作推导的前提。 两侧各自成立，互不代替。

---

二 · 对象门与操作门

第一道门的内容是这样：

两条已经成形的自己的法，在同一情境里要求不相容的行动，而三件事同时成立：不能互相认证到达，不能替对方立法，也不能由任何一方单方退出而完成结算。

这句话若只这样放着，会锁死。

锁在哪里。若必须先确认对方那一端确实是自己的法，我才有权启动，那么我永远无权启动，因为认证对方的到达这件事，下一节就要禁止。而若不确认即可启动，那么句子里的两条法就不是一道操作的门，只是一句本体上的描述。

解开的办法是把两层分开。

对象层：这里所论的对象，在本体上确实是两条各自形成的自己的法，在同一情境中要求不相容的行动。

操作层：我只能确认自己这一端。另一端以对方第一人称承担并持续组织其行动的一项要求出现。我不认证它是否已经取得自己的法的资格，也不以未能认证为由替他撤销。

两层分开之后，两句话都成立，且互不侵犯。冲突处理按不可代位的原则启动。若后来确认对象类型并未成立，那就退出，而不是倒判对方。退出说的是这件事不归这里管；倒判说的是对方那一端不够格。前者是关于射程的，后者是关于人的，而后者这里没有权限作出。

现在看那三件收窄条件。

不能互相认证到达。 两个在该关系域各自运行于 15DD 的第一人称位置，即使推出了同样的律，这不是共识，不是共享，是各自到达；我不能拿我的到达去确认他的到达，也不能拿他的到达来支撑我的到达。这一条《自己的法》已经立过。它在这里的后果是：两条法之间没有一个位置可以站上去说这一条比那一条更真。不是那个位置还没找到，是那个位置的存在本身与两条法各自的第一人称来源相矛盾。

有一个听上去很客气的说法：不说谁更要紧，说两边一样要紧。它同样不成立。一样要紧仍是一把尺上的两个刻度，它没有取消那把尺，只把两个读数放在了同一处。

这里要划一道总边界，否则整句会被读得太宽。不能比较的是两侧第一人称的来源地位，不是所有行动事实。 安全风险，伤害，权利边界，法律权限，程序优先与资源条件仍然可以比较，也可以被裁决。这些裁决安排行动，却不能倒写哪一条法在第一人称上更真。

不能替对方立法。 法可以说你不可以压垮对方；法不能说你该让步。后一句是替对方立法，那就是殖民。这一条法学 P1 已经立过。

但这一条须紧跟一句分判，否则它会把该留的东西一起封掉。我可以陈述理由。我可以提出请求。我可以说明边界。我可以提出行动方案。真正被禁止的是另一件：我的法不能单凭自身取得替你结算「你必须让步」的权限。把我的理由当成已经替你成立，才是替你立法。 不作这一分判，说理与请求与共识与外部仲裁会被全部提前封死，而那不是这一条要禁止的东西。

不能以单方退出完成结算。 这一条要写准。它说的不是物理上离不开。它说的是：不能由任何一方单方面退出，便把这场不相容在其第一人称责任，共同关系，第三方责任或制度后果上结清。

这样写才容纳得下亲密关系，共同养育，组织角色与公共责任，也才与末卷允许双方各自离开不冲突。离开可以发生，离开本身却不使那些责任结清。而单纯不喜欢离开，沉没成本，或者关系的惯性，都不足以单独开这道门。

---

三 · 这一端是不是我的法

上一节已经把两条法写进了对象。这一节只查我这一端。

因为有一件事是可能的：我以为是我的那一端，可能并不是我的。

它可能是被装进来的。它可能是某一段关系留下的形状。它可能是一种反复出现却从未被我认领的要求。《本体之间如何可能》§3.1 处理过这种结构，并给出三项线索。

第一，撤掉外部条件之后它是否还成立。没有人看着的时候，没有人会因此夸我或怪我的时候，它还在不在。

第二，它能否与我其他重要的目的共存，而不把它们全都吃掉。真的那一项倾向于在别的事情旁边站着；装进来的那一项常常要求把别的全清空。

第三，暂时中断它的时候，出现的是懊悔与想回去，还是一种整个人塌下去的感觉。前者支持它是我的一件重要的事；后者支持我可能是靠它撑着，而那是另一回事。

三项只能支持或者削弱某一种判断，不是充分必要条件。它们不保证判对。

而且三项全都要求第一人称的内部读数。撤掉外部条件之后那件事还在不在，只有站在里面的位置读得到。主体拥有不可替代的第一人称读取位置，不等于他必然读准，也不等于他此刻必然已经知道。 把这三项拿去核对方那一条真不真，是替对方立法。

这里须把三档分清，因为后面还会回到这个形状。有线索，没有可保证闭合的终局判准，另有可以从外面看见的痕迹。

第一档是上面那三项。第二档是说：不给判准，不等于没有线索；被否定的只是可保证闭合的那一种。第三档在这一问上也有：在可得的材料里，那项要求是不是只在某一个人在场，可被看见或者会得到评价的时候才出现；是不是随一段关系的结束而消失；有没有在无人注视时仍被执行的证据。这些是推断性的痕迹，支持或者削弱判断，不判真伪。 它们看的是可得材料中的条件相关性，看不见一切无人知道的事。 三档的完整形状到第二十六节还会用一次。

而在线索与关系事实与自我回看之间作出裁断，是判断力的事。判断力如何形成，如何执行，交由行动论。主体判错也是可能的，这里不担保。

若这一端的确认不成立，处理的方式很要紧。不宣布那项要求是假的。也不启动后面的四条。只登记一件事：本案的对象类型尚未取得。随后转入殖民与内摄的诊断，那是另一套工作。

这不是某几条定理落空。是全篇从未启动。

---

四 · 界与两账

前面那对住在一起的人，僵持了几个月之后，有一天一个人说了一句很重的话。那句话说出来之后，房间里的事就不只一件了。

现在有两个不同的问题。

第一个问题：那两条法的行动要求是否仍然不相容。第二个问题：刚才那句话是否已经造成并且须要承认的伤害。

第一个问题是这里主审的。第二个问题不是，它另有裁断处，见道德律 P4《道德庭程序》。

要紧的是：这两个问题不是二选一，而是两张账并行记。 同一件案子完全可以同时有它们。一场冲突也可以一次伤害都没有发生，只是两条法撞在一处，谁也没说重话。

两张账并行，就有两条不能走的路。

道歉不消解两条法的不相容。 那句重话被认了，被修复了，两个人和好了，而家里的事该不该先商量这一问，一个字都没有动。把伤害的了结当成冲突的了结，是用一件已经结算的事去覆盖一件尚未结算的事。

也不得反向使用。 不得以那件大事还没解决为由，把说出去的那句话挂起不认。两件事各自结算。

还有一段，两套裁断机器都不结算它：冲突正在进行时新发生的，双方都还没有自判的伤害。 已发生并已自判的那一种，另有裁断处；尚未自判的这一种，那一处也接不了。

这一段这里明文留置，不假装它已经被安置。

留置不等于无事。留置不阻止事实核对，不阻止保护性退出，不阻止相应的权利或安全机制先行。这些动作也不替当事人完成另一套裁断所需要的自判。尤其要说明：单向路的继续不构成对这一段的处理。

而伤害这一账保有它自己的权限。道歉，承认事实，修复，赔偿，安全保障，这里不把它们立为一切案件的普遍义务。但伤害账，权利与安全机制，既有程序，或者主体自己的法，完全可能把其中某一项列为这一案件的必要条件。 这里不越那道账界。

另有一格也要划开。两条法在 14DD 上相遇，已论于法学 P1。那是对象来源的归属，不等于这里处理时主体位置降至 14DD。

---

五 · 分级交接

有一种情形，处理它的不该是这一篇。

前面那对人，若其中一个开始这样做：把另一个人的收入管起来，把他与朋友的联系一点点切断，每一次讨论都由自己定题目与时间，让另一个人越来越难以自己走开。这时房间里发生的已经不只是两条法不相容。

当不对称的操作开始实质压缩我的中断，退出，表达或基本条件时，这里不再单独主审。 此时与权力论，权利机制，法与安全机制并行或交接。保护可以先于任何互向处理。

再往前一级另有门槛。战争论自陈它只处理 14DD 一方的体系目的扩张或扩张准备，且以战争之外的整条处理谱系已经穷尽为退到极端态的条件。只有当对象进一步满足这两条，并且我这一侧的涵育条件已经被实质卷入，才转入《SAE 战争论》。

两级不得跳接。个人关系中的控制，资源不对称，议程操纵，依赖制造与退出压缩，通常止于第一级，不自动达到第二级的门槛。把它们直接交给战争论，等于让一段本该先接保护的关系去等一个永远达不到的门槛。

这里只用权力论 P1《论权力的起源》与权利论 P5《论基础权利的限与侵》的辖域边界与交接条件，不展开它们的内部机器。

---

六 · 冲突不能免除

两条法各自内生，各自到达，不由同一道程序产生。既然如此，没有任何理由保证它们在一切具体情境里的行动要求总是相容的。

而互凿之后的残余余项不必然比之前小。道德律 P5《正和》用分形海岸线说过这件事：勾勒得越细，露出的边界可能越多。两个人谈得越深，可撞的面并不一定越少。

由此可以说明回避是什么。回避不是冲突已经得到处理的结算方式。 不相容没有因为回避而结清，它仍然留在共同的场域里。它可能延宕，可能转形，可能暂时去激活。

这里不由此预判后果一定更坏。说的只有这一件：回避没有结算它。

而不能免除不等于不该避免。能不冲突就不冲突。改一个安排使那一步不必发生，是好事。

这里要按时态分账，因为第八节还要用同一个动作。对象尚未成立时改变条件，叫预防或者避免。对象已经成立之后改变条件，使当前的不相容解除或者去激活，叫行动安排。 动作可以相同，理论身份不同：前者在冲突之前，后者是冲突处理的一种。

全篇从避免在这一件事上失效之后开始。

---

七 · 两种关系条件，与由此分出的两条路

分路不按人分，按当前的关系条件分。

互向条件是这样两件事同时成立：在这个关系域上，双方当前都能承认，对方第一人称承担的那项行动要求不由我替他立，也不由我替他撤；并且我自己的判断可以受对方追问。

这里确认的是不可代位的关系条件，不是彼此对已经到达的资格认证。这一点与上一节的操作层同版。

互向条件当前可用时走的路，称互向路。互向条件当前不可用时走的路，称单向路。单向指的是约束只在我这一侧运行，不是指我单向控制对方。

两条定义都只说条件，不说人。条件是当前的，是可修订的，是由这段关系里已经发生的事显出来的。

选路的机制要写准，因为这里最容易滑回判人。

不作前置判定，只指不判层级，不判内心。它不指不核安全，不核退出，不核资源不对称与程序操纵。 上一节那道保护门与基本权力条件检查通过之后，先按互向路起手。

起手之后，共同深入要么发生，要么不发生。若他不进入互凿，若每一次都退回把我的法当成可以被压垮的东西，那么在这个关系域上互向条件事实上不可用，转单向路。

转的根据是当前互向过程不可用这件事本身。 共同深入没有发生，可以因为不愿，因为暂时没有能力，因为时点不对，因为不信任程序，也可以因为对方确实把我的法看成可以压垮的东西。路线转换只需要这些的共同结论，不必再加一项关于对方内心或层级的解释。

不安全要分两种。 若只是不适合此刻进入共同深入，而尚未命中主体位置，退出，安全与权利那几道保护条件，可以转单向路。若不安全已经命中第五节的第一级交接，或者命中后面要立的停止门，那就先暂停，先保护，或者移交，不进入单向的介入。

这样写还有一个理由。一个还没有走到那一步的人，与一个走得到却选择不走的人，在单次或局部的互动里可能呈现相同的行为签名。长期的关系事实可以改变我的判断权重，但不能完成认证。而路线转换本来就只依赖当前互向过程是否可用。

有一条结构上的解释可以说明互向条件为何常常失效：一个还在 14DD 上运行的位置，没有结构上的理由承认另一条法具有与自己同样的第一人称绝对性。这解释了这种失效为何在结构上可能发生。它不说明某一个当前案件中互向条件已经不可用，也不承担分路的任何一根柱子。

---

八 · 行动安排与法不是同一层

回到那对人。

有一天他们不再争论该不该先商量，而是把家里的事分了三类：一类各自做完再说，一类做之前说一声，一类必须两个人一起定。分完之后那件事就不再每天发生了。

这里发生了什么，须要说准。

两条法一条都没有改。 一个人仍然认为被绕过去的人最后会被当成不存在；另一个人仍然认为什么都要请示的人最后会失去做事的资格。

改的是别的东西。时间，资源，角色，程序与行动条件被重新安排，使当前的行动不相容解除或者去激活。而安排本身不生成第三条自己的法，也不凭自身改写任何一方的法。

不立这一层，那天发生的事会被误读成四件之一：某一条法被修订了；出现了第三条共同的法；一方替另一方立了法；或者只是把问题拖着。四种误读都不对，而它们之所以出现，是因为没有一个位置容纳「两条法都没变而问题解决了」。

这一层里有一种特别的：在双方各自的不可让渡之间，可以找到一个解，它不等于任何一方的初始位置，而它不要求任何一方放弃自己的不可让渡。经济学 P1《经济理性的四个层级》立过它的产生条件：双方都有不可让渡；这个解不等同任一方的初始位置；寻找它依赖对双方底线的真实探询；而若这样的解不存在，未形成这个解可以被登记为过程的有效结果，不因未成交而自动判为失败。它是产出，不是妥协，因为它是双方进入过程之前谁也没有的新解。三类分法就是这样一个解。

但它是这一层里的一种，不是这一层的总名。单方的自侧调整，保护性安排，外部程序的裁决，都属于行动安排这一层，都不是那种共同生成。

单向路上这一层仍在，只是没有共同探询。单方能做的是改变自己这一侧的时间，资源与行动条件，以及提出方案。不能单方宣布一个共同生成的解已经成立。

安排如何形成，如何选择，如何执行，如何回看，交由行动论。这里只做类型的分界。

---

九 · 手段位与目标函数

处理冲突是手段。方向是同时保存两侧主体位置的开放性：对外不替他立法，对内不让冲突替我改写。前者叫涵育，后者叫自我涵育。

两侧同时写出来，是为了挡住这样一种读法：为了涵育对方，所以我要处理他。

目标函数因此不是赢。真正的输是这一件：

冲突中，我的法可以被我改写；不能由冲突替我改写。

这句话须要两处解释，否则会被读成两个极端。

一处是它不禁止变化。 冲突当然可能给我理由，给我材料，给我压力，给我新的事实，并因此触发我的修订。被禁止的不是冲突对我有因果影响；被禁止的是它占据最终的第一人称确认位置。冲突可以触发，促成并改变我的可见材料；不能替我完成「这一改写确实是我的」这一构成确认。

互向路更是明确要求我在对方的参照系面前重新审视自己的构，而这可能导致我修订自己的法。那是路的正常运行，不是失守。

另一处是它不给判准。 这是我认同的修订，还是我被殖民了，由主体自己判，靠判断力。这里不给可保证闭合的判准，主体也可能判错。

我这一侧继续做第一人称修订，继续受问的那个位置，写作自我涵育。它对应的反面是自我殖民。它与对外涵育同名而不同向，凡出现处逐处标明。

还有一件不在选项里的事要说明。

一个人有时会想：我干脆装作没看见对方也有他自己的位置，把这件事当成一次利益谈判来打，会不会省事得多。

这一步不在选项里。 不是因为它不道德，而是因为我已经在本案中确认，看见对方的主体位置属于我自己的法。因此我可以暂停，可以撤出，可以移交，却不能把装作没看见写成同一位置中的一项处理策略。另有两件事可以作旁证：余项一旦产生就不可清零；降级会失去对他人意图的分辨深度。这两件都是旁证，不承担不能降级的构成性模态。

---

十 · 三处与战争论不同

写到这里，与《SAE 战争论》的差别已经可以说清。那一篇处理的是体系目的压迫下的极端状态，这一篇处理的是个人尺度上两条法的相遇。三处不同影响后面各卷的写法。

第一，第一人称责任不能换人承担。 个人冲突里，决定的人，执行的人，承受的人未必是同一条命，伤害可能落在第三个人身上。这里能取得的只有一句：我不能把自己行动的第一人称责任换人承担。

第二，没有统一可仪式化的战时边界。 不像那一篇有清楚的战时与平时之分。但在冲突尚未激活时，仍然可以建立程序，可以约定退出方式，可以设置第三方机制。

第三，没有终止仪式。 冲突常常只是不再提起。拖着不终很容易成为默认的路径。后面立第三条约束时，这一点会比在战争论里重得多。

另有两个名目须要归位。有一种状态是那件事根本没有形成需要处理的局面，它是未激活，属于冲突之前的那一段，不是后面单向路上的一种办法。而动用关系里的不对称，在这里属于第二条约束所说的手段与权力问题，它的辖域交接见第五节第一级。

---

中卷 · 四条约束

十一 · 这四条是哪一种不得不

下面四条都以不得不起句。起句的形式要先说清楚，否则后面四节全部会被读错。

这四条不是四条规范。 它们不是四项应当遵守的准则，也不是四种美德。它们不发出规范效力。

它们也不是因果自动式。 不是说冲突一出现，这四个动作就自动成立了。

它们是条件—构成式：

只要我仍以第一人称的位置处理这场已经过门的冲突，并仍承认自己的法不能由冲突代写，那么正视，深入，向终，受问就是这个处理位置本身不能取消的条件。

于是共同语法不是「我有权正视」「我有权追问」。不是我取得了四项主权，而是我自己的法使四项豁免不再属于我可以承认的处理方式。

这一写法有一个后果必须当场说明，因为它常被读反。

我可以停止这一操作。 我可以暂停，可以退出，可以转交程序。停下不证明这四条错了，只表示这项处理不再运行。被禁止的只有一件：我不能在仍然宣称占据同一位置的时候，取消其中任何一项。

停下之后是什么状态，也要写准。停止处理以后，前三项不再以进行时运行；第四项也不产生继续接触或继续回应原对方的要求。但停止，移出，仲裁或暂停这一决定本身，仍不得被原则性地置于理由审查之外。这里继续下去的是一条更上游的效力，不是原来那场冲突处理仍在运行。

四条只在几件事都已成立时启动：那场碰撞过了两道门，我这一端的确认已经成立，并且第五节那道保护门已经通过。普通的意见不合，可退出的关系，可由既有规则直接结算的争执，都不在这里。

四条全部是我的第一人称约束。它们不给我判对方的权限，也办不到让对方也这样。这是它们与一切冲突处理技术的分界。

正视与深入不要求维持接触，也不取消安全撤离。伤害那一账另行并行结算，四条都不结算伤害。

四条并置，穿透的深度依次加深：进入，延续，方向，自身位置。它们不是四张并列的清单，是四个同心圆。

---

十二 · 不得不正视

冲突已经发生时，我不得不正视它。操作层上的正视，是承认：我这一端已经确认的自己的法，与对方以第一人称承担并持续组织其行动的那项要求，在当前情境中不能共同实现。我不把这一不相容完全还原为误会，时机，情绪或单方过错，也不由此认证对方已经取得自己的法的资格。

为什么这是一个条件而不是一项建议。

若我用误会，用时机，用某一方的过错去完全还原或替换一个已经过门的双主词结构，我不是在解决它，而是在改写对象。改写之后我处理的已经不是原来那件事。正视是保持对象同一性的最低条件。

这里的分寸要拿准，因为过窄会封掉不该封的东西。

被禁止的是完全还原，不是讨论。 误会，情绪，时机与过错，完全可能与真实的不相容并存。那对人确实有几次是话没说清，确实有一次赶上了很坏的时点，确实有人说了重话。讨论这些不取消双主词结构。伤害那一账尤其可能必须判断过错。 被禁止的只是以其中任何一项完全取代已经成立的不相容。

改写有五种形状。

改写成误会：我们只是没说清。改写成时机：这事以后再说。改写成情绪：他今天心情不好。改写成对方的错误。改写成我的错误。

后两种要单独说，因为它们看起来相反，作用却相同。

两者都在把一个双主词的结构问题换成单主词的过错问题。 归咎是把它交给对方，自责是把它揽到自己身上。自责看起来像担当，甚至看起来比归咎更负责，可它同样取消了不相容：若这件事是我不好，那么两侧的要求不能共同实现这一结构就不必再面对了。

回避正是被这一点吸引过去的。过错有解，而不相容没有那种解。 一个人宁可承认自己不对，也不愿承认那里有一件谁都没有做错而仍然解不开的事。

正视之后有三件事不随之而来。

正视不等于开口。 正视是我这一侧的承认。说不说，什么时候说，怎么说，是行动的事，这里不给。

正视不等于立刻处理。

正视不要求在场。 这一条要写死。未承认冲突而用另一个名字盖住它然后离开，是回避；先行撤离危险，停止接触或保护边界，不因此构成回避。正视是第一人称的承认，它可以在撤离之后发生，也可以在很远的地方发生。

由此可以说清走开这个动作。走开只是一项外部动作，它本身不取得技术身份。 同一个离开，按离开时冲突处理处于哪一种状态，至少分四种：

未正视而换一个名字离开，是回避，违反这一条。为安全，边界或事实核对而先行离开，是保护性撤离，不违反。没有取得双边之终而终止这场处理在原关系中的运行，是移出；只是暂时中止而尚未转交的，是暂停。双方取得终之后各自离开，是终之后的行动。

分判不只看有没有正视，更要看离开时这场处理处在哪一种状态。而若仍有共同资源，第三方责任或程序义务，离开本身不结清它们。

---

十三 · 不得不深入理解

已经正视，我不得不深入理解这场冲突。

理由是直接的：若不澄清不相容究竟落在哪一组行动要求上，任何调整都无法知道自己改的是冲突还是别的东西。那对人若不弄清争的是「先商量」还是「被当成不存在」，他们改一百次做法也只是在原地换姿势。

理解的方向要说准，否则会滑成两种偏差。

一种是把力气用到外面去。 有六种形状，性质并不相同，须分开说。

把这一件事扩成「你总是这样」，以及从这件事漂移到「这段关系还值不值」，这两种较明确地更换了对象。原来要澄清的是一组具体的行动要求，现在要澄清的变成了一个人的性格或一段关系的总账。

拉第三方与翻旧账则要留余地。拉第三方有时是取得证据，或是识别一个反复出现的结构；翻旧账有时是判断当前这一次是不是同一个模式的又一次。它们可以是漂移，也可以不是。

升级语气与手段，以及动用关系里的不对称，这两种主要不是对象问题，是手段与权力问题。它们的辖域交接在第五节第一级。

另一种偏差是把理解当成前置条件，什么都不敢做。 这也要挡住。保护性退出，临时暂停，资源隔离，保持现状，这些可逆的安排可以先行。若必要的保护措施是不可逆的，它的权限由权利，法或者安全那一账给出，不由这一条授权，也不冒充这场冲突的终。 这些安排都不冒充终局处理。若冲突处理继续或者重启，对对象的澄清仍须完成；若已经负责任地移出，这里不另发持续澄清的要求。

理解的对象是不相容的形状，不是对方这个人。

而理解有底。深到不相容看清为止，不是深到理解一切为止。 余项不因理解而清零，谈得越深可撞的面并不一定越少。若把理解一切当成目标，这一条就成了一件办不到的事，而下一条要求的向终会随之落空。

深入也不要求持续接触。这一条是我的第一人称约束，它不生成对方持续谈，持续听，持续解释的义务，也不生成我对对方的这种要求。

最后一句回指第三节：我这一端的自查仍然只能我做。双方即使进入同一互向过程，也不使我这一端的自查变成共同的自查，那三项线索也仍然不得拿去核对方。

---

十四 · 不得不向终

只要冲突处理仍在运行，我不能把冲突的维持本身抬成目的。

向终是这项反自我维持条件的名称，不表示每一条路都能取得技术意义上的终。互向路中，它落实为朝向共识或互相承认不可通约；单向路中，在原对象仍然成立且处理仍在运行时，向终的反自我维持要求落实为：使当前不相容经行动安排或者激活条件的改变而真实去激活，或者把处理负责任地移出。若我这一端后来发生真实的第一人称修订，使原对象消解，那不是本条要求我朝向的第三种结果，而是本条自此失去对象；它作为射程退出另行登记。止点，移出与对象消解都不因此改名为终。

压垮对方不是终，是把冲突换成了压制；拖着不结也不是终，是把冲突换成了常态。

理由在第九节已经给过：处理冲突是手段，因此不能把冲突的维持本身抬成目的。这是共同核，下面两条路上的落法都是它的展开，不是对它的例外。

终只有两种，而且只在互向路上取得。 共识，或者互相承认不可通约。两类终都要求双方进入同一互向过程，不要求物理同场或者持续接触，而两者都不是我一个人能够取得的东西。

单向路上没有双边之终，这一点要正面承认，不要含糊过去。也正因为没有终，那条路才必须有底。底是什么，留到末卷说。

单向路另有一种不由本条要求，却会使本条停止适用的射程退出：我这一端完成真实的第一人称修订，使原对象消解。

这里几句要一起写死。本条不禁止真实的修订。本条也不要求我为了结束冲突而修订自己。修订一旦真实发生并使对象消失，四条不再以冲突处理的进行时运行：前三条失去原来的冲突对象，第四条也不生成继续接触或者回应的要求；而那一次修订与停止的决定，仍处在更上游的理由回看之下。这里没有取得第三种终。 它不是处理策略，不是终，也不由向终推出。它的身份留到末卷。

有一句核句要写在这里，因为它最容易被上面那段推翻。

关系可以终止，通道可以关闭；不能因此终审这个人。

现实的关系与接触的通道可以永久终止。这一条不要求保留联系方式，不要求保持可接近，也不要求随时允许重启。它禁止的是另一件事：把对方当前的不可互向，写成他作为主体的终局。

有两种结果尤其容易被误认成终，须单独排除。止点，移出与暂停同样不是终，但它们各有技术身份，留待末卷分账。

压垮。 它确实会使这件事不再被提起，可它是把冲突换成了压制。

拖着。 它的形状是不再提起，绕开，变成家里所有别的事的底噪，以忙碌覆盖。个人冲突没有降书，没有停战协定，常常只是有一天不再说了。因此拖着很容易成为默认的路径。 这一条约束在个人尺度上比在战争论里重得多，正是因为这里缺一个把终标出来的仪式。

向终还有两件事不包含。

向终不等于求快。 为了尽快取得一个终而承认一个我并不承认的共识，是用这一条去违反前面那一条。求快的诱惑是回避的另一种形状。

向终不等于我促成那个终。 终需要双方。我能做的是我这一侧每一步都朝向它。要求对方同意，就回到了替对方立法。这一条在四条里最容易破，因为朝向与促成在感觉上很难分开。

最后要给移出一个准确的位置。移出终止的是这场处理在原关系中的继续运行，它不称作朝向了结。 这一条不因移出而被违反，但也不能把移出改称为终。

---

十五 · 不得不受问

我不能把自己在冲突中的行动原则上置于理由审查之外。

第一句先写死：受问是我受问。 这一条不给我追问对方的资格。四条里只有这一条在句法上主词可以被读反，因此它必须放在最前面。

为什么这是一个条件。分两步。

第一步：可修订性要求理由原则上能够到达我。 一个把自己原则上置于一切理由审查之外的位置，与「我的法可以被我改写」不能共存：免检的位置使修订失去了发生的入口。若没有任何理由能够到达我，那么我的法就不再有被我修订的机会，只剩下被我持有。

第二步：防止我独占理由的选择，解释与裁断，要求至少有一条不完全由我控制的外部复核路径。 第一步只推得出理由须能到达，推不出这条路径必须在我之外；而一个由我全权决定哪些理由算数的通道，等于没有通道。

这个证明只覆盖原则性的免检。它不覆盖任何一次具体的不回应。

道德律 P1《道德律四条基础定理》的第四条与它同形：不得不让自己的不殖民结构暴露在追问中，以防自我殖民。那一条同时给出了一个双层的精度，这里直接沿用。

第一层是资格与有效性的筛选。 我不能以自己为最终裁判，预先宣布某一类追问者永远没有资格。这一层我不能豁免。一旦我取得了这个权限，我就成了自己的终审。

第二层是回应的路由与安全。 什么时候回应，由谁转达，是否暂停，如何保护隐私与安全，这一层我必须决定，也只能由我决定。

两层的关系是这样：适格，可用，且不被争议一方控制的制度存在时，主要的筛选可以由制度负担，因为共同体的程序可以区分真实的追问，噪音，操纵与伤害。制度缺席时，主体可以延期，可以转介，可以设置边界，但不能把这些安排升级为原则性的永久免检。

这一条不等于四件事，须逐件排除。它不等于任何人随时有权盘问我。它不等于我必须即时回答。它不等于私密信息必须交出。它不等于对方因此取得了判定我内心的资格。它要求的是不把自己置于审查之外，不是随传随到。

维持通道也不等于我必须去问他。通道开着说的是我不关；它不生成我向对方发问的义务，更不生成审讯权。

关闭通道有五种形状：你不懂；你没资格问；等这件事过去再说；我已经仁至义尽；我都是为你好。第四句在单向路的暂停期尤其容易出现，第五句同时是殖民的标志。但这五句只有在被用来终止一切复核时才是关闭的形状，它们本身不是自动判据。 一个人说了一次「等这件事过去再说」，可能只是那一天说不动了。

追问有四个来源，分两类身份。对方与第三方是外部的可问性。当时的我与日后的我是补充的来源。自我追问不能替代外部的追问通道，否则这一条会从被问滑回封闭的自我反省，而封闭的自我反省正是它要防的那件事。

所需要的不一定是原对方直达我的那一条通道。 现实的关系与直接的接触可以永久关闭；外部的可问性可以由适格的第三方，程序，或者后来的公共复核承担。而只有适格，可用，且不被争议一方控制的制度，才有资格负担主要的筛选。

最后一句要写准，不能写过头。这一条是四条里最容易留下公共痕迹的一项，也是一项不可替代的防闭合条件。但它不足以认证前三条是否真的发生过。 一个人完全可以保持通道开着，同时没有正视，没有深入，也没有朝向任何一个终。

---

十六 · 四条之间

四条的穿透深度不同。

前三条各自约束一个环节：进入，延续，方向。第四条不同，它约束的不是某一个动作，是我作为主体在这场冲突里的存在形式。因此它最深。

有一个二分要避免：说前三条对外而第四条对内，是不准的。四条全部是我的第一人称约束，这是它们的发生位置，都在我这里。 分别在于主要对象：前三条主要面向冲突的对象与我的行动，第四条主要面向我自己的免检位置。

也因此四条同时护住我这一侧。一个把自己原则上置于审查之外的人，已经打断了自己的自我涵育，无论他在那场冲突里表现得多么克制。

而四条全部是第一人称的，这就决定了这里给不出一套调解术。四条都办不到让对方也这样。若有人想拿它们去要求对方正视，要求对方深入，要求对方向终，要求对方受问，那已经不是这四条，是把第一人称的约束外推成了对人的伦理，而那一步在《自己的法》里已经被判为结构上非法的越界。

这四条能做的只有一件：使我在处理这场冲突时，不至于把对象改写掉，不至于让冲突替我改写我自己的法。 它们不使冲突变少，也不保证冲突有解。

---

下卷 · 两条路

十七 · 两条路怎么分

四条约束是同一套。落在两种关系条件中，落法不同。

分路按当前的关系条件分，不按人分。这一句在第七节已经立过，这里再说明它的三个后果，因为下面两节最容易在这三处滑回判人。

第一，路线选择是暂时的读数。 它随关系状态修订。它不是一次判定的产物，也不需要一劳永逸。

第二，路可以换。 换路不需要重走正视那一步，因为路本来就是过程中的读数，不是入场时的一次分类。互向条件后来恢复，就转回去；后来失效，就转过来。

第三，路名不免责。 两条路上都可能发生伤害，走在哪一条上都不构成对伤害的豁免。单向路尤其不提供忍受压迫的义务：第五节那道保护门在两条路上同样有效，任何一条路都不能盖过它。

还有一句要先说明。经济学 P2《桥的另一端》在经济语境里展开过这样一个象限：我看得见对方的位置，而对方看不见我的。那一篇处理了非对称的感知，涵育，承认不可涵育，以及退出成本与制度替代。下面所说的单向路是同一个象限在冲突处理这一侧的落法。那一篇的结论这里直接用，不重立。

自己的法可以由自己修订。后面要立的那几道门守的是改写由谁完成，不是内容不能变。

---

十八 · 互向路：共同深入

互向路上，双方都进入同一个互向过程。这不指物理上同场，也不指持续接触。

深入在这条路上是共同的。它不是各说各的，而是在对方的参照系面前重新审视自己的构。那对人若真走到这一步，可能发生的不是一个人被说服，而是双方各自看见了自己那一条先前没有说明白的地方。

这里可以形成一种收益，须写成可能式。互凿可以使双方的立法更活跃，这是道德律 P5 立过的。它可能带来的不是分了什么好处，而是勾勒本身激活了双方各自的构。

但同一件事也可能带来别的：疲惫，伤害，以及新暴露出来的不相容。一次真正的共同深入之后，两个人可能比之前更清楚彼此哪里合不上。

因此有三句限定要跟上。

第一，收益不指伤害消失。 这条路上仍然会发生伤害，第四节所说的两账并行在这条路上同样有效。不得用「我们这是正和」去否认「刚才那句话确实伤到人了」。

第二，余项变大不自动判失败，也不自动判精度提高。 谈完之后两个人看见的差异比之前多，这本身既不说明谈砸了，也不说明谈好了。要看的是那些新增的差异能不能定位，能不能重述，能不能被追问，以及对象有没有保持同一。能，才有资格把它登记为精度可能提高，而这一登记仍可被后来纠正；不能，就只是又多了一片模糊。

第三，我这一端的自查仍然只能我做。 双方进入同一互向过程，不使我这一端的自查变成共同的自查。那三项线索也仍然不得拿去核对方。

而上一卷第八节所说的那种共同生成的解，只在这条路上有产生的条件，因为它要求对双方底线的真实探询，而探询要求双方都进入同一互向过程。

---

十九 · 互向路：向终

这条路上的终有两类：共识，或者互相承认不可通约。共识内部有两种发生史：安排共识与修订共识。另有两种可能都不是终。一是深入之后发现对象类型原本没有成立，那是第二节那道退出的事后发生，撤销的是入场。二是对象成立过而此后由一方的真实修订使它消解，却未形成共同读数。两者都属射程退出，与共识的两种发生史在末卷分别处理。

这里只说一句：两条法都没有改变而问题解决了，这一类是终，不是拖着。

两种终之间不排序。不写哪一种更好，也不写它们同样好，因为后一句仍是一把尺上的两个刻度，它没有取消那把尺。

这条路上也可能不终。双方都在互向条件之内，走到了，而没有取得共识，也没有取得互认。这与转路要分开：

不终是互向条件仍然可用而没有取得终。转路是互向条件当前不再可用，无论它从未启动，还是启动之后中断，还是一度可用而后来失效。

前者留在这条路上，后者转到下一条。

---

二十 · 单向路中的涵育

换一个场景。

一个人在一处做了很多年的事，与一位在其之上的人有一件长期解不开的分歧。他认为一件事该怎么做有它自己的理由；那位在其之上的人则把每一次不同意都读成态度问题。他解释过，对方听完之后说的是「我知道你不容易」。他退让过，退让被记成了「他终于想通了」。他试过把话说重，说重之后那件事变成了对他这个人的评价。

这里发生的不是他表达得不好。是他在这段关系里说什么都会被译成另一套东西。互向条件在这个关系域上当前不可用。 而他走不掉，或者说走开并不能把那件事结清。

单向路不是涵育的同义词。 它还包括自侧的行动安排，边界保护，不再施加，暂停与移出，这些动作不都以改变对方那一侧为对象。

只有当我仍对对方一侧作积极介入时，这项介入若要留在冲突处理之内，它的方向才是涵育：为他者发现并发展他自己的法提供潜在条件。 涵育不构成介入的义务，也不把单向路改写成一项对他人的成长工程。

四句话要紧跟着说清楚，因为这个词最容易被读成它的反面。

第一，涵育不是教化，不是改造，不是把对方变成什么。 这条路的名字里不放涵育两个字，正是因为一旦叫作涵育之路，它就会被读成「我判你较低，所以由我来涵育你」。名字管入口，内容管做法，两件分开。

第二，第一步不是让他看见我。 经济学 P2 §4.1 说过：一个更展开的位置只能从一个稳固的位置里长出来。因此第一步是帮他更稳地站在他自己那些放不下的事上。在此之前，让他考虑我的位置的努力有高风险被译成压他让步的策略信号。上面那个人每一次解释，对方听到的都是「他在争」。

第三，有一条事前的结构条件。 这一条第一人称可核，不需要判对方的层级或内心：

干预不能压缩对方拒绝，中断，修订，退出与重新表述自己的权限，通道及现实可行性。

提供理由可以改变他愿不愿意拒绝，却不能降低他能否拒绝。这一分别要紧：若不作，摆出根据这个动作本身会被判成殖民。

一项介入若压缩对方说不的权限，通道或者现实可行性，不管它看上去多像帮忙，都不属于这里所说的涵育性介入。

第四，有一组事后的迹象，但它们只是支持性的。 对方更能说明自己，更能拒绝我，这些可以支持「这里发生了涵育」的判断。它们不是充分必要的判准，更不是我认证他成长的凭据。

由此可以说清依赖这件事。由干预制造并锁定的，使对方不能对干预者说不的那一种依赖，是殖民的温柔形态。 但并非一切依赖都是殖民。人本来就互相依赖，孩子依赖照顾，病人依赖医治，学徒依赖师傅。被判为殖民的只有那一种：它是被造出来的，而且它锁住了说不的通道。

最后两句：涵育不预设成功。它也不预测对方会变。做这些事的理由不在预测里。

---

二十一 · 单向路中面向对方的四种位置

战争论 §5 立过五个名目，作为战争之外的处理谱系，未逐一展开。其中最后一个说的是那件事根本没有形成需要处理的局面，它属于冲突之前的那一段，已在第十节归位。这里展开其余四个，落到个人尺度上。前三种是介入的操作，第四种不伐是不介入的姿态。四者都不是单向路上一切动作的穷尽分类：自侧安排，边界保护，暂停与移出都不在这四种之内。那一篇把五个立为一条上行的序列，这里只取名目，不取序列身份，理由见下。

伐谋：改情境，使那不相容的一步不必发生。上面那个人可以把需要当面争的事改成书面的，把两个人的事改成三个人在场的，把每次都要临时决定的改成事先定好的。

伐辩：把我这一条的根据摆出来，不要求他接受结论。

伐思：只求那个问题在他那里立住。

不伐：不再施加，只维持条件。

这四种是不同的位置，不是一条从低到高的序列。有一种读法说越往后施的力越小，可是没有一个共同的量纲支持它：改情境有时比摆理由强势得多，因为它动用的是资源，空间，时间与第三方的结构。这里只主张这四种不同，不主张它们递减。前三种是操作，第四种是姿态。

伐辩与伐思的分界要写在伐辩上，不能只写在伐思上。 辩是把根据摆出来，思是只要那个问题在他那里立住。两者都不要求他接受结论。 若只在伐思上写这一句，伐辩就会与说服术同形，而说服术正是这一整条路要避开的东西。

伐思还有一道限定，因为它最容易变质。那个问题必须保持可拒绝，可改写，可中断。 一个不能被对方拒绝的问题，不是伐思，是伐辩的伪装，或者已经转入了施压与权力操作。把一个问题种进别人心里让他甩不掉，这不是让问题立住，这是别的东西。

伐谋也有一道限定。改情境改的是那不相容的一步的发生条件，不是对方的日程，关系或选择。这里要说准：不能仅凭「我在处理冲突」或者「我在涵育你」取得单方安排对方生活的权限。另有的角色权限，法律权限或程序权限，分别按它们各自的辖域结算，与这条路无关。

最后一句关于力的去处。当前的对象已经去激活之后，继续追加的力不再属于对这一场冲突的处理。它若仍然发生，转入权力，伤害或者其他行动的账。这不是谦让，是那一份力已经不对着任何一件正在处理的事。

---

二十二 · 单向路：单方理解的界

这条路上没有足以形成共同读数的共同深入。对方仍然可能在某几个点上参与，但那不构成共同的读数。

由此有这条路上最要紧的一句：

我理解了他为什么这样，不等于他同意了我的理解，更不等于这件事就此说定。

把单方的理解当成共同的共识，是替对方立法。而这是最容易被当成体谅而放行的一种：我这么替他着想，怎么会是殖民呢。上面那个人若在心里把对方的行为全部解释通了，然后按这个解释安排后面所有的事，他做的就是这一件。

单方理解也不得当成我已经尽力的凭据。 「我都理解他了，还要我怎么样」，这句在形状上就是关闭追问通道的第四种。理解不产出免检权。

---

末卷 · 终 止点 移出

二十三 · 三类技术去向与两种射程退出

先分两层。

对仍在射程内的处理，可登记的技术去向分为终，止点与移出三类。若对象校正或者对象消解先使案件退出射程，则不进入这三类。

对象校正：后来确认入场判断原本没有成立。撤销的是此前的分类。

对象消解：入场判断原本成立，但此后我这一端完成了真实的第一人称修订或者退出，使原来的行动不相容不再成立。若尚未形成双方的共同读数，它不是终，只是原对象自此退出射程。 若双方随后共同确认新的兼容，才登记为修订共识。修订共识从共同确认形成的那一刻取得，不追溯把此前单方发生的对象消解写成当时已经存在的共同终。

操作层还有一条限定。我可以确认我这一端的真实修订；若变化只发生在对方一端而互向过程仍不可用，我不能替对方认证来源层已经改变，只能按可见的行动关系重新判断对象是否仍被激活。

下面回到三类技术去向。这三类不是三种结局，它们的身份不同。

共识与互相承认不可通约是终，由双方取得。局部止点不是终，它是当前行动不相容的去激活。移出也不是终，它是这场处理在原关系中的运行终止或改由程序承接。

下面五个词各有专用，不得互换。日常话里的结束，收场，了断，不再计较，都不足以指定它们中的任何一个。

身份	内容
终	互向路中由双方取得的共识或互相承认不可通约
止点	已形成一项可说明的安排或条件，使当前的行动不相容真实去激活；不冒充终；激活条件复现时可重开
移出	原关系不再承接这场处理，或者处理交由适格的程序承接
暂停	单向路因停止门命中而暂时中止，具备条件时可重新判路
拖着	没有形成任何技术收束，只以沉默，回避或背景化替代处理；不是合法的第四种终

暂停不是移出的一种形状，两者分列。对象仍成立且取得可登记的技术去向时，分为终，止点与移出三类。处理也可能暂时停在暂停，或者始终没有取得技术收束而拖着；对象校正与对象消解则属于射程退出。

有一句要放在全卷最前面。

一场冲突有可识别的对象边界，并可能取得终。这个人没有终。

这里所收束的是这一件不相容，不是这段关系，更不是这个人。

还有几句从前面继承下来的，在整个末卷都有效。三类技术去向与对象消解都不追认既往，也不结清伤害那一账；对象校正只撤销这里的入场分类，不倒写在误入场过程中已经发生的行动，关系与伤害。 承认对方的来源位置不签发行动许可。仲裁只接行动上的争议，不接第一人称的构成。而一次终了不保证以后不再冲突。

---

二十四 · 共识的两种发生史

走到这里之前，可能先发生两件事，它们都不是技术去向。

谈到后来发现，那里并没有真实的行动不相容：一个人以为对方要的是事事请示，另一个人以为对方要的是所有事都得开会，而两个人真正在意的东西从来就没有撞上过。

这是第二节那道退出的事后发生：这件事不归这里管。 它取消的是入场，不是收束。当初判进来的那件事，本来就不该进来。若把它算作一种终，会让人误以为深入总能把不相容化掉，而那不是深入能做的事。

另一件是对象消解。 那件事确实成立过，而后来一方真正由自己的法完成了一次修订，原来的要求不再要求不相容的行动。若双方共同确认了新的兼容，那是下面第二种，是终；若没有形成共同读数，它只是原对象退出射程，不是终。

下面两种才是共识的发生史，两种共识都是终。

第一种是安排共识。 两条法都没有改变，双方共同形成了一个解，使当前的行动要求可以共同实现。家里的事分三类，就是这一种。

这是终。而且这一类必须在这里有位置，否则第八节新立的那一层就没有出口：两条法都没变而问题解决了，这一类会无处安放，只能被误读成拖着。

第二种是修订共识。 一方或者双方分别完成了第一人称的修订，新的行动要求取得了兼容。这里要写准是一方或者双方，不是必须双方都改。一个人在对方的参照系面前重新审视自己的构，改了；另一个人没有改；而两条现在合得上了。这仍然是终。

两种都不由说服的动作构成。说服可以提供理由，并因果地促成一方修订；而说服者的成功本身不构成共识。 共识的构成落在双方各自完成的第一人称承认上，或者落在双方共同形成并分别承认的行动安排上。

两句限定要跟上。

共识不自动生成第三个第一人称的立法者。 双方完全可以从此有一条持续的规则，家里的事就照三类分下去。但那条规则的约束力须分别经过双方各自的法，或者来自双方已经承认的程序。它不是一条凌驾在两条法之上的第三条法。各自独立持有这一条，不构成互相立法。

共识不追认既往。 它了结的是此刻这一件不相容，不改写此前任何一方做过什么。

---

二十五 · 互相承认不可通约

第二种终要把内容写准，因为它最容易被写成一句空话。

互认所共同确认的，不是彼此已经替对方认证了到达。那正是第二节禁止的动作，若终局要靠它，全篇就自相矛盾了。

互认所确认的是这样两件：

当前两项由各自第一人称承担的行动要求不能共同实现；且任何一方都无权替另一方撤销其承担。

这与操作层完全同版。对象层说的是两条法，操作层说的是我只确认我这一端；而互认所认的是承担与不可撤销，不是资格。我不需要认证他到达了，才能承认我不能替他撤销。

由此有四句要跟上。

只认来源位置，不签发行动许可。 承认不等于他的行动因此可被允许，不等于我须承受它的后果，不等于公共规则不得限制它，也不撤销伤害与权利那一账。上一节那个人承认他的上级也有他自己在意的东西，这不使那些译码变得正当。

承认不可通约不等于承认对方是对的，也不等于我错了。 这一句要写死，否则整个第二种终会被读成各打五十大板。

不可通约不是理解失败。 它是在深入所规定的对象边界内，已经看清行动不相容的形状之后的读数。它与「我们没谈明白」的分界，就在深入那一条有没有走完。

承认之后各自保留，关系可续可各行。 双方各自离开可以发生在这一终之后；离开本身不构成这一终。若仍有共同资源，第三方责任或程序义务，离开也不结清它们，只能转入安排或仲裁。

---

二十六 · 局部止点

单向路取不到双边之终。它在原关系内部唯一能够取得的技术收束是止点。 除此之外它还可以暂停或者移出，但后两者不是这条路取得的终，也不是止点。

局部止点是已经形成一项可说明的安排或条件，使当前的行动不相容真实去激活。 上一节那个人可能把需要当面争的事全改成了书面的，从此那件事一年也不发作一次。

它不冒充终。 构成这场冲突的那种行动要求关系没有被来源层改写；激活的条件若复现，两侧仍可能要求不能共同实现的行动，原对象可以重新进入处理。

止点改的是行动安排或者激活条件，不是两条法的来源地位。而止点可以来自非共同生成的自侧安排，即他一个人改了自己的做法，对方并没有参与。一旦共同探询恢复并形成了共同生成的解，就该切回互向路，登记为安排共识。

现在到了全篇最难的一处，须直说。

止点与拖着从外面没有可保证闭合的判准。

拖着，是没有形成针对不相容本身的任何安排，只以不再提起替代处理。止点，是已经形成了一项可说明的安排或条件。两者在第三人称那里可能看起来一模一样：此刻都没有不相容的行动，都不再提起，都很平静。

仅凭表面的沉默或不再提起，第三人称不能保证分开这两者。 外部可以取得支持性的证据，比如那些安排是不是公开的，是不是稳定的，是不是可以重复观察到的。但没有一个可保证闭合的终局判准。

有一样东西可以从外面看见，须写准它的份量。

宣布止点而同时关闭一切复核通道，是这个止点可能只是拖着的一项强风险信号。它足以使这个止点声明的公共可信度显著下降，不足以替主体完成内部的分类。

这里三档要分清，全篇在三处用到同一个形状。有线索，即那三项自查。没有可保证闭合的判准，这一点在第三节已经立过。另有从外面看得见的痕迹，即这一条。三者不能互相冒充：线索不是判准，痕迹也不是判决。

---

二十七 · 移出

第三类去向是把冲突移出这段关系。

移出是原关系不再承接这场处理，或者由适格的外部程序承接，它不叫终。外部仲裁是移出的一种。

单向路暂停不是移出。 暂停是这条路因停止门命中而中止，尚未转交，也尚未终止原关系作为承接的场域。两者不得互换。

外部仲裁要先划一道限定，因为这里最容易被反向利用。

把裁决权让渡给外部，不是主体性的失败，是主体性的边界意识。但这只对某一类外部成立：双方已经承认的程序，依法有权的机制，或者为保护基础权利而有正当权限的制度。

把裁决权交给任意的外部者，并不自动体现边界意识。 由强势的一方单独选定的第三方，可能只是那一方权力的延伸。上一节那个人若被要求把分歧交给上级指定的仲裁人，那不是移出，那是同一件事换了个名字继续。

让渡出去的东西也要写准。让渡的是行动争议的程序裁决权，不是自己之法的构成权。 仲裁可以决定谁在什么时候使用哪一份资源，哪一项行动可以实施，程序怎么安排。它不能决定哪一条法在第一人称上更真，也不能决定谁的主体资格更高。

制度的位置由此也清楚了。经济学 P2 §5 立过：制度不创造我们，制度降低退出的成本。制度可以改变信息，激励，接触的结构与安全的条件，因而可能帮助承认发生；但它不能替主体完成承认，也不能保证承认。法说「你不可以把她当作手段」，这是本体命令的外部强制版：它划出禁区，违反有后果。它是我们不在场时的替代与脚手架，不是我们本身。

---

二十八 · 三道停止门与重启

单向路可以一直走下去吗。不可以。有三道门，任何一道命中，这条路暂停。

门一：我的主体位置或者退出条件正在被主动压缩。 这是安全门。它命中的时候同时触发第五节的分级交接，因为那已经不只是冲突处理的事了。

门二：我的第一人称改写位置正在被压力，依赖或者恐惧夺走。 这也是安全门。要注意它守的不是我的法有没有变化，而是这一次改写由谁完成。

门三：当前的关系域已经取得不可涵育的判断。 经济学 P2 §4.3 把它写成关系诊断而不是道德判断，也不是能力判断：当前条件下，继续投入的预期边际收益低于预期边际成本。这是主体对当前关系的局部判断，不是一把跨主体的共同量尺，也不要求数值化。

三道门要并置，因为每一道单独立都会被读偏。只写门一，会读成只要对方还没明着压，我就得一直走下去。只写门二，会读成我不舒服就可以停。只写门三，会读成算得过来才做。

门一还要跟一句：缺少互向条件本身不构成恶。 一个人在这个关系域上运行不到那一步，这不是恶。主动取消另一主体的位置，或者压缩他的退出条件，才进入恶的操作的诊断。

暂停之后是什么。

暂停只是暂停。 人会变化，条件会变化。但这一句不保证以后一定还有机会，它只说明暂停不被写成终局。暂停的是这条路，不是对这个人的终审。

重启的条件必须把两账分开，否则会越界。

这里不把道歉或者原谅规定为一切案件的普遍重启条件。 那属于另一套裁断，不由冲突处理来立。

但也不排除：伤害那一账，权利与安全机制，既有的程序，或者主体自己的法，完全可能把某一项承认与修复列为这一案件的必要条件。

冲突处理只负责重新判一件事，就是路。若互向条件恢复了，转回互向路。若互向条件仍然不可用，而三道门都不再命中，单向路可以重启。 而其他账目所列的必要条件若尚未满足，这里不得越账宣布重启。

最后一件事要说，而且要紧接着把它封住。

《本体之间如何可能》§3.3 立过：一个位置从单一走向多元，通常需要一次断裂，而涵育的一项关键功能，是在断裂尚未发生的时候，为主体预留足够的空白空间。它没有说撤出创造空白。

这里另立一个可能式：在某些关系中，持续占据通道可能把那段空白填满，而撤出可能参与形成新的可用空间。这一句在这里另立，不是上游的结论。

这里要把三件事分开。

三道门给出的是单向路必须暂停的触发条件，它们不穷尽一切移出的理由。 主体也可以依自己的法，依关系的承接条件，或者依适格的程序，选择移出。

避免占满空间可以参与说明为什么此时这样退出，但不得被写成对方后来成长的原因或者保证。 不能把对方后来的成长预写成我的作品，也不能宣称我知道这一次撤出将怎样完成他的断裂，更不能据此预测对方会变。可能是可能，不是判断。

而暂停期间，暂停这个决定仍然可以被回看。这不是原来那场冲突处理仍在运行，也不是原对方直达我的通道仍然开着。

---

二十九 · 收束

在三类技术去向，射程退出与暂停状态中，四条约束的地位并不相同。

取得共识或者互认之终以后，前三条已经完成或者满足，不再以进行时要求继续深入。那个停止的决定本身仍然可以回看，但它不生成继续接触的要求。

停在局部止点上，前三条暂停在后台，冲突再激活时恢复。宣布止点这个决定不得被原则性免检。

移出或者交给仲裁以后，前三条作为我在原关系中进行的这场第一人称处理终止。程序承接的是行动争议，不是前三条，不是自己的法，也不是第一人称的确认。 移出这个决定本身不得被原则性免检。

对象消解以后，前三条不再有对象可以运行；而那一次修订本身仍然可以回看：它是不是我自己完成的，是不是被压力，依赖或者恐惧替我完成的。

对象校正以后，四条不登记为完成，也不登记为失败；这里只确认这一案件本不属于射程。先前的误入场判断，处理行为与退出决定仍然可以回看。发现原来不是冲突，不等于处理中说过的话，做过的事与发生的伤害都被倒写为没有发生。

单向路暂停期间，前三条暂停在后台。暂停这个决定不得被原则性免检，重启按上一节。

上述各项第二栏说的都是同一件事，而它不是「那四条一直在运行」。继续下去的是那个停止决定的可回看性，是一条更上游的效力，不是原来那场冲突处理仍在跑。

三类去向都不保证未来不再发生同型的冲突。同样两条法，在别的情境上仍然可能要求不相容的行动。不相容不能事先排除，这一条在收束处再说一次。

那么这一整套东西到底做了什么。

它不保证一个人一生中的冲突数量减少。它也不保证我不变。 它约束的是每一次冲突的处理方向，以及我的法若在其中变了，那一次改写是不是仍然由我完成。

有些事交了出去。判断力如何形成，如何执行。行动安排如何形成，如何选择，如何执行，如何回看。这一整道次序在具体的一天里怎么走。这些交给行动论。这是一笔前指的债，记在这里，将来公开兑现，改派，或者撤销。

---

三十 · 这一篇没有做的

先说它真正守住的是什么。

这里守住的，不是两条法永远不变，也不是一切冲突终于达成共识，而是几项不能互相代：关系里的行为不能替我判我这一条法是不是我的；单方的理解不能冒充双方的读数；行动安排不能冒充任何一方的第一人称修订；任何一方都不能替另一方认证或者撤销他的承担。

也因此，这里没有提出四种应该怎样处理冲突的美德。它做的是把冲突处理限定为一种第一人称的构成操作：对象不能被改写，自己的法不能被代写，对方的法不能被代认，安排不能冒充修订，单方理解不能冒充共同读数。

以下各项这里明确不做，逐条列出，不留给读者去推。

不给判定对方层级的程序。不认证对方那一端是否已经取得自己的法的资格，也不以未能认证为由替他撤销。不给可保证闭合的判准来判定我自己那一条的真伪。不给共识的达成方法。不给道歉，原谅与和解的规范，也不排除其他账目把它们列为某一案件的条件。不处理已经发生并已自判的伤害，那另有裁断处。不处理进行中而尚未自判的伤害，那一段明文留置。不处理一个人内部两项不得不的相持，最清楚的一类已论于价值论 P5。不给可保证闭合的外部终局判准来区分止点与拖着；外部证据，风险信号与可能准确的支持性判断仍然成立。不替对方认证他那一端的来源层是否已经改变，即使他的行动确实变了。不展开权力论与权利论的内部机器，只用它们的辖域边界。不重立经济学 P2 已论的涵育，不可涵育与制度替代。规范效力一律不支取。

最后回到开头那对人。

他们可能走到三类分法，两条法一条没改而事情不再每天发生。他们也可能走到承认这两条合不上，各自保留，把家里的事重新安排，或者各行各的。他们还可能一直没有走到任何一处，只是那件事从某一天起不再被提起。

第三种从外面看，与前两种未必分得开。对每一位承担者而言，他可以在自己的处理账上形成分判；第三方也可以依据是否存在公开，稳定，可重述并真实去激活对象的安排，取得支持性判断。 双方的读数可以不同，外部的判断也可能错。这里否定的是任何一方或者任何一道程序拥有可保证闭合的终局判准，不是否定外部证据与准确判断的可能。

有一样东西从外面看得见：他们有没有把这件事的复核通道关掉。这不是判准，是痕迹。

而无论走到哪一处，有一句在这里都成立。这一场冲突有它的边界，可以取得终，也可以取不到。这两个人没有终。

---

致谢

本篇为 SAE（Self-as-an-End）哲学体系补充层的一篇单论，与法学系列 战争论同位，其哲学论证由作者独立完成。全部论证 取舍与最终定稿均由作者裁决，文责由作者独任。成稿过程采用四方交叉评审方法：内部精度审计，本体极限与矛盾检测，边界测试，结构闸门与发布前签署。

起草辅助与上述四方评审均借助大型语言模型进行。起草与内部精度审计由 Claude（Anthropic）承担，本体极限与矛盾检测由 Gemini（Google DeepMind）承担，边界测试由 Grok（xAI）承担，结构闸门与发布前签署由 ChatGPT（OpenAI）承担。谨向这些系统及其背后的研究与工程团队致谢。它们提出的意见由作者逐条裁定，采纳与驳回均记录在案。

本篇若干承重处经评审改定。射程由单门改为双门，并新立操作层：本体上的对象仍是两条各自形成的自己的法，而实际启动只确认自己这一端，不认证对方是否已取得资格，也不以未能认证为由替他撤销。两条路的名称由按主体层级命名改为按当前关系条件命名，路线转换的根据由一次关于对方的判定改为当前互向过程是否可用。四条约束的模态由外力式改为条件—构成式，并写明停止处理不证明定理错误。单向路的收口由终改称局部止点，定理三的总陈述随之重立共同核，两条路的落法作为它的分路展开。行动安排与法分层，另立为一层，其互向路上的共同生成型取自经济学 P1，末卷因此新增安排共识一类。停止门由两条改为三道，并把重启判路与他账所列条件分开结算。

另有五处经正文轮改定：定理一的题面由对象层语法改回操作层语法；单向路撤去以涵育为整条路总名的写法，涵育只作面向对方之积极介入的方向；终，止点，移出，暂停与拖着五者的身份一次写定，走开不再按二分处理；末卷另立对象校正与对象消解两种射程退出，使一方真实完成第一人称修订而未形成共同读数的情形取得地址；定理三的题面随之写明，真实修订不由向终命令，一旦发生并使对象消解，本条自此失去对象，而不是取得了第三种终。

完整的版本沿革见发布包中的《版本说明与归档清单》，此处不备载。

本篇续行上游转述审计这一常设产出，逐条比对本篇的转述与上游原句。该表由起草席编制并初核，待独立闸门复核；未经该表签署不发布。四方评审显著提升了本篇的精度与一致性，谨此致谢。

---

参考文献

以下文献均存档于 Zenodo，中英双语版本同记录。所引一律为概念 DOI，始终指向最新版本，故不在此另注版本号。逐句核验所据的版本 DOI 与快照哈希另见发布包中的上游引用逐句核对表，不与本表混列。全部文献单一作者署名（秦汉）。起草与评审辅助已在致谢中说明。文献按上游系列的层位排列，非按正文首次出现的次序。

秦汉. (2026). 自己的法：SAE 对伦理与道德的批判 [预印本]. Zenodo. https://doi.org/10.5281/zenodo.19037566

秦汉. (2026). 本体之间如何可能 [预印本]. Zenodo. https://doi.org/10.5281/zenodo.19297847

秦汉. (2026). 自己的法遇上自己的法：14DD 对决中法的结构起源（SAE 法学系列 Paper 1） [预印本]. Zenodo. https://doi.org/10.5281/zenodo.19548237

秦汉. (2026). SAE 战争论：涵育如何不被战争本身吞噬 [预印本]. Zenodo. https://doi.org/10.5281/zenodo.19656617

秦汉. (2026). 道德律四条基础定理（SAE 道德律系列 Paper 1） [预印本]. Zenodo. https://doi.org/10.5281/zenodo.20011018

秦汉. (2026). 道德庭程序（SAE 道德律系列 Paper 4） [预印本]. Zenodo. https://doi.org/10.5281/zenodo.20066987

秦汉. (2026). 正和：15DD 互凿的本体根基（SAE 道德律系列 Paper 5） [预印本]. Zenodo. https://doi.org/10.5281/zenodo.20081078

秦汉. (2026). 经济理性的四个层级：维度序列理论对理性人假设的诊断与扩展（SAE 经济学系列 Paper 1） [预印本]. Zenodo. https://doi.org/10.5281/zenodo.19358010

秦汉. (2026). 桥的另一端：15DD 面对不进入 15DD 的对方时的经济理性（SAE 经济学系列 Paper 2） [预印本]. Zenodo. https://doi.org/10.5281/zenodo.19381666

秦汉. (2026). 论权力的起源（SAE 权力论系列 Paper 1） [预印本]. Zenodo. https://doi.org/10.5281/zenodo.20370224

秦汉. (2026). 论基础权利的限与侵（SAE 权利论系列 Paper 5） [预印本]. Zenodo. https://doi.org/10.5281/zenodo.21212595

秦汉. (2026). 论价值的冲突（SAE 价值论系列 Paper 5） [预印本]. Zenodo. https://doi.org/10.5281/zenodo.22122819

秦汉. (2026). 论价值与他者（SAE 价值论系列 Paper 6） [预印本]. Zenodo. https://doi.org/10.5281/zenodo.22191071