The Implicit Treaty

I. Endpoint Words

Some advice is true as a description and useless as an instruction.

He’s just insecure. Set boundaries. Communicate openly. Be authentic. Use common sense. Just ask directly. Be a team player. Trust your gut.

Each of these names a state in which a problem is already solved. None describes the path to that state. The advice points at where someone needs to be standing. It does not describe the ground between here and there.

A person told “he’s just insecure” hears a label that functions as if security were a switch the other person could flip on by recognizing the label. A person told “set boundaries” hears an instruction whose execution presupposes an entire catalogue: knowing what one’s needs are, distinguishing them from preferences, naming them in a moment, tolerating the other person’s reaction, holding the position when challenged, repairing the relationship afterward, updating the boundary based on what happens next. None of that catalogue is in the phrase.

The defect is structural. These phrases compress a working machine into a single pointer to its output. For someone who already runs the machine, the pointer is sufficient — they read it as shorthand for the operations they perform without thinking. For someone who does not, the phrase is an empty pointer.

Phrases like “communicate openly” fail because they are endpoint labels. They describe the relationship after a hidden stack of perceptions, skills, timing rules, emotional distinctions, and repair protocols is already running. To someone who has the stack, the phrase feels obvious. To someone who does not, it is an instruction with the mechanism removed.

II. The Hidden Stack

To execute “communicate openly,” a person must do the following, often in real time, often under emotional load: notice what matters before naming it, distinguish a feeling from a belief, identify whether the issue is a request, a boundary, a question, or an observation, choose the timing so the other person can receive it, choose a phrasing that names the issue without coding it as attack, tolerate whatever response the other person produces, check whether the message landed as intended, and update future behavior based on what came back.

That is eight separate operations, every one of them requiring its own subskill, every one of them invisible from outside the head doing them. The instruction “communicate openly” includes none of this. It points at the output of the eight-operation machine and treats the pointer as the recipe.

This is why the advice is reliably useless when it is most needed. The person who can already run the machine produces the output without coaching. The person who cannot run the machine cannot bridge the gap by trying harder. The gap is operational, not motivational. There is no runnable part of the machine available to them.

Consider the analogous instruction in another domain. “Just run the company well.” For the executive who has already absorbed the dozen sub-skills of running a company — reading a balance sheet, hiring against a brief, declining a deal, holding a meeting that ends, knowing when to delegate and when not to — the instruction is shorthand for what they do. For someone with none of those skills, it functions as a tautology dressed as a plan.

The compression is the same. The endpoint label compiles a long sequence of specific operations into one phrase whose readability depends entirely on whether the reader already executes the sequence. The phrase carries information for one reader and noise for another, and the speaker often cannot tell which is which.

The mechanism is general. Endpoint labels conceal the operations they refer to, and they conceal them most effectively from the people most fluent in the operations. Fluency is its own form of blindness — not to the world, but to the visibility of one’s own machine.

III. Why Fluency Hides the Mechanism

When a fluent communicator hears the phrase “Implicit Treaty” and answers “Isn’t this just open communication, said with fancy words?”, the response is not a failure to understand. It is the phenomenon itself. The endpoint label is transparent to someone whose stack already runs the operations. They read the new label, recognize the output, and reach for the closest existing pointer in their vocabulary. Because the new label and the old label both refer to states they already know how to inhabit, they conclude the labels are equivalent.

For the person whose machine does not run those operations, the new label and the old label refer to different things. The old label (“communicate openly”) points at a destination. The new label (“Implicit Treaty”) points at a structure that, if recognized, gives them somewhere to start. The difference is operational: one phrase tells you to land the plane; the other begins to describe the cockpit.

Fluency in any operational domain produces this selective blindness: the operations that constitute fluency are precisely the operations that drop out of conscious vocabulary.

When someone fluent at a relational protocol meets a vocabulary that names the protocol’s operations explicitly, the natural reaction is to compress the new vocabulary back into the existing label. This is what fluency makes available. The decompressed terms feel ornamental because the decompressed operations are invisible from the inside.

The skeptical compression also reveals a treaty clause. It says: new vocabulary in the relational domain is suspect until proven not to be relabeling. The clause is reasonable in many contexts. Enough pop-psychology vocabulary is relabeling that the clause often works and feels like discernment.

It fails when the new vocabulary is doing decompression rather than decoration. Two people can both value “open communication” and still disagree about whether a new term names a real mechanism or merely renames the obvious. If that disagreement is never stated, it becomes another treaty collision.

IV. The Implicit Treaty

An Implicit Treaty is the unconscious constitution a person carries into every interaction: the catalogue of expectations they treat as obvious, the operating defaults that govern what counts as care, respect, honesty, support, autonomy, directness, fairness, competence, and adulthood. The treaty sits underneath stated positions: the felt assumption about how reasonable adults of course behave.

Every person runs one. The treaty is invisible to its bearer in the same way the assumptions of grammar are invisible to a fluent speaker. It surfaces only when violated. When the other person fails a clause, the violation is experienced as a moral failure rather than a constitutional difference. They don’t care. They aren’t paying attention. They’re being unreasonable. The available diagnoses all locate the problem in the other person’s character.

There is no sufficiently detailed universal human treaty for the conflicts that matter most. There are broad human regularities, but the clauses that decide concrete conflicts — what counts as care, what counts as pressure, what counts as honesty, what counts as withdrawal — vary enough that treating them as universal is the first error. Each person’s compression of thousands of micro-decisions produces a treaty. Another person’s compression produces a different one. Both are coherent. Both feel obvious from inside. Each can violate the other without ever intending to.

Validation before analysis. If I express distress, you provide presence first; analysis comes later, if at all, and only after I have signaled openness to it. Reasonable inside a treaty whose operating principle is co-regulation: the distressed person’s nervous system needs to settle before the verbal cortex can engage. Useless or worse inside a treaty whose principle is repair: the distressed person came to you specifically because you might see what they are missing.

Solitude as restoration. If I need time alone, that is energy management, not commentary on you; you do not interpret my withdrawal as a judgment of our relationship. Reasonable inside a treaty where autonomy is a precondition of presence. Violated immediately by a treaty in which time apart, especially unrequested, is itself a relational signal — in which togetherness is the unit of love and chosen solitude is, by definition, a withdrawal of love.

Reasons as real reasons. If I state a reason, you treat it as my actual reason unless there is concrete evidence otherwise; you do not automatically translate it into a hidden motive, excuse, defense, or maneuver. Reasonable inside a treaty where explicit self-report is the basic unit of trust. Violated immediately by a treaty in which reasons are often treated as social objects: face-saving, pressure, avoidance, or retrospective justification. One person experiences disbelief as epistemic violation; the other experiences literal self-report as suspiciously overproduced.

Risk evaluated for growth, not only danger. If I propose a risky opportunity, you weigh the upside as carefully as the downside; you do not default to the conservative position and require me to prove the risk is acceptable. Reasonable inside a treaty where stagnation is the defining failure mode. Destabilizing when crossed with a treaty in which the family’s accumulated stability is the central asset to protect, and any voluntary risk to it is a betrayal of the work that built it.

Planning as care, improvisation as presence. If I produce a detailed plan, that is service; if I improvise, that is presence. You do not interpret my plan as control or my improvisation as recklessness. Each half is reasonable inside its own treaty. Both halves are routinely interpreted as proof of the partner’s defective character by someone running the other — the planner reads the improviser as chaotic, the improviser reads the planner as suffocating, and each is performing what their treaty defines as the correct way to show up.

Each of these clauses is a perfectly serviceable principle in some context. None is universal. When two people carrying incompatible treaties interact, the violations are real on both sides. The solitude-seeker is genuinely hurt by the togetherness-treaty’s interpretation of their need as rejection. The togetherness-seeker is genuinely hurt by the solitude-treaty’s interpretation of their bid for connection as suffocation. Neither is performing — both are reading the encounter through their own constitution and finding the other person guilty of breach.

The conflict feels moral because the constitution is invisible. Repair begins when the invisible constitution becomes visible enough to be amended.

V. The Mask

When voice and exit are available, a treaty collision can become negotiation or separation. When the relationship contains power asymmetry — child and parent, employee and supervisor, financially dependent partner and provider — exit may be unavailable or too costly to function as a real option. The lower-power party often has to survive inside the higher-power party’s treaty, and the remaining survival strategy is to adopt that treaty as their own performed signature.

This is the Mask: a relational operating protocol that is not native to the person running it, sustained because the environment treats the protocol as mandatory.

The Mask is something more specific than lying or surface-level adjustment to social context (the way one speaks differently to a colleague than to a child). It is the sustained, system-wide performance of a constitution that does not match the underlying architecture. The native protocol still exists. The native impulses still arise. They are simply not permitted to reach behavior.

The cost has four parts. Suppression blocks native impulses before they reach behavior. Performance generates the expected protocol. Translation converts native responses into the required interface in real time. Monitoring watches for slippage. These run continuously, at metabolic cost, for as long as the Mask is held, and they consume substantial cognitive capacity that would otherwise be available for external action. The phenomenology is recognizable: exhaustion that does not respond to rest, paralysis at tasks that should be manageable, a baseline of low-grade vigilance that does not switch off in safe contexts because the system has lost the calibration that distinguishes safe contexts from threatening ones. The internal coherence of the person drops, and the energy that had been available for the world is consumed by the management of the discrepancy between the native architecture and the performed one.

A communication method that works for one person can become another person’s Mask if the method is treated as a universal default rather than a negotiated agreement. Consider the partnership in which one person believes that “healthy communication” requires direct questions, named feelings, explicit requests, and reflective listening. For someone whose native architecture runs that protocol comfortably, this is just how they communicate. For their partner, every interaction now involves running an operating system that is not theirs — generating the interface their partner expects, suppressing the responses that would come naturally, translating between native and counterfeit in real time, monitoring for slippage, repairing performance when it slips. The communication is explicit. It is also, for the partner, a Mask. The cost compounds. The partner reports unexplained exhaustion, withdrawal, eventual collapse. From inside the first person’s protocol, the partner appears to be failing at “basic communication.” From inside the partner’s substrate, the partner is performing someone else’s communication, which is more expensive than communicating natively, even when natively imperfect.

A communication method that unmasks one person can become another person’s Mask.

Ordinary communication advice often treats the mechanism as one-sided: one person needs to learn to express, the other needs to learn to listen, both need to “improve their communication.” The unilateral framing skips the diagnostic question: whose protocol is the conversation being conducted in, and what is the cost to the person not running their native one? When the question is not asked, the person whose protocol is being run experiences the conversation as healthy adult behavior, and the person running someone else’s protocol experiences the conversation as exhausting performance whose cost they cannot articulate without being told they are resisting basic communication.

The Mask is rarely removed by insight alone. The pressure that produced it persists; without an alternative architecture, the Mask regrows. The repair has to happen at the level of the protocol itself.

VI. The Shared Skeleton

The repair mechanism for a treaty collision is the construction of a Shared Skeleton: a bilateral constitution that sits above both individual treaties and binds them both. Both parties’ native defaults remain different. The Shared Skeleton arbitrates between them through explicit articles that both have agreed to honor, rather than through whichever treaty happens to win the moment.

The Shared Skeleton is a shared operating protocol: a small set of agreed articles that name the recurring collisions and specify how each will be handled. The articles do not eliminate the underlying difference between the two treaties. They make the difference manageable by giving each party a procedure they can run in the moments where their native treaty would otherwise fire automatically.

The operative move is narrower than the slogan “make things explicit” suggests. That slogan can become another endpoint label. The mechanism is concrete: convert a recurring “of course” into a testable article. A useful article names the trigger, the expected behavior of each party, the reciprocal obligation, and the condition under which the article has succeeded or failed. “Of course you should support me when I’m upset” is still too compressed; it is the felt violation, not the article. The article is what survives translation into a sentence both can act on the next time the trigger fires.

The goal is not total explicitness. A relationship with no implicitness would be unlivable. Most shared defaults should remain silent because they work. The treaty needs excavation only where the same “of course” keeps producing injury, confusion, resentment, or Mask-cost. Surfacing what already runs cleanly is overhead without payoff. At worst, total explicitness becomes its own avoidance strategy: a way to manage intimacy instead of inhabiting it.

Construction proceeds in three stages.

The first is mutual recognition. Each party identifies the parts of their own treaty that have been treated as universal: the clauses they have been enforcing without ever stating them. This is harder than it sounds. The clauses are invisible from inside; the parties may have been together for years without ever having had to articulate the rule that has been silently governing their reactions. The work is to bring at least the clauses that produce recurring collisions to the surface.

The second is collision mapping. Each recurring conflict is examined as a treaty difference rather than as a moral failure. The argument about whether to take the job offer, the argument about whether the in-laws stay over Christmas, the argument about whose interpretation of “we’ll talk about it later” was operative — each becomes data about which clauses of the two treaties are incompatible. The point is not to win the past argument. The point is to identify the clauses that will produce future arguments unless something is built to govern them.

The third is article drafting. For each mapped collision, the parties draft an article that both can inhabit. The article has to be specific enough to actually constrain behavior in the moment, and it has to leave both parties’ native expression intact wherever expression does not violate the article.

A worked example for the validation-before-analysis collision:

When one person expresses distress, the other provides presence first. Analysis is offered only after consent, or after a previously agreed cooling interval in which both have accepted that analysis may be offered. In return, the distressed person agrees to treat later analysis as care rather than attack.

The article does not declare validation correct and analysis incorrect. Both responses are preserved as legitimate. What is removed is the automatic firing of either response as the universal default. The validating partner does not have to suppress the eventual analysis; they have to defer it. The analyzing partner does not have to suppress the analysis itself; they have to sequence it. Both are now bound to a procedure neither of their native treaties would have produced. Both have surrendered the right to enforce their treaty automatically. In exchange, both have gained a procedure that will not require either to mask.

“Article” here does not mean paperwork. Some couples may write the clauses down. Others may only need a sentence both remember, a shared phrase, a ritual, or a one-line rule. The form is irrelevant. The test is whether the next instance of the pattern runs differently from the last one.

The articles compound. A handful of them, each governing one recurring collision, produces a coherence in the relationship that neither party’s individual treaty could have produced alone. The relationship is no longer being run by whichever treaty wins the moment. It is being run by the Shared Skeleton, which both have authored and both have agreed to.

The Shared Skeleton requires maintenance. New contexts produce collisions the existing articles do not cover. Existing articles reveal failure modes that have to be patched. A regular review surfaces what is not working and what new article would govern it. Amendments require both parties’ consent. The Skeleton can be revised, but the revision has to be bilateral, or it ceases to be a Skeleton and becomes one party’s treaty asserted against the other again.

The work is not glamorous and produces no slogans. Most of it consists of two adults stating things they had assumed were obvious and discovering that the other person had assumed something different.

VII. Diagnostic Questions

The diagnostic questions are short and specific. They are queries against the system — the kind of questions one runs against a process whose behavior is unexpected.

What “of course” was violated? Identify the unstated rule behind the reaction. The visible act matters, but the recurring heat usually comes from the clause the act breached. Name the clause in one sentence.

Was the rule ever stated? If not, you do not yet know whether the violation was deliberate. The other person may not have had the information that would have let them avoid it. This does not mean the rule is wrong. It means the rule was not yet part of the shared treaty.

Is this a disagreement, or a treaty collision? A disagreement is two parties applying the same rules and reaching different conclusions. A treaty collision is two parties applying different rules. The two require different repairs. A disagreement is repaired by argument; a treaty collision is repaired by construction of a shared article.

Is one person’s “healthy communication” functioning as the other person’s Mask? The test is asymmetric exhaustion. If one party reports the conversations as draining and the other reports them as working, suspect a unilateral protocol. The party who finds the protocol natural is unlikely to detect the cost from inside.

What article would prevent the next collision? The article should be specific enough that both parties can recognize when it is being honored and when it is being violated. “We will be more respectful” is not an article. “When one of us asks for thirty minutes alone after work, the other does not interpret it as withdrawal and does not attempt to repair it” is an article.

Was the answer integrated? A question that returns unchanged after being answered was probably not functioning as a question. A request that is acknowledged but does not change future behavior did not enter the shared treaty. A treaty collision that produces an article no one runs differently is theater. The test of communication is not whether words were exchanged, but whether the next instance of the pattern goes differently.

VIII. Every “Of Course” Is a Treaty Clause

Every “of course” is a treaty clause. Of course you should ask. Of course you should know. Of course you should validate first. Of course you should treat my reason as real. Of course you should not need to be told.

The relationship breaks when both people mistake their own “of course” for shared law. Each interprets the other’s deviation as evidence of a defect — a defect of attention, of care, of competence, of basic adulthood. Each is correct that their own treaty has been violated. Neither is correct that the violated treaty is universal.

The repair is not more speech in the abstract. The repair is constitutional: make the treaty visible, test whether both people can live inside it without masking, and amend it when reality disagrees. The mechanism is small and specific. It does not require either party to abandon what feels native. It requires both to stop enforcing what feels native automatically, and to construct a small set of agreed articles that govern the places where the two treaties would otherwise collide.

The endpoint label was always describing the result of this work. It was never describing the work itself. The phrase “open communication” pointed correctly at the end state. It just left the entire path between here and there to whoever happened to be standing in the dark.

Related essays:

• I Feel Alive When… — Wonder as a compass for whether you’re creating or consuming complexity
• The Spore Strategy — Hyper-localized syntropy when external ports are blocked
• Holistic System Rotation — The cognitive process behind seeing what domain experts miss