Notes on the Mechanism

These notes are not a glossary. Each one explains what a word is carrying — which mechanism it locks — and why it can't be swapped for the smoother near-synonym. They run in the order the essay descends, layer by layer, from the outermost illusion down to the ground.

In the essay itself, every word answers to one thing: precision about the mechanism. Nothing is softened for ease, nothing stops mid-sentence to explain itself. If you read it and a word snags, this is where the key is.

The test for any word: is it carrying a mechanism, or an atmosphere? Words carrying a mechanism answer only to precision. Words carrying atmosphere are where style and how-it-reads come in — and the valuable part of this text is the mechanism, not the atmosphere.

One principle runs under all of it. The essay never states it; here it is, as the key to these notes: don't infer motives when structure is observable. Every test in the essay obeys it — the observable (behavior, the shutting door, the echo, whether they pour in under hardship) is read directly; the unobservable, where a person's center sits, is only inferred from behavior, and always declared as inference, carrying the room to be wrong. That is why it watches the clock, the echo, the shutting door — and never "what they're really thinking."

The Frame

mirror — The other person is the medium that shows you yourself, not the object under analysis. The essay's premise: reading the other and reading yourself is one act, not two.

running on — What your choices run on underneath — a forward reference to the deepest line, where belief becomes ground. It's the closing word of the opening, and it lets the whole chain close at the end: relationship → event → belief → ground → machine. Not what you believe in or what you value — those are beliefs you can see and name; running on points at the base you can't see and take for reality.

The Early Illusion

intensity / direction — Scalar versus vector. Intensity is temperature, magnitude, how hot. Direction is where it points. Intensity tells you the size, not the heading. This pair is the axis of the whole essay.

not data about direction — The early signal is data — it tells you intensity; it just isn't data about direction. Not "not data" at all, which would contradict the next line. A thermometer produces data too; it just can't measure heading.

baseline / resting state — The default steady value a system returns to once the driving force is removed. It stands against the peak, the momentary response driven up by the chase. Not true self or real personality — those are static essence; the baseline is a reading taken after the force is gone, a current value, later rewritten by events.

pull — The externally applied driving force, courtship's traction on a person. It drives them off baseline up to the peak; remove it and they fall back. Not attraction or chemistry or feelings — those are internal states; pull is the external input. Swap it and you lose "remove the force and it falls back."

speed → roots — Formation speed is inversely proportional to structural depth. Fast means shallow, because depth takes time to grow, and deep things form slowly and come apart slowly too. Not a value judgment, not "fast equals bad" — fast means no deep structure yet, and so fragile. Roots names whether there is deep structure, not decoration.

The Center

center / converge / hub — The mechanism is convergence. Everything a person does in a relationship — the spokes — converges on one point, the hub, and that point's location decides what the behavior means. This is the load-bearing idea of the section, and the root of "you can't tell which is which": the same spokes can converge on different hubs. Not what they're after or their goal or their desire — those keep only the heading and lose "many behaviors converging on one point." Not core or true self or essence — that is a fixed internal core, static; this center is a moving point of convergence, grown from gaps and expectations, rewritten by events. Converge and hub make the geometry explicit, so the center stands on geometry, a meeting point, not psychology, a core.

source — Replaceability. When the center converges on what they get, the other is the supply's source — and a source is defined by "whoever supplies the same thing can take this position." Not object or target — those name the person pointed at, and lose "interchangeable with anyone who supplies the same."

a part in a structure — Indirectness. The other is a component, wanted through "the structure still holds," not wanted directly. Not partner or the relationship itself — part carries "you are a component, your value is in keeping the structure, and what gets protected when threatened is the structure, not you."

not seeing, inferring — The honesty marker, and the whole method's anchor. The center can't be observed; every judgment is inferred backward from behavior, always carrying the room to be wrong. Not seeing through or reading — those disguise error-prone inference as certain sight, and the essay's honesty rests on "I can only infer, and may be wrong."

The Refusal Test

the door shuts / staying open — An observable, binary behavior, used to check the unobservable center. The key: the open pole is staying open, staying in — behavior, they didn't cut off — not understanding, an internal state, whether they "get it." The power of the section is "no mind-reading": you read whether the door shut, which is visible, not whether they truly understood you, which is not. Making understanding the binary pole would smuggle the observable signal back into unobservable guesswork.

three states: open / bolted for good / keeps shutting — Shutting is not binary but three states, each pointing somewhere different. Bolted for good — single, total, irreversible — is a clean signal: the default is to cut completely; it hurts, but it's clear. Keeps shutting — closes, leaves a crack, closes again — means the door isn't dead, because someone fully done doesn't keep closing, once is enough; so it can be opened, but the cost of opening it is you. This state connects directly to "Change Is Possible, but Expensive." Don't collapse it back to binary — that loses the most draining, most easily-misread-as-hope middle state.

How the Machine Turns

a product, not a sum / multiply / zero / factor — An event is two states multiplied; if one is zero, the whole is zero, and your input does not remain. This is the mechanism's body. Not "it takes both," and not a sum: "takes both" doesn't say what happens to your input when one side is zero, and a sum (0+5=5) leaves your input standing, while the truth is it goes to zero (0×5=0) — the mathematical root of "holding on is useless." Use a sum and the whole conclusion collapses. Zero is a state description, one side gives nothing, not an operator. The math stays not because it's cold but because the smoother word deforms the mechanism. (The lens line at the section's start already says: these aren't numbers to plug in, they're description; what you verify is your own relationship, not the math.)

rewrite / overwrite — The event's action on a state is irreversible overwrite, leaving a mark. Not influence or affect or shape — influence can come and go and cancel out, reversible; a rewrite leaves a mark you can't get back behind, so that after the collision you are no longer the you from before. The irreversibility is the mechanism.

product side / maker side — The determinism brake. You are both the product of every past event — made, passive — and the maker of the next, since your choice now is a new event — active. The thing made and the thing that makes are the same you. This is the root of the essay's exit: it blocks "everything is determined by events, so people are powerless." Seeing is useful precisely because seeing is itself a new event that can rewrite what's downstream.

congeal — An experience, at some point, going from fluid to fixed on its own, unnoticed, the way a liquid sets solid at a temperature. Not form or develop or decide — those are too generic and lose the unnoticed phase-change; decide is wrong, since you don't decide to hold a belief, it sets on its own. Congeal locks "the belief's making is a passive phase-change, not a choice."

sink into ground / floor — A congealed belief sinks to the bottom and becomes the floor you stand on but never look down at. Three things at once: it sinks, it becomes the base, it goes unexamined. It writes out the deception: the belief hides as reality, and you can't question what you don't know you hold. Don't replace it with a flat abstraction like hidden assumption — the image, ground, carries all three; the abstraction carries one. Runs through to the last two sections.

installed — A belief can — without your passing through any event — be put straight into your head, then sink and become you, while you take it for original. Not taught or imposed or influenced — taught is too gentle, a teacher-student relation; imposed too adversarial, you'd know; installed carries "put into the system, you don't know, taken as native." And it has to get in through a wanting you already have, which is why it's silent. This is what "thinking for yourself" precisely means here.

lifted back / knocked over / loosens — The one place this machine can be reached. Ground is recognized as only a belief — lifted back — and once it's a view again it can be knocked over; knock one over and the connected ones downstream loosen, not fallen, but no longer firm. This is the physical process of what "seeing" means at the end. Loosen isn't change or break — it precisely means "loosened, not toppled."

Two Machines Meet

multiply two centers — This section is the two-person extension of "an event is two states multiplied," not a new taxonomy. Multiply two centers and the outcomes that grow are only a few — derived by the product logic, not an arbitrary list of relationship types.

built-in ending / take root — For supply against supply: replaceable, and passing through no difficulty, so the ending is built in, not accidental. Not "they didn't love enough" or "didn't try enough," a moral or effort attribution — it's a structural necessity, because the replaceable can't take root.

asymmetric grief / the bill isn't split evenly — Replaceability makes the depth of investment asymmetric, so the pain at the break differs by worlds. The deep one wanted this specific person, a deep investment and an irreplaceable loss; the shallow one's want could be connected elsewhere, a shallow investment, a replaceable loss, reconnected before long. This is the emotional consequence of "replaceable," the emotional version of "holding on is useless." No one did anything wrong, but the bill isn't split evenly — no blame, and the unfairness is structural, not anyone's fault. It answers the recurring "was it my fault or theirs": no answer; the thing with an answer is "two machines, not turning the same thing."

When Hardship Comes

more factors / can't be exited — Going deep means more factors in the product, some of which can't be exited, which fundamentally changes the machine. Early it's A×B; deep it's A×B×child×responsibility×…, and a child is a living, growing, back-rewriting factor that can't be exited. Early, one factor goes to zero and both walk away, a clean stop; deep, one goes to zero, but because of the factors that can't be exited, the event doesn't stop cleanly — it becomes an unleavable, ongoing drain. This is why a deep-stage mismatch is crueler than an early one.

reverses the direction / pour in — Hardship reverses the relationship's direction from two-way to out-only, which is why it tests the center. Ordinary times are two-way, give and take both present, and you can't easily see whose center is where; hardship is one-way, only pouring in, no return for a long stretch, and "wants to take" and "wants to give" finally split. Hardship isn't "a test of how deep the feeling is," which is vague; it's "reversing the direction to force the center out," which is precise.

three replies with results / refutable — These three outcomes are checkable, refutable predictions, not descriptions of attitude. A claim that can't be refuted is worth nothing — the framework dares to predict that a supply-center withdraws in pour-in, no-return hardship, and if reality refutes it, if they stay, it doesn't dodge: it gives two further-checkable possibilities, that the center was read wrong, or that it was rewritten. This is the essay's direct answer to "is this an unfalsifiable, all-absorbing narrative": giving a refutable prediction at the hardest point is the proof that it isn't.

rewrote their center — The determinism brake again, connecting forward. A supply-center is predicted to withdraw, but if they stay, one possibility is that this hardship rewrote their center — hardship is the kind of high-intensity event that can rewrite a state. It holds the line against fatalism, since people aren't permanently pinned by their center and hardship can rewrite, without going cheap, since rewriting is real and observable, and is the expensive force of the next section.

False Solutions

load — The thing that has to be absorbed — a quantity that can be carried, transferred, and conserved. Not deficit or debt or burden — deficit is accounting; debt implies repayment, but moving isn't repaying, it's shifting onto someone else; burden is emotional. Load is the neutral, transferable, conserved engineering quantity — it has to support "conserved," "moves to the next node," "spends the one it's moved to."

absorbing vs transferring — The only source of win-win is "someone genuinely absorbed the load" — passed through it, grew new capacity; moving merely transfers, doesn't absorb. The two are mutually exclusive — the load-bearing point of the section's first half.

conserved / moving / weakest node — The positive answer to "how could consuming others ever be win-win": moving doesn't remove the load, only changes who carries it and when it comes due; the total is conserved; each move spends the one it's moved to; and it finally lands on the weakest, least-able-to-pass-it-on node. A system kept alive by moving can never close or be win-win — its premise is "don't absorb, only transfer," and win-win's premise is "someone absorbed it." Checkable: follow such a system and see who the load lands on. Not "selfish" or "irresponsible," which is moral — it's a consequence of conservation. "The most silent, the one least likely to argue, carries the final reckoning" is a line with cold weight, but a mechanism statement — the load lands where there's least bargaining power — not an accusation.

swapping / runs both ways — Swapping isn't one-directional; it's two models passing the buck, neither absorbing its load. Direction one: a relationship load — is their center on you, do they hold up — translated into money, "with enough money none of this matters." Direction two, closer to the spine: money is a problem of production — whether you can make, earn, produce value with the world, a problem that's yours, solved only by growing a capacity, slow and costly. The swap is to not solve it — to translate "I can't produce enough" into "I need someone who can supply it," turning the other into a source, the supply position again, covering your production gap with the relationship. Both directions park a load belonging to one model in the other and cover it. Connects to source and the swappable supply position.

money as a problem of production — Money's load is, at bottom, whether you can produce, create, exchange value with the world — a problem that is yours, to be solved in the production model by growing a capacity. Not the shallow read, "not enough money, so go get money." Precisely: money is the expression of productive capacity, and the money problem is the capacity problem.

the snobbery brake — Money solves money problems — a medical bill is a medical bill, a genuine solution; swapping is specifically "using one model to cover another model's load." The dividing line: after the money is spent, is that load actually gone, absorbed in its own model, or just covered, parked, out of sight, due later. Don't dismiss all use of real means, including money, as avoidance — that's its own snobbery; only "covering model B's load with model A" is swapping. (On the gender point: the text never names a gender or uses any accusatory term — it stays at the structural "supply position." The resentment, if it goes anywhere, goes into the mirror, not into an accusation of any group.)

several models running — People rarely live one unified, coherent life; they run several models at once in the same body — relationship, economic, answering-to-oneself — translating, moving load, sometimes fighting, while taking it all for one thing. "Running" connects to running on and machine. One layer of seeing is recognizing which model you're running, whether what it solves is the problem you take it to solve, or whether it's only, decently, busily, skirting the core you never dared open.

Holding On, and the Cost of Change

supplying more / cementing / runs the wrong way — Holding on is, at bottom, supplying more — and supply is the fuel a supply- or structure-center runs on. So holding on isn't useless; it's worse — the wrong direction. The more you supply, the more firmly you cement them into the position they were already in, confirming that "taking from you" is right and sustainable. Cementing, not reinforcing, locks the irreversible fixing-in-place — you fix them where they are with your own hands.

factor is zero / multiplies to zero — The mathematical body of "holding on is useless": their factor is zero now, their center isn't on you, and however large you raise your own, the product stays zero. What you're pushing is a thing already at zero. This is the cashing-out of "a product, not a sum": with a sum, your effort would partly remain; with multiplication, it goes to zero.

ground translates "refused" into "I'm not enough" — The deepest reason holding on can't stop: under the feet is a sunk belief, "as long as I'm good enough, they'll stay," that automatically translates every "refused" into "I'm still not enough," so the more they're refused, the harder they hold — the loop's root is the unrecognized ground. The only way out is to recognize that ground as only a belief, and lift it back.

the conclusion to guard against — "Since it can't be changed, see it and go" is the essay's most dangerous conclusion, because it's too convenient — it can package "pulled back too early" as "saw too clearly." This section exists for one reason: to block that conclusion and put a price on leaving. No blade takes sides — giving too much and pulling back too much can both be you.

undiscounted cost / certain for uncertain / self-cementing — The cost of change has to be stated honestly and undiscounted: you stake the certain — your time, your state, possibly a whole life — for the uncertain — whether they'll truly be rewritten, with no guarantee. You're being the continuous external force against their self-cementing direction; since self-cementing is the default and change is uphill, it's expensive. Don't write change as "love is enough" or "patience is enough," which is cheap hope — it's certain cost for uncertain result, and that asymmetry must be on the table.

two accounts, eyes open — Going and staying are both no longer easy options; they are two accounts, each with a cost, both to be reckoned with eyes open. The symmetric load-bearing: leaving isn't the default-clever move, it has a cost, and staying and changing isn't a fantasy, it has a cost. Each side has one lovely word that lets you skip the reckoning — "they weren't worth it," "I have patience" — and this section exists to expose both.

The Ground

the whole continuous floor — Gathering the scattered sunk beliefs into one continuous floor — and that whole sheet is who the person is right now. The base of the tower: not a few scattered blocks, but a continuous floor, which together is them.

choices by the floor, not by right and wrong — The coldest line: people don't choose by right and wrong; they choose by the floor under their feet — a choice consistent with the floor feels right, and they don't regret it. So "no regret afterward" isn't "it genuinely wasn't wrong," it's "it stepped on their own floor." This takes "why people don't regret" from firmness or will, a wrong attribution, to "the floor filters out the regret signal," the mechanism.

a wrong floor raises no error — The core of the base, the coldest cut. The floor itself can be wrong; and a wrong floor, having filtered out regret, has no self-correcting signal, no brake. Someone on the right path stops over the occasional pain to look down; someone on a wrong floor, precisely because every step feels right, never looks down to check — so it carries them, the whole way, into harder ground while they stay certain. "Self-cementing" is the surface of this; the wrong floor that raises no error is the real mechanism underneath. (This is how the section carries, without ever using the term, the idea of an optimization target that's wrong and never self-corrects.)

outside force / heave the floor up — The only thing that interrupts a no-error floor is outside force — and that force isn't more supply, since supply is poured along the floor and only thickens it, but "what you wanted, you didn't get, and the other didn't give way," or a hardship that won't pass unless it's passed through. The one heaved by that strike, who bends to look at their feet, grows outward; the one who, struck, presses or withdraws, tamps the floor firmer. Points to the last section's look down.

After You See It

the loop / the loop's product — You think seeing puts you outside, making a clear-eyed decision; in fact you're always inside a turning loop — a feeling rises, a thought congeals, a choice, the other's reaction, fed back into feeling — and those after-the-fact "I thought it through, so I did this" are products generated inside the loop, not judgments from outside it. Rationalization isn't a label stuck on at the end; it's a product throughout.

seeing = knowing you're inside, not getting outside — The closing's cashing-out of "understanding, not solving": no one can stand outside the loop; seeing isn't transcending or controlling or escaping, it's knowing you're inside. This refuses "so what do I do now" — refuses to hand over a method for operating from outside the loop.

screen / remake / pull back — Three explicitly refused actions, three kinds of fixing. The essay doesn't teach screening people, remaking anyone, or pulling a relationship back — it only lets you see how the machine turns. Anything that teaches you what to do to change the outcome, the three fixes, isn't in this essay; this essay only gives seeing.

slacken — After seeing, the force that's been dragging you slackens. Connects to loosen, the same slack — a relationship without investment loosens, and seeing lets the force dragging you slacken. Seeing changes neither the other nor what's happened; it changes whether you're still being dragged.

deepest way out = looking down — Going and staying are still choices made on your own floor; the deepest way out isn't which one you choose, but looking down at the floor itself. The concrete act: recognize what you take as "just how things are" as only a belief, congealed or installed, and tell apart how much of the floor you walked in yourself and how much was poured by someone else. The abstract "seeing" lands on a concrete, doable act — not "examine yourself," but "look down, and lift the floor back into a belief."

"the moment you see it — it has already stopped being ground" — The closing, and the cashing-out of the lever. Seeing is recognizing it as only a belief, which is it turning from ground back into a view, which is already loosened. The act of seeing completes the first step of lifting it back, so it has "already stopped being ground" — not a sentimental aphorism, but the mechanism cashed out: to see is to have made the first move. It ends on the cold side with a trace of warmth, the instant of seeing-as-changing, and doesn't turn back to point at whoever wrote it — the close stays under the reader's own feet.

The standard, in one line: precision is not keeping the original word, precision is keeping the original mechanism. If a word turns out to be inaccurate, it should be changed — including to a more mathematical word, if that's more accurate. The one thing that can never be a reason to change a word is "will the reader be scared off" — that is no longer editing the mechanism; it's editing the reader.