What Is Ambivalence

The ordinary account of ambivalence is a subtraction. Two forces, opposite signs, one line. The job at plus seven, the old life at minus seven, the sum at zero, and zero is why you cannot move. Everything that follows from that account is familiar, because it is what people say: break the tie, add weight to one side, and the sum will stop being zero. Decisiveness is the virtue and the tie is the disease.

The picture is so natural that the word has half-collapsed under it. People now say they are ambivalent when they mean they do not much care — which is exactly what the line predicts they should mean. On a line, (+7) + (−7) and 0 + 0 are the same number. A model in which full-strength wanting and total indifference are indistinguishable is not a subtle model. It is a broken one, and we have been using it to give each other advice.

What follows

1. Two ordinary cases the line cannot hold
2. What the foundation puts there instead
3. The arithmetic of being torn
4. Which axis your conflict is on
5. Why you do not turn the dial
6. What it costs, and what this cannot do

Two ordinary cases the line cannot hold

The resolution that does not resolve

On the subtraction account, the difficulty is the tie. Break it and the difficulty ends. But the most common report from people who decided cleanly, decided early, and were right, is that the pull continued — for years, at close to full strength, with no accompanying doubt about the decision. A tie that has been broken should not still weigh what a tie weighs. Something is still there, and the line has nowhere to put it.

The conflict with no line on it

You want to spend a year abroad and you want to plant an orchard. Name the line that contains both. There isn’t one. You can build one — convert both into money, or into a decade of your life, and now they trade against each other — but you built it, and it will report that they are opposites because you constructed it so they would be. Before the conversion there was no axis on which more orchard was less travel. The opposition was an accounting device, imported and then mistaken for a finding.

The line is not a description of the conflict. It is a tool for forcing a decision, and it does that by first destroying the information that made the decision hard.

What the foundation puts there instead

Begin at the shared core. S1: when the unconditioned is conditioned, a circle appears, and every point on that circumference is a condition. Not a list, not a ranking, not a line — a circumference, which is infinite, and which no enumeration closes. Two conditions therefore do not stand in a relation of more and less. They stand at an angle.

S2 says the plane of that circle is ℂ, and says it for reasons rather than for convenience: ℂ is the smallest algebraically closed extension of ℝ, and a phase requires two real dimensions to exist at all. One real dimension gives you sign. Two give you direction. Everything in this piece turns on the second thing being available, which is why the plane cannot be traded for a line without losing the phenomenon.

A word on the physics, stated carefully, because the careless version is everywhere. Renou and colleagues showed in 2021 that real-valued quantum mechanics differs from the standard formalism in multipartite settings, and experiments followed in 2022. Since 2025 that reading has been contested: real-valued reconstructions exist that reproduce all the predictions under a different composition rule for independent sources. What survives the dispute is the useful part — complex arithmetic is not eliminable from the description, only relabelled. It should never be written as a flat experimental falsification of real numbers, and it is reported here, not leaned on.

Then P6 and P9, which are what keep this from becoming mysticism. A condition is a prerequisite in order for something to happen or exist. It does not cause, does not make possible, and does not happen. A condition conditions — a real relation, and not agency. So nothing out there is pulling at you. There is no tug-of-war and no rope. By P4 the load sits in the denominator, which holds prediction and ideas together, and by R8 what a person experiences is never the numerator or the denominator but the quotient. What you feel at the sink is R. The two conditions are not in the room.

Last, the measurement. The spine already fixes what alignment between an artifact and an exemplar means: fidelity is cos(φ_artifact − φ_idea). That is not new here. What is new is running it against two exemplars at once.

The arithmetic of being torn

Put two named conditions on the circumference, separated by an angle θ. Measure everything from the point halfway between them, so one sits at −θ/2 and the other at +θ/2. Now take any artifact whose phase is α from that midpoint. Its fidelity to the first is cos(α + θ/2); to the second, cos(α − θ/2).

Add them and the sum is 2 cos(α) cos(θ/2), which is largest at α = 0. Ask instead for the best worst case — maximise whichever of the two fidelities is smaller, which is what a person torn between two things is usually trying to do — and the answer is the same position, because the smaller of the pair is largest where the pair is equal. Two different objectives, one answer:

The best fidelity jointly available to two conditions separated by θ is cos(θ/2) to each. It depends on the angle and on nothing else. Not on effort, not on sequencing, not on how much you want either one.

Figure 01

Three angles, three completely different situations

• θ = 0°
• Fidelity 1.00 each
• Not a conflict. One condition named twice, and the argument you are having is with your vocabulary.

• θ = 90°
• Fidelity 0.71 each
• A real cost and no betrayal. Roughly seven-tenths of each is the ceiling, and it is a ceiling, not a failure of nerve.

• θ = 180°
• Fidelity 0.00 each
• Antipodal. Any positive fidelity to one is an exactly equal negative fidelity to the other. There is no clever middle.

The antipodal case deserves its exact statement, because it is the one people spend years trying to escape. At θ = π, the two fidelities are cos(α + π/2) = −sin α and cos(α − π/2) = +sin α. They are negatives of each other at every position. Not approximately, not usually — identically, everywhere on the circle. Whatever you gain in one you lose in the other at exactly the same rate, and at the midpoint you have nothing of either.

Figure 02

What the angle costs

θ = 0°1.00

θ = 30°0.97

θ = 60°0.87

θ = 90°0.71

θ = 120°0.50

θ = 150°0.26

θ = 180°0.00

• Near enough to serve together
• A priced conflict
• Approaching antipodal

Notice what the curve does at the top. Small angles are almost free: two conditions thirty degrees apart can both be served at ninety-seven per cent, which is why a person with a well-arranged life can hold a dozen commitments and feel no strain at all. The curve is nearly flat there and then it falls off a cliff. Most of the cost of a conflict arrives in the last sixty degrees.

• Marked construction 1 · joint fidelity treated as a single scalar objective
• Marked construction 2 · the midpoint result cos(θ/2)
• Marked construction 3 · θ between two named conditions taken as well-defined

Those three are new, and they are marked rather than asserted. Construction 1 is the weakest of them: it assumes the two fidelities can be traded against each other as numbers. The one piece of comfort is that the two most natural versions of that assumption — add them, or protect the smaller — give the same answer, which is a mild sign that the result is about the geometry and not about the objective. Construction 3 is the one I would bet against, and I say why at the end.

Which axis your conflict is on

The angle is not a mood. S3 gives the compass two orthogonal axes and therefore four poles, and a conflict sits somewhere definite on it.

Which gives a diagnostic you can run in about a minute. Does an answer to one of your two conditions change what would count as an answer to the other? If it does, they are interrogating the same question and they sit on the same axis — near-antipodal, expensive, and possibly unsolvable. If it does not, they are on different axes, and the geometry says you are entitled to about seven-tenths of each and no more.

The case that matters most to anyone who makes things: you want your work judged fairly against the standard, and you want it to be unmistakably yours. The first is an ordering question and lives on the real axis. The second is a distinguishability question and lives on the imaginary one. On this compass they are not opposites at all. But a career spent feeling torn between them has usually been spent on the assumption that one had to be sacrificed, when what the geometry says is that both survive at roughly seventy per cent, permanently, and that the seventy per cent is the ceiling rather than a consolation prize.

Two cautions, both of which are open problems in the foundation and neither of which this piece closes. First, the orthogonality of the two axes is supported by the observation that all four quadrants populate, and that is evidence rather than argument. Closing it would mean showing from the primitives that an ordering question cannot be re-expressed as a distinguishability question — that no rotation carries one axis onto the other — rather than noticing that examples land in all four corners. Until that is done, the 0.71 is exact given orthogonality and merely indicative without it. Second, fairness being the antipode of hierarchy rather than a neighbour of it is assumed here and is itself unsettled, since meritocratic fairness produces rank. Every antipodal reading below inherits that assumption.

Why you do not turn the dial

It would be easy to read the last two sections as instructions: find the midpoint, sit there, collect your cos(θ/2). That reading is the ownership error in new clothes, and the spine forecloses it. Phase is relational. It is fixed by which idea has the actualizer — not by the mark, and not by the maker. Nobody sets their own φ by deciding to.

What does happen is that the relation rotates over time, which is the trajectory of a life around the circumference. So the arithmetic is a map of what a position is worth, not a control you operate. It tells you what is available where you are standing. It does not hand you a way to stand somewhere else.

That is not fatalism, and the distinction matters. P5 holds that the host can occupy a position from which its response to a readable condition is revisable, and R13 puts conscious responsibility at exactly that site. Agency is relocated by this framework, not abolished. What is gone is the fantasy of choosing your angle. What remains is everything that happens after you can read it.

What it costs

Three unwelcome things follow, and softening any of them would make the piece useless.

The twenty-nine per cent is not recoverable. At right angles, 0.71 is the ceiling, and the arrangements, schedules and integrations on offer all promise to beat it. None of them can. What they can do is discover that θ was smaller than it looked — which is a genuine finding and worth having — but a claim to have obtained full fidelity to two orthogonal conditions is a claim to have found cos(45°) = 1.

At π there is no synthesis. People do report synthesising antipodal conflicts, and what usually happened is a rotation plus a rename. The spine has a word for it: corruption is phase drift, and at the antipode the actualizer serving himself produces fairness-artifacts and calls them hierarchy. The tell is that the description of the conflict changes while the conflict does not. If your account of what you were torn between quietly improved after you resolved it, look again.

And it was capped before the angle got involved. By R10, every artifact is a miss; the denominator was never ours to command. The cos(θ/2) does not introduce the shortfall, it discounts a shortfall that was already there. Which is why R11 is the only evaluative question that survives contact with any of this: fidelity, not success. Not did you get both, but what kind of miss did you make.

One more, aimed at this piece. By R14, every idea is biased toward itself, and this one is no exception. Hand somebody an angle and every difficulty in their life starts presenting as an angle. The specific failure to watch for is the flattering one: a diagnostic that keeps telling you your conflicts are orthogonal and therefore nobody’s fault is not a diagnostic, it is a comfort. Antipodal conflicts are real, and nothing in the foundation suggests they are rare.

What this cannot do

1. Knowing the angle does not reduce the pullReading a mechanism has never yet stopped one. The value here is that a correctly identified orthogonal conflict stops generating the second layer — the private verdict that being torn means you are weak, indecisive, or lying to someone. That layer can go. The pull stays.
2. Nothing here says the distress is a reasoning errorThe line is a bad model. The feeling is not the model. A person at an antipodal conflict is in an actual bind with an actual loss in it, and the geometry does not make the loss smaller — it only stops you searching for the door that was never in the wall.
3. There is no scheduleNaming a conflict antipodal does not start a clock on being over it. Grief for a road not taken is not shortened by having correctly classified the intersection.
4. If it is looping, this is the wrong toolSame evening, same two options, no new information for months — that is a mind grinding, and a metaphysics cannot treat it. It can take shame off a mechanism and it should not be offered as though it could do more. Talk to someone who works with people, not with circles.

What the geometry gives

A single number, cos(θ/2), that survives two different ways of asking the question, and a one-minute test that separates a conflict with a price from a conflict without a solution. It also explains why deciding does not end the pull: the decision changed the artifact, and the artifact was never what the angle was between.

What it does not give

No method for choosing φ, which is relational and not yours. No claim that θ is measurable in any instrument sense. No licence to demote a real antipodal conflict to a misunderstanding, which is the most likely way this gets misused.

Where I am probably wrong

Construction 3. If naming is what makes a condition readable as an idea, then θ is partly a property of the naming, and two people torn over what looks like the same thing may be standing at different angles — with the arithmetic still exact and every number reporting on a namer rather than measuring a world. That would leave the shape intact and make every quantity in this piece a self-report. I do not have a way to rule it out, and I would want it ruled out before anyone treated 0.71 as a fact about their marriage.