The Is-Ought Problem

In 1739 Hume noticed that moral arguments slid, without saying so, from what is to what ought to be. One paragraph described the world. The next prescribed. The step between was never accounted for. No description, however complete, entails an instruction.

Moore sharpened it. Define good in natural terms — pleasant, useful, survival-serving — and someone can always ask: but is that good? The question stays open. So the word was never defined.

Ethics, on this reading, floats free of the world.

Deny that moral claims say anything. Murder is wrong expresses disapproval; it is not true or false. But people argue morality as if they can be wrong, and revise when shown to be. The theory describes a grammar, not a practice.

Add a second domain. Moral facts are real, just not physical — reached by a faculty the inventory of the world does not list. That is a new metaphysics, unconnected to anything the axiom produces.

Make it feeling. Murder is wrong means murder, ugh. Part right, and it throws away the reasoning that moral life visibly does.

Declare it an error, or make it relative. All moral claims presuppose objective values, there are none, so all are false. Or: true only inside a culture. Both accept the framing and then find the second domain empty. Both are unlivable.

Each is tracking something. None has grounded ethics.

Two domains. Facts here. Values there. How do you get across?

Every response above accepts that picture. The deeper assumption is separation itself — the same assumption the last chapter found under free will.

The earlier chapters have removed it. The self is a loop, not a substance. The inside the loop reads from is one, not a private room. The I in you is the I in the other.

That is not a value laid over the facts. It is a fact about the coupling between windows.

Carry it forward and the second domain disappears. The ought was never in a realm above the facts. It is in the same coupling the is describes — read at a layer the tradition did not have.

Every coupling has a shape. At each moment the operator reads its surroundings and its own records, and among the trajectories available, couples with one. The shape of that space — what is available, what each path would produce — is the coupling’s geometry.

It has layers.

The raw geometry is what physics and habit alone give you. Causal law, prior records, the architecture’s wiring. A ball rolling downhill has raw geometry. So does a reflex. Most of what any organism does is raw-geometry coupling.

The full geometry is the raw geometry plus everything else the coupling is actually reading. For a self-aware window that includes what the self-model takes itself to be doing and to have committed to. It includes the whole-body state — exhaustion, fear, arousal. And it includes one structural fact more. The interior is one. The other windows this coupling is coupling with are local expressions of the same inside it is a local expression of.

The is the tradition named is the raw geometry.

The ought the tradition named is the full geometry, correctly read.

No gap between them. Two readings of one coupling. One incomplete. One complete.

That is the hinge.

The objection arrives at once. You have put a value into the description by calling the shared interior part of a correct reading. Who decides what counts as correct?

The shared interior is not a value. It is a structural fact, derived two chapters ago. One break, one inside. Whether a reading includes it is a question of completeness, not of virtue. A map that leaves out half the territory is incomplete. That is a fact about the map, not a judgement about the map-maker.

The consequences are normative, because the full geometry weighs the paths differently from the raw one. But the consequences follow from a fact about readings. Nothing was imported.

Every coupling carries gradients against its own dissolution. Not values it holds. What it is, read at the level of persistence. A configuration that did not hold itself against collapse would not last long enough to be a configuration.

A cell is self-preserving. A tree. A simple animal. That is not ethics. It is what makes continued coupling possible.

Now ask what its own refers to in its own dissolution.

At the raw layer: this local architecture. These cells, this nervous system. Correct, as far as it goes.

At the full layer that answer is incomplete. What the coupling is, structurally, is a local expression of the one interior. Without the interior there is no coupling left to preserve. So the same gradient, correctly specified, is a gradient against the dissolution of what the coupling expresses. That favours the continuation of other expressions of the same interior. Damage to them is damage to the structure of expression itself.

Nothing added. The same gradient, weighed with the coupling’s own structure correctly drawn.

That is the derivation. Compassion is not a value laid on the raw geometry. It is what the raw geometry’s own self-preservation favours once the shared interior is in the reading.

A reader who wants to hold the strong claim lightly has a narrower one available. Even without identifying what is preserved with the interior itself, a reading that leaves the shared interior out reads less than is there. Damage to another window registers as damage inside the structure this coupling is part of. Care follows on either reading. On the strong one, as identity. On the narrow one, as continuity.

The pull you felt at the start of this chapter is not imported. It is what the full geometry of a self-aware coupling weighs.

The tradition thought the weight came from outside the facts. It came from reading all of them.

No fallacy, because no step from is to ought was taken. A descriptive premise — the interior is shared. A descriptive fact — the full reading weighs differently from the raw one. A descriptive result — the full reading favours care. What the tradition called the ought is what correctly read geometry favours.

There was no gap. Only a reading at the wrong layer.

But reading correctly is not automatic. The shared interior is present in every coupling between windows. What varies is whether the self-model’s scope includes the other windows. A self-model that holds only its own group as fully real reads the geometry with the interior excluded for everyone outside.

Reading correctly means the scope set to all windows.

One limit, stated plainly. This chapter does not derive that every coupling ought to widen its scope. It derives that, for a coupling whose scope already includes other windows, compassion follows from the reading. Why the scope widens is a question about cognition and history, not a premise.

Widening it is an achievement, and most of history was lived without it. What the species calls moral progress is, at this layer, the widening of the scope. The enslaved recognised as windows. Women recognised as windows. The current widening runs across species. These were also political and economic fights against real interests. The structural layer is what made the widening sayable and defensible at all — once the interior is seen, the narrower reading cannot be read as complete again.

The reading is the work.

Two more words, so the next chapter can use them.

Sub-optimal execution: the coupling commits to a path that is not the strongest-weighted on the correctly read geometry. It could have taken what the reading favoured most. It took something else, and paid.

Override-capacity: the structural capacity to do that. The self-model holds commitments that can weight the coupling towards a path the reading alone would not select. Simpler organisms do not have it. Self-aware windows do.

Now the distinction.

Compassion is computation on correctly read geometry. Care is the strongest path and the coupling takes it. No override needed.

Love is sub-optimal execution under the self-model’s weights. This one, at this cost, against the architecture’s own preservation weighting. Because they are mine. Because I committed. The generic care the reading favours may not survive the worst a particular relation can produce. The raw weighting says withdraw. Love is the self-model holding the coupling to this window across that pressure.

Both are ethics. Both are structural. They differ in which layer is doing the work.

The hardest case. Two windows. One resource. It cannot sustain both.

First a caution. Nearly every justification for serious harm in history has claimed zero-sum. Whether a situation is in fact zero-sum is empirical work — economics, ecology, medicine, politics — and most situations narrated as zero-sum have paths the narrator’s frame left out. A reader who hears this section as a licence has read it wrong.

Where it is genuine: zero-sum is not a failure of the reading. It is a condition the reading reads correctly. At the narrator layer it is tragic, and the tragedy is real — one loop ends. At the operator layer the ending window’s records propagate into the environment, into other windows, into whatever couples with their trace. The interior both were expressions of is undiminished.

What separates that from parasitism is whether the condition is real. Parasitism is taking when sharing was available. A coupling that narrates a non-zero-sum situation as zero-sum, to license extraction, is reading incompletely — in a different way from the reader who leaves the interior out, and just as incompletely.

Genuine zero-sum, correctly read, is not an ethical failure of the window that continued. It is the situation’s own shape, executed. Not harmonious. That word dismisses what the living carry.

Hume was right at his layer. Raw geometry produces no instruction. What he could not see was the layer above it.

Moore was right at his. The raw reading always leaves the question open. It closes when the reading is complete.

The emotivists were right that moral claims carry feeling. Feeling is one input to the full reading, not the whole of it.

The error theorists were right that there is no realm of values above the world. They missed that the absence of a second realm is not the absence of ground.

The is was never separated from the ought.

There was one coupling.

The reading is the work.