The Foundation That Isn't

Consider a simple experiment. Place one coin in a bowl. Add another. There are now two coins. This is true regardless of whether anyone agrees with it, believes it, or is present to observe it. Remove every human being from the universe tonight, and 1+1 will still equal 2. The fact does not wait on consensus.

Now try the same move with a moral claim. Remove every human being from the universe tonight. Is slavery still wrong?

The question produces a strange sensation — something between a category error and a vertigo. We want to say yes, obviously. But the mechanism by which the wrongness persists in an empty universe is not obvious at all. When 1+1=2 persists in an empty universe, we know what is doing the persisting: a logical relationship that holds between abstract objects, independent of minds that might contemplate it. What is the analogous thing for slavery's wrongness? What kind of entity is a moral fact, and where does it live?

The Detection Problem

The way to press this sharply is to ask what observation would distinguish two universes: one in which slavery is objectively wrong, and one in which everyone merely believes slavery is wrong. If no instrument could register the difference, no experiment could be designed to test it, then the hypothesis of objective wrongness is not doing any explanatory work. It is not describing anything additional beyond the belief.

This is not a rhetorical trick. In physics, claims earn their ontological seriousness through detection. Mass bends spacetime and we can measure that bending. Charge deflects particles and we can track the deflection. Temperature is kinetic energy and we have thermometers. The entities in these theories justify their place in our picture of the world because they produce observable effects that distinguish their presence from their absence.

Moral properties produce no such effects. A universe where murder is objectively wrong and a universe where everyone believes murder is wrong are, at the level of observable phenomena, identical. The corpse is the same corpse. The grief is the same grief. The punishment is the same punishment. Nothing in the physical or social world differs. This is not a limitation of current technology. It is a structural feature of the hypothesis: moral properties, if they exist, are causally inert.

The moral realist typically responds that mathematics is also causally inert and we still believe in mathematical truths. This is correct, and it suggests that causal inertness alone is not disqualifying. But mathematical truths are grounded in formal systems — derivable through proof, which is a method available to any sufficiently powerful reasoner. There is no analogous derivation in ethics. You cannot prove that suffering is bad from first principles the way you can prove the Pythagorean theorem. Every derivation in ethics eventually requires you to assume something moral. The theorem follows from axioms. The moral conclusion assumes a moral axiom.^[1]

The Foundation Problem

This is where moral philosophy has spent most of its effort, and the effort has been remarkably consistent in its failure. Every proposed foundation for objective morality turns out, on inspection, to require prior moral commitments.

Consider suffering. The proposal: suffering is objectively bad, and from this foundation we can construct a moral system. But why is suffering bad? Because it reduces well-being. Whose well-being? A thief who suffers less than he gains from stealing is still considered a candidate for punishment. So aggregate well-being — each person's interests counted equally. But the decision to aggregate across persons, to weight each person's well-being equally in the sum, is itself a moral commitment. It is not forced by the observation that suffering is unpleasant. It is a choice about whose interests count, and how much. The proposed foundation already contains what it was supposed to prove.

Consider rationality. The proposal: any rational agent, reasoning clearly, would converge on certain moral principles. The slave owner of the eighteenth century thought he was being quite rational — his system maximised his family's welfare, conformed to the norms of his time, and was integrated into a functioning economic order. To exclude this as genuinely rational requires adding conditions: impartiality, ignorance of one's social position, the famous veil of ignorance.^[2] But impartiality is a moral value, not a logical requirement. The demand that a rational agent reason from behind a veil is a normative demand. It is assuming that impartiality matters before we have shown why it should matter.

The Metric Problem

There is a cleaner way to see what is happening. Every moral system picks a metric — a dimension along which persons, actions, or states of affairs are evaluated. The choice of metric determines everything downstream.

If the metric is power, you get hierarchy, monarchy, and the political orders that dominated most of human history. If the metric is productivity or capital formation, you get meritocracy and the justifications for market economies. If the metric is equal moral worth, you get liberal democracy and its attendant commitments to rights, autonomy, and dignity. Each metric generates a coherent and internally consistent system. Each system can explain why its own metric is the right one. None can explain this from outside the system without borrowing from a further metric.

Modern liberal societies often proceed as if equal moral worth is self-evident — not a choice but a discovery, like the circulation of blood. It is anything but. Persons are observably unequal in nearly every measurable dimension: intelligence, strength, productivity, social contribution, reproductive success. Equal moral worth is not read off from observation. It is stipulated. The stipulation may be enormously productive — it enables the kind of cooperation and social trust that liberal institutions require. But productive stipulations and objective truths are different things.

The point is not that feudal societies were right to value hierarchy. The point is that the meta-question — which metric should we use? — cannot be answered from within any particular metric without circularity, and cannot be answered from outside all metrics without having already made a choice about what matters. The choice of metric is the moral move. Everything else is computation.

Commitment Is Not Evidence

The standard response to this kind of skepticism is to invoke the intensity of moral conviction. People who die for their beliefs, the argument goes, must be tracking something real. The abolitionist who risks his life to free enslaved people is not simply running a social calculation — he believes, sincerely, that slavery is wrong in a way that demands sacrifice.

This argument proves too much. Slave owners fought with equal ferocity to maintain slavery. Monarchists died for monarchy at exactly the moment when revolutionaries died against it. Conquerors and resistance fighters faced each other with the same conviction that they were on the right side of history. If willingness to sacrifice established the truth of a moral claim, then mutually contradictory causes would all be true simultaneously. The argument demonstrates depth of conviction. It says nothing about whether conviction corresponds to an external fact.

The evolutionary account of this intensity is more parsimonious than the realist alternative.^[3] Humans are highly social animals whose survival has always depended on group cohesion. The suite of behaviours that look like moral seriousness — the willingness to punish defectors at personal cost, the extension of loyalty to strangers, the visceral disgust at norm violations — are exactly the behaviours that improve group survival. Natural selection does not produce this intensity because there is a moral fact to track. It produces this intensity because groups whose members feel strongly about their shared norms outcompete groups whose members do not.

The hero who dies for strangers is not detecting a moral law of the universe. He is expressing an instinct that was selected for because it produced more surviving groups. This does not make his sacrifice less real, less admirable, or less worthy of the gratitude of those he saved. It does mean that the intensity of the experience is not evidence for the metaphysics of objective morality.

Enforcement Is Not Justification

The same structural point applies when the argument moves from moral philosophy to religion. Divine command theory offers what looks like an escape from the foundational regress: moral facts exist because God's commands define them. The wrongness of slavery is not a brute fact floating in a Platonic realm; it is constituted by what God has decreed. The foundation is finally external to human minds.

This solves one problem by relocating another. How do we know what God commands? Through scripture, tradition, religious authority — all of which are human institutions transmitting human interpretations. And if God's commands are good independent of the commander, then goodness is prior to God, and the divine authority becomes redundant. If they are good because God commands them, then goodness is arbitrary in the worst possible sense: God could have commanded cruelty, and cruelty would have been good.

But the deeper observation is structural rather than theological. Courts enforce laws. Armies enforce political orders. Religions enforce moral codes through promised consequences and social sanction. In all three cases, an enforcement mechanism exists that sustains a norm across time and population. The existence of the mechanism does not constitute evidence for the objective truth of the norm it enforces. Courts can punish theft without establishing that property rights are objective features of the universe. Religion can reward virtue and punish vice without establishing that the underlying distinction tracks something beyond social convention.

What enforcement mechanisms explain is not why a norm is true, but why it persists. This distinction also offers an account of religion's remarkable cross-cultural persistence that has nothing to do with its metaphysical claims. Every human society faces the same coordination problems: how to sustain cooperation among strangers, how to regulate behavior when no authority is watching, how to transmit norms across generations, how to maintain trust at scale. Religion provides an unusually elegant solution to all of these simultaneously by introducing an omniscient, ever-present observer whose authority is not bounded by physical presence or institutional reach. Unlike courts or rulers, this observer requires no enforcement apparatus at the moment of decision. Compliance monitoring is internalized through belief itself.

This makes religion not a failed attempt at cosmology but a highly efficient social technology. The belief that actions are continuously observed and ultimately judged can shape behavior even in the complete absence of human witnesses — which is precisely the case where all other coordination mechanisms fail. Whether the supernatural claims are true becomes orthogonal to their social function. The persistence of religion across human societies is then exactly what we should expect from an institution that solves one of social organization's hardest problems: not the metaphysical problem of what is good, but the engineering problem of getting large groups of people to behave as if they agreed on what is good, even when alone.

The Unemployed Hypothesis

The standard formulation of Occam's Razor is that entities should not be multiplied beyond necessity. The sharper version is this: existence and explanatory necessity are orthogonal. Dark matter might exist. That question remains genuinely open. But if modified gravity fully accounts for the observed galactic rotation curves, dark matter becomes explanatorily redundant — a posit we no longer need to invoke, regardless of its ontological status. The question of whether it exists has not been answered. The question of whether we need it has.

Applied to metaphysics, this move is devastating precisely because it sidesteps the unprovable existence question entirely. You do not need to prove that God does not exist. You do not need to prove that objective moral facts do not exist. You only need to show that everything we observe — moral behaviour, social coordination, historical change, motivational force, religious persistence, the feeling that some things are simply wrong — is fully explained by evolutionary, sociological, and power-based accounts without remainder. If the naturalistic account is complete, the metaphysical entity has no work left to do.

The metaphysical entities are not disproven. They are unemployed. And an unemployed explanatory entity is the closest thing to a refutation that epistemically honest inquiry can produce. Pure absence of evidence is weak — absence in domains where evidence should exist if the entity were doing genuine explanatory work is something else entirely. If objective moral facts were real and causally relevant, we would expect some phenomenon to be explicable only by reference to them. We have found none. This is not conclusive. But it is the shape that a successful skeptical argument takes.

The Productive Skepticism

None of this is an argument for moral indifference, or for the view that all social arrangements are equally acceptable. The skeptical conclusion about objective moral truth is compatible with strong moral commitments and vigorous opposition to practices one finds intolerable.

A vegan may believe with real conviction that eating animals is wrong without claiming that vegetarianism is a law of the universe. Someone may oppose slavery, support democratic institutions, value freedom of expression, and act consistently on those values — without believing any of these commitments correspond to objective moral facts waiting to be discovered like undiscovered elements. The intensity of a commitment is independent of its metaphysical foundation.

The strongest historical arguments against slavery were not metaphysical. They were arguments about internal contradiction — about societies that proclaimed human dignity while practicing human bondage. They were arguments about suffering, about the incompatibility of certain institutions with the social trust a liberal order requires, about what kind of society you become when you rely on coercion to sustain your economy. These arguments do not need objective moral facts to do their work. They need only the values their audiences already hold, applied consistently.

Moral progress — and there does appear to be something that functions like progress, at least in the direction of expanding circles of moral consideration — may be better understood as a social and argumentative process than as a discovery process. We do not find moral facts the way we find planets. We construct moral frameworks through conflict, persuasion, institutional design, and the slow accumulation of arguments that reveal the internal contradictions of existing systems. Abolition required abolitionists to accumulate enough influence to change institutions. That they were right to do so is something most of us believe. The grounds for that belief appear to be social and historical, not cosmological.