The Near Misses

Parfit, Rand, and objectivity in the wrong place

Derek Parfit spent the last decades of his life on a single wager: that the great moral theories of the modern era — Kantian duty, consequentialist utility, contractualist fairness — were not rivals at all but climbers on the same mountain, converging on a single summit of objective moral truth. On What Matters¹ is the record of that wager, a heroic attempt to rescue moral realism from the wreckage of modern philosophy by welding the three traditions into one convergent moral law: the Triple Theory. It is a brilliant work of philosophical engineering, and it fails — not at the edges, where critics usually probe, but at the foundation, in a way worth understanding precisely.

I rejected four proposed foundations for agent-independent value — God, reason, evolution, intuition — in the myth of objective value. But subjectivism has two rivals that deserve more than a paragraph, because each gets something importantly right before going wrong: Parfit’s convergence realism, and Rand’s Objectivism. I call them the near misses. Each identifies constraints or agreements that survive scrutiny, then treats them as sufficient for moral authority. The correction is not that propositions cannot be objective. It is that public procedure can make a conditional application transparent without making its normative premise agent-independent.

The Four Tells

Start with how Parfit argues, because the method puts pressure on the project before the metaphysics does. On What Matters treats moral realism and convergence as commitments to defend, and four features limit how much independent evidence the defense supplies. Here tell names an argumentative pressure point, not evidence of a hidden motive.

First tell: the undesirability of a view bears argumentative weight against it. Parfit argues for irreducible normative truths — mind-independent facts about what we have reason to do — while also stressing that anti-realism would make morality “less important than it really is.” That consequence may explain what is at stake, but it cannot establish the ontology. The universe is not obliged to be arranged so that morality stays inspiring.

Second tell: the convergence is constructed from revised theories. The Triple Theory synthesizes Kantian, rule-consequentialist, and Scanlonian formulations, and Parfit presents their agreement as confirmation of realism — three expeditions arriving at the same summit. But selecting nearby theories and revising their formulations until they converge limits the independence of that evidence. The synthesis may reveal deep unity among the reconstructed views; it cannot count as three untouched theories independently detecting one moral realm.

Third tell: counterexamples are often absorbed by revision. Where the Triple Theory’s components conflict — in coercive-sacrifice dilemmas, lifeboat scenarios, or distributive cases — narrowing “reasonable rejection” or revising what counts as “making things go best” can restore alignment. Such revision may improve a theory, but it also makes convergence less discriminating unless the revisions face independent constraints.

Fourth tell: motivational utility remains distinct from truth. Parfit’s concern that rejecting realism could weaken motivation is philosophically important, but usefulness is not truth. A belief can motivate while being false, and a true belief can fail to inspire. The practical cost of anti-realism cannot carry the evidential burden for realism.

Four tells, one narrower diagnosis: convergence inside Parfit’s reconstructed theories does not independently establish the realism used to interpret it. The project may offer arguments for realism elsewhere, but convergence cannot serve as its own confirmation when selection and revision help produce the convergence. A robust moral framework should also survive the falsity of realism. Mine aims to do so by grounding normativity in explicit, conditional value-commitments.

Coherence Is Not Ontology

Set the method aside and grant Parfit everything he wanted: suppose the three traditions really do converge. What would that prove? Here the deeper error comes into view, and it is a category error: Parfit mistakes coherence for ontology.

Look at what each component of the Triple Theory actually does. Kantian ethics optimizes internal consistency — could you will your maxim as universal law? Contractualism optimizes mutual justifiability — could anyone reasonably reject the principle? Rule consequentialism optimizes aggregate outcomes — which rules make things go best? These are three different optimization functions, and the fact that they frequently return the same answers on human problems is not evidence of a moral realm they are all detecting. It is evidence of common constraints on the optimizers: a shared cognitive architecture, evolved for cooperation under uncertainty, running on agents with overlapping needs, vulnerabilities, and horizons. The same phenomenon appears in machine learning, where different loss functions yield similar policies once they are constrained by the same environment. Convergence is computational, not transcendental. Agreement among rational agents is evidence of common constraints on cognition and communication — constraints that are perfectly real, and perfectly epistemic, not moral.

Conditionalism supplies the general form of this diagnosis. Interpreted claims require background conditions, and normative conclusions additionally require standards. Moral disagreement between fully informed, rational agents need not be an error; it can reflect different value premises, exactly as agent-binding predicts. When premises are shared — as among humans they often are — convergence needs no further metaphysical explanation. The Triple Theory may therefore identify important agreement shaped by human evolutionary, cultural, and cognitive conditions. The procedure can expose and test that agreement. It cannot by itself show that the resulting content binds every possible agent.

Reasons Without a Realm

Parfit’s most seductive claim deserves separate treatment: that reasons exist “out there,” accessible to intuition the way mathematical truths are. The analogy is meant to reassure — nobody thinks arithmetic is subjective — but it cuts the other way. Mathematics is not discovered in the Platonic ether either; it is an internally consistent game built on axioms chosen for their utility and coherence. The theorems are objective given the axioms, which is precisely the conditional structure the realist was trying to escape.

So what is a reason, if not a metaphysical entity waiting to be intuited? A reason is a locally stable pattern of preference propagation across agents. That is the whole of it. Reasons are emergent invariants in the information flow of agency — regularities that hold across a wide range of agents and conditions, stable enough to be named, argued over, and relied upon. When we say “we have reason to prevent suffering,” we are saying that under nearly all viable agency-preserving conditions, the suppression of unnecessary suffering increases mutual predictability and flourishing. That is a model of equilibrium, not a revelation of essence. The moral realist’s ontology of reasons is a projection of the agent’s own deliberative architecture onto the universe — the shadow of the procedure, mistaken for furniture.

This is the constructive takeaway, and it is worth flagging where it leads. If moral principles are evolutionary equilibria in the space of choice — stable strategies for agents seeking to preserve their agency — then the right way to study them is not intuition-mongering about eternal truths but the analysis of what survives when agents with different values must coexist. That is the ultimate metagame, and Part VI of this volume is built on it. Parfit sought the laws of ethics. I am after the physics of choice.

Rand’s Half-Victory

The other near miss approaches from the opposite direction. Where Parfit built objectivity out of convergence among moral theories, Ayn Rand claimed to have derived it from the requirements of life itself — and her Objectivism is no museum piece. It retains serious defenders who restate it in rationalist terms, Peter Voss’s formulation being the strongest I have engaged, and the restatement inherits both the system’s genuine achievement and its structural flaw.

The achievement first, because it is real. Objectivism presents itself as a corrective to moral systems that elevate duty, sacrifice, obedience, or collective authority above the individual — and those systems deserve the correction. Rand diagnosed genuine pathologies: the erosion of agency, the sanctification of self-denial, the use of moral language as a tool of control. Against mysticism and moralized suffering she proposed a virtue-centered ethics of reason and rational self-interest — morality as a guide to flourishing rather than a mechanism of guilt. A great deal of this survives scrutiny: the emphasis on agency, the rejection of self-sacrifice as a moral ideal, the insistence on coherence between values and action, the recognition that virtues function as internalized policies. As a description of how certain agents may live well under certain commitments, Objectivism is often good advice.

The difficulty arises at the exact point where Objectivism stops offering a personal ethics and claims to have discovered objective morality itself. The system grounds its principles in effectiveness — their capacity to support survival, productivity, psychological health, long-term flourishing — and then treats that effectiveness as sufficient for legitimacy. But these are different questions. Instrumental success answers what tends to work, given a goal. Ethical authority concerns which reasons may claim standing over an agent at all. Objectivism treats the two as interchangeable, and the conflation is the engine of everything that goes wrong downstream.

The clearest symptom is the handling of terminal value. Human life, flourishing, and rational self-interest function in Objectivism as self-justifying ends — discoveries about reality rather than commitments made by agents situated within it. But no terminal value works that way in this framework. Once an agent adopts an end, the world imposes constraints on what can achieve it, and conduct can be assessed against the stated end, factual model, and decision rules. Prior to adoption, the end has not acquired authority over that agent. This is agent-binding again: endorsement must be attributed with evidence, while application asks what follows if the premise is fixed. Rand’s system oscillates between presenting its ethics as a rational choice and presenting it as moral law, and the two postures cannot be held simultaneously without tension. Chosen, it is a recommendation. Imposed, it requires a further account of authority.

Two further fragilities follow from grounding ethics in effectiveness. The first is the egoism argument. Objectivism holds that predation, deception, and exploitation undermine the agent who employs them — trust erodes, relationships destabilize, self-esteem decays. Often true, for psychologically typical agents in open social environments. But it is a convergence claim, not a constraint. There are contexts where asymmetric power, enclosure, or short horizons insulate predatory behaviour from its costs, and there are agents whose psychological architecture cancels the internal penalties the argument relies on. A foundational ethics must remain intact under adversarial modeling; an ethics that holds only where predation happens not to pay is a fair-weather ethics.

The second is the treatment of consent. Voluntary interaction — exchange, trade, contract — carries the system’s moral weight, and voluntariness is certified by the absence of overt force: a binary threshold. But consent is not binary. Agency can be impaired by manipulation, desperation, information asymmetry, or preference distortion while formal choice remains intact; consent admits degrees and defeasibility, and any framework that cannot say so is blind to coercion as it actually appears in real systems. The same bluntness infects the political extension: principles derived from rational self-interest are taken to ground legitimate institutions, with authority justified by correctness rather than delegation — no internal mechanism distinguishing guidance from governance. An ethics that aspires to inform collective enforcement needs explicit criteria of legitimacy, and without them its principles are standing invitations to appropriation by power.

Objectivity in the Right Place

Set the two systems side by side and the shared error is unmistakable. Parfit locates objectivity in convergent moral content — propositions certified by the agreement of theories chosen for their agreement. Rand locates it in correct moral content — flourishing and rational self-interest promoted from effective strategy to discovered law. Both are reaching for something real, and both grab the wrong part of it.

Two kinds of public assessment remain in the vicinity of ethics, and neither generates a terminal value. First, a bound claim can be tested: given a normative premise, attributed commitments, a factual model, and conflict rules, does the verdict follow? Second, descriptive inquiry can assess whether a framework preserves an agent’s capacities to evaluate, consent, revise, and act under stated conditions. Those capacities make endorsement and refusal possible, but their structural role does not itself prove that every agent must value reciprocal protection. An ethics that erodes them may be self-defeating under its own avowed commitments; a coherent predator may not share those commitments at all. The Ethics of Viability therefore adds the agency-protection premise openly rather than presenting reflective structure as a fixed point that legislates for every valuer.

Ethics, on this view, can function as chosen infrastructure: it states which reasons and protections a framework recognizes before advising what to pursue. Reframed accordingly, both near misses keep what they earned. The Triple Theory survives as a map of where reconstructed human moral theories agree — a useful discovery about shared architecture, not by itself metaphysics. Objectivism survives as a recommendation for agents who endorse flourishing, autonomy, and long-term coherence as ends — a useful strategy, not by itself law. Systems that conflate success with authority, or coherence with ontology, become especially dangerous when scaled, enforced, or weaponized. A philosophy fit for those transitions must distinguish descriptive constraint, normative guidance, and legitimate authority.