Moral Luck

The Philosophical Problem

Moral agents live among causes they initiate, inherit, misunderstand, and encounter. One driver acts carelessly and reaches home without incident; another acts with an equivalent degree of carelessness and kills a pedestrian. One person never faces the invitation to participate in political atrocity; another, formed in similar dispositions, confronts that invitation and accepts it. One artist abandons intimate obligations to undertake a difficult project and succeeds; another makes the same sacrifice and fails. These differences seem morally momentous, although the differences between the cases can be traced partly to fortune. The moral luck problem asks whether the resulting differences in moral judgment are justified.

Its philosophical force arises from commitments that appear independently attractive. We judge persons for what they do, sometimes with reference to outcomes, achievements, missed opportunities, and harms. We also hold that a person cannot deserve blame simply because an event beyond that person's control occurred. A theory of moral responsibility must make room for both convictions or explain why one should be revised. The dispute is therefore about the conditions of justified evaluation, the objects of evaluation, and the moral significance of contingency. Its implications reach conceptions of persons, social institutions, legal punishment, practical rationality, retrospective interpretation, and the meaning of a completed life.

The intellectual problem predates its contemporary name. Aristotle investigated fortune and the fragility of flourishing, Epictetus distinguished what is up to us, Kant defended the unconditional worth of a good will, and Adam Smith gave a sustained analysis of the influence of fortunate outcomes on praise and blame. Bernard Williams and Thomas Nagel transformed these antecedents into a modern problem with a distinctive architecture in their adjacent contributions to the 1976 supplementary volume of the Proceedings of the Aristotelian Society (Williams, 1976; Nagel, 1976). The subsequent editions in Williams's Moral Luck (1981) and Nagel's Mortal Questions (1979) are important sources of the debate. Their projects differ: Nagel emphasizes the threatened coherence of ordinary moral assessment under a control condition, whereas Williams investigates how luck bears on projects, justification, regret, and the ethical significance of a life. An interpretation that treats their essays as one identical thesis loses the central philosophical tension.

The inquiry must begin with precision about judgment. To say that a collision was worse because someone died is one proposition; to say that the driver was more blameworthy is another. To say that the driver owes compensation introduces a third. To say that she understandably feels a deeper form of regret states a fourth. A morally serious theory must determine which of these propositions requires control, which requires causation, and which depends on an outcome even when desert does not.

Definition, Objects, and Scope

Moral luck occurs, in the standard philosophical formulation, when factors outside an agent's control make a difference to the moral evaluation of that agent in some respect. The qualification 'in some respect' matters. Without an explicit object of assessment, the definition can absorb almost every instance of ordinary causal dependence. In a rigorous analysis, evaluation must identify whether it concerns action, intention, judgment, disposition, harm, responsibility, character, achievement, relationship, or desert. The Stanford Encyclopedia of Philosophy's contemporary exposition treats moral luck as a tension between morally assessing persons and acknowledging that significant assessed factors exceed their control (Nelkin, 2025). This account offers a diagnostic starting point rather than a conclusion that every such assessment is legitimate.

Luck is relational to a domain of control. An event can be statistically likely yet still beyond an individual's power to determine; an extremely unlikely result may still follow from a voluntary risk. The fact that an agent deliberately exposes others to a foreseeable hazard establishes control over the risk-taking decision, though not necessarily over which particular consequence follows. The resulting harm may be causally traceable to the decision while remaining partly dependent on independent circumstances. Neither causal connection nor inability to guarantee an outcome by itself settles responsibility.

Several further distinctions organize the subject. 'Moral luck' is a technical designation for the evaluative role of fortune; it does not mean that the lucky person has been morally benefited. Bad resultant luck may be terrible for a victim and morally complicating for the agent. 'Moral desert' concerns what praise, blame, or treatment someone merits; 'moral status' concerns which beings deserve moral consideration. 'Responsibility' can designate causal involvement, attributability, answerability, accountability, or liability. These senses overlap but have different conditions. A person may owe repair because of causal involvement without having acted culpably. Conversely, a dangerous intention can attract moral criticism without ever becoming an outward injury.

The research subject is the general structure of luck-sensitive moral assessment. It includes control, desert, character, contingency, retrospective meaning, and institutional treatment. Detailed accounts of free will, criminal attempt law, responsibility for attitudes, and distributive justice enter when they test its central proposition, while retaining their own independent philosophical questions.

The Principle of Control and the Paradox

The control principle states that persons are morally assessable only insofar as what they are assessed for depends upon factors under their control. The strongest intuitive argument begins with fairness. Blame claims that an agent deserves disapproval, resentment, sanction, or some negative moral response on account of a feature appropriately imputable to that agent. If the relevant difference could not be affected by the agent, making that difference decisive for blame looks arbitrary. Similar conduct in similar conditions ought to attract similar fundamental assessment. The control principle can therefore be reconstructed as a condition of equality in desert, not merely a psychological preference for sympathetic judgment.

The opposing premise is equally familiar: our moral evaluations vary with outcomes and opportunities. Two culpably reckless drivers receive different social and often legal condemnation when only one kills. The agent whose ambition yields an admirable achievement appears retrospectively different from the agent whose analogous ambition produces only abandonment and failure. People may be praised for courage only because circumstances demanded courage. The paradox is the apparent coexistence of the control principle with morally significant assessments that vary with uncontrolled conditions.

A deliberately austere argument is helpful. Let A and B be equivalent in every feature they control relevant to an act. Suppose an external contingency changes only the realized result. If every deserved assessment supervenes exclusively upon controlled features, the agents are equally blameworthy. If realized harm increases blameworthiness under otherwise identical conditions, the agents are unequally blameworthy. These judgments cannot both hold for the same respect, standard, and degree of blameworthiness. The contradiction disappears only if one rejects one premise, identifies a genuine controlled difference, or discovers that apparently rival judgments concern distinct evaluative objects.

That third strategy must earn its explanatory force. It would be evasive to call an outcome 'wrong' while refusing to investigate why its actual harm matters to the wrongdoer, victim, and community. It would be equally careless to infer heightened culpability directly from heightened suffering. The philosophical task is to construct a vocabulary in which the impact of results and the equality of agents can both be articulated without concealment. Nagel's argument shows the difficulty of drawing a completely secure boundary around control because causal histories, available situations, and character formation themselves have contingent components (Nagel, 1979, chap. 3). The point widens a puzzle about accidents into an investigation of moral agency.

Ancient Ethics and the Vulnerability of the Good Life

Aristotle's Nicomachean Ethics offers a foundational analysis of contingency without formulating the modern fourfold taxonomy. In Book I, chapters 8–11 (especially 1099a31–1101a21), he argues that flourishing, eudaimonia, fundamentally consists in activities in accordance with virtue, while significant external goods make its full realization possible. Friendship, political resources, and favorable circumstances provide occasions and instruments for excellent action. The fortunes of Priam and the question attributed to Solon, whether a life can be judged happy before its end, illustrate the exposure of a life to reversals (Aristotle, Nicomachean Ethics I.8–11).

Aristotle's achievement is to distinguish the center of a good life from the material conditions of its exercise. Fortune supplies possibilities and threats; virtuous activity supplies its distinctive ethical orientation. A person who meets tragedy with courage can display excellence, yet repeated catastrophe can restrict the very activities through which a flourishing life unfolds. The relation is neither reducible to a calculation of pleasant outcomes nor exhausted by insulated intention. The value of an actual life depends upon a history in which the agent exercises capacities under conditions that the agent did not wholly choose.

His analysis of voluntary action in Book III adds the indispensable moral-psychological dimension. Coercion, ignorance of relevant particulars, and mixed actions under pressure are ethically distinct. Such distinctions prevent every unfortunate outcome from counting as culpable. What matters is how choice, knowledge, and the surrounding conditions enter the action. Aristotle thereby anticipates both the modern concern with control and the observation that moral opportunity and achievement depend on fortune. His discussions of habituation in Book II also show why formed character is historically produced through practices and social settings; the present moral agent has a developmental history.

These ancient materials are antecedents to the modern debate rather than early statements of Nagel's exact conceptual schema. Their central question often concerns flourishing and virtue rather than the distribution of blame. That difference makes them philosophically instructive: a theory that protects equal culpability across outcomes may still need to acknowledge outcome-dependent accomplishments and damaged lives.

Stoic Control and the Domain of Judgment

The Stoic discipline articulated by Epictetus begins with the division between what is up to us and what is not. The Enchiridion, §1, distinguishes judgment, impulse, desire, and aversion from body, property, reputation, and office. The Greek formulation ta eph' hēmin, things dependent on us, provides a powerful ancestor of a moral focus on agency. A person who treats external fortune as the measure of ethical worth places what matters most beyond the proper province of choice (Epictetus, Enchiridion 1).

The argument has a practical dimension. Concentrating on one's judgments and volitions can reduce the tyranny of unpredictable events over conduct. It has a metaphysical and evaluative dimension as well: if virtue is located in the sound exercise of the faculty of choice, an adverse result need not destroy its moral worth. This protection answers a powerful aspect of the moral luck paradox. A life may suffer catastrophe while the agent remains ethically steadfast.

Yet the very domain described as 'up to us' remains a contested locus of control. How did the agent acquire the capacity for sound judgment? Which opportunities enabled reflection and discipline? Do coercive conditions and severe deprivation impair the exercise of choice? A Stoic can distinguish the normative location of virtue from the causal origin of dispositions, but doing so leaves a question concerning constitutive luck. The modern challenge presses this question even where ancient Stoic psychology provides reasons to defend freedom in judgment.

Kant and the Unconditional Value of the Good Will

Kant gives the anti-luck intuition its most famous modern expression. In the first section of the Groundwork of the Metaphysics of Morals, Akademie 4:393–397, the good will has unconditional moral worth, whereas talent, power, happiness, and the success of action can be valuable or harmful depending on the will governing them. Kant's famous thought experiment imagines a worthy will that, through the disfavor of fortune or inadequate natural endowment, accomplishes none of its chosen ends. Its moral worth remains intrinsic to its willing rather than determined by successful effects (Kant, 1785/2012, 4:394–396).

This is a claim about moral worth, not a general denial that consequences matter to practical reasoning. A duty of beneficence requires attention to what helps others; reckless indifference to foreseeable outcomes can reveal a defect in willing. Kant's contrast concerns the condition of worth: favorable fortune cannot turn an ill will into a good one, and failed execution cannot annihilate the worth of a will acting from duty. The best interpretation distinguishes an outcome's relevance to the rightness of practical choices from its authority to manufacture good will retrospectively.

Kant's own framework also sharpens rather than resolves every form of moral luck. Agents vary in temperament, formative experience, opportunity, and their ability to perceive duties. If the moral worth of a will is evaluated by its practical orientation, it remains to specify how that orientation can appropriately be attributed to agents whose histories were never chosen. The Kantian emphasis on autonomy is a resource for analyzing this issue; the argument about causal constitution demands further work. Williams's comparison with unconditional moral value illuminates how modern moral thought attempts to secure an evaluative domain that fortune cannot undo (Williams, 1981, pp. 20–39).

Hume, Smith, and the Fortune of Moral Sentiment

Hume's Treatise of Human Nature analyzes moral approval and disapproval in relation to durable qualities of mind, passions, and the motives expressed in conduct. In Book III, part III, section I, his account of natural virtues places character and the sentiments it evokes at the center of evaluation (Hume, 1739–1740, T 3.3.1). This attention to motivation helps explain why the same external performance can have different moral meanings. It also invites inquiry into the natural and social processes through which spectators acquire the information required to evaluate character. Effects serve as evidence, yet evidence of an intention must be distinguished from the content of the intention itself.

Adam Smith offers a particularly direct historical anticipation of moral luck in The Theory of Moral Sentiments, Part II, Section III, explicitly devoted to fortune's influence on judgments of merit and demerit (Smith, 1759/1790). Smith states the equitable maxim that praise and blame properly trace the intention or affection from which an action proceeds, rather than accidental external consequences. He then investigates why actual success and failure nevertheless alter gratitude, resentment, and the felt force of desert. The gap between an acknowledged rule and actual spectator sentiment is the central object of his analysis.

Smith's analysis is unusually important because he does not simply record inconsistent emotions. Consequences organize how spectators perceive injuries, benefits, intentions, and social obligations. The public moral world contains actual sufferers and actual beneficiaries. A harmless attempt and a completed assault can issue from comparably wrongful intentions while generating different relationships of resentment and repair. Smith's proposed account of fortune's moral psychology helps distinguish an explanatory claim about sentiments from a justificatory claim about deserved blame. The transition from what people feel to what they are justified in feeling is precisely where the contemporary argument takes place.

The history thus contains three enduring motifs: virtue and flourishing require a contingent world; the will can be evaluated by standards independent of realized results; and actual results influence moral sentiment even among observers who endorse intention-sensitive fairness. The philosophical conflict emerges where those motifs are treated as competing answers to one undifferentiated question.

Williams: Projects, Justification, and Agent-Regret

Bernard Williams's 'Moral Luck' addresses an instability in the authority of moral judgment over a life. His example of a fictionalized Gauguin considers an artist who leaves personal obligations in pursuit of an uncertain creative undertaking. The assessment cannot be settled simply by projecting a moral ledger from the present moment into an unknown future. If the project succeeds, its achievement can become part of the artist's retrospective justification; if it fails, that resource disappears. Williams does not thereby claim that artistic success converts every morally objectionable abandonment into a permissible act. He examines whether a person's practical identity and the possibility of justifying a life-defining decision can be radically hostage to a future the person could not know in advance (Williams, 1976; 1981, pp. 20–39).

The argument must be reconstructed carefully. First, some commitments are so central that the agent's life would be different in kind if the projects were abandoned. Second, a decision about such commitments must be undertaken from within an uncertain history, without a guarantee that the projects will justify the sacrifices they demand. Third, a retrospective assessment of whether the decision was worth making may legitimately take achievement or failure into account. The resulting dependence upon luck threatens an image of practical justification as wholly available from the agent's deliberative position at the time. This is a challenge to an ideal of rational self-sufficiency as well as to the alleged insulation of morality.

Williams develops a further notion through cases of unintended harm: agent-regret. A person who inadvertently injures another in circumstances in which no reasonable precaution was omitted can still experience a distinctive practical sorrow because the injury came about through her agency. The regret differs from regret felt by a bystander who merely wishes the harm had not occurred. It can motivate apology, acknowledgment, or an attempt at repair, even when the agent's culpability is absent. The presence of such regret shows that the morally significant relation between an actor and the consequences of acting extends beyond assessments of deserving blame.

Consider two drivers who exercise precisely the same reasonable care. An unforeseeable failure causes one of them to injure a child, while the other arrives home without incident. If one focuses on culpability, both may be blameless. Yet only one has a concrete relation to an injured child and the burden of living with having caused the injury. The divergence is ethically real, although the verdict concerning their culpable wills is equal. Williams forces moral philosophy to explain why agency has a retrospective dimension that a theory of choice at an isolated instant cannot exhaust.

There are serious objections. A retrospective success may be an unreliable vindicator of a sacrifice made on poor evidence; even spectacular art cannot automatically nullify a person's obligations to family. An argument that confuses achievement with moral permission would license a dangerous retrospective favoritism toward successful agents. The defensible reading therefore distinguishes ethical justification of an entire life or project from the rightness of particular conduct, and keeps moral accountability for obligations in view. Williams's power lies precisely in showing why ethical evaluation is broader than a simple, invariant score for motives.

Nagel: Four Dimensions of Moral Luck

Thomas Nagel asks how people can remain responsible when so many determinants of their actions escape their control. The question begins with an ordinary premise: it appears unfair to hold someone morally answerable for what the person could not determine. Yet Nagel observes that the boundaries of moral judgment repeatedly extend into uncontrolled territories. His paper, published in its original form alongside Williams's in 1976 and revised for Mortal Questions in 1979, distinguishes four dimensions: luck in outcomes, circumstances, character, and causal antecedents (Nagel, 1979, pp. 24–38).

Resultant luck concerns what emerges from an act. Circumstantial luck concerns what a person happens to face. Constitutive luck concerns what kind of person one is, including tendencies and capacities shaped by forces outside initial choice. Causal luck concerns the antecedents that make a person's decisions and actions occur. Each dimension challenges the view that moral desert can be cleanly assigned to an independently self-made agent. The fourth dimension broadens the other three: causal history can undermine the supposedly secure domain of intentional choice itself.

The fourfold classification is analytic rather than exhaustive of every moral experience. The kinds overlap. A driver's negligent act is shaped by character and training, occurs in circumstances containing some hazard, issues from a causal history, and produces a realized outcome. The same case therefore belongs to several categories. What makes the classification useful is the distinct question posed at each stage: would we judge differently if the result, opportunity, disposition, or causal background were different while holding fixed the relevant other features?

Nagel's challenge also raises a dilemma for rigorous control theories. Narrow control to conscious intention and much ordinary moral judgment becomes unavailable. Widen control to include character, action, and reasonably predictable effects, and it becomes difficult to show why these are genuinely under the agent's control when their origins reach beyond the agent's making. Moral evaluation may rely on a practical perspective that cannot be reconciled completely with the impartial view of a world of causes. Rather than offering a quick solution, Nagel exposes a structural tension between our conception of persons as accountable agents and our recognition of persons as parts of nature.

Resultant Luck and the Moral Difference of Outcomes

Resultant luck is clearest when two agents make equally culpable choices yet different results follow. An attempted murderer whose bullet misses because of a gust of wind and a murderer whose bullet hits have the same relevant intention and may have exercised the same degree of effort. One victim survives; another dies. An outcome-sensitive theory says that completed harm adds a morally relevant fact. An intention-centered theory says the wrongfulness of the purpose and the culpability of the agent were equivalent. The cases invite disagreement about which features are genuinely moral and which are evidence of social consequences.

The strongest argument for equal blameworthiness is a fairness argument. If the relevant choices, intentions, beliefs about risk, and degrees of control are fixed, no agent-controlled moral property differs. Where blameworthiness is intended to measure the quality of will displayed in making a choice, assigning unequal blame would make desert depend upon the indifferent trajectory of a projectile. This conclusion remains consistent with the judgment that the completed homicide is a greater harm, produces additional duties, and constitutes a legally distinct offense.

The strongest argument for outcome sensitivity appeals to the completed exercise of agency. Acting wrongfully is not an isolated inner event but a worldly undertaking. A person who actually destroys a life has participated in a grave moral reality, not merely desired or attempted it. The harm is partly the agent's doing, even where its exact occurrence was uncertain. The issue is whether participation in realized harm properly increases the degree of blame deserved, or whether it creates another kind of responsibility: liability, accountability to a victim, obligations of repair, and a changed practical relationship.

A possible reconciliation separates the wrongness of what was done, the culpability of choosing it, the severity of injury, and the duties that follow. But reconciliation is not automatic. Some theorists defend genuinely luck-sensitive degrees of blameworthiness; others deny them while admitting that realized effects are morally important. The difference is substantive: the same account cannot treat outcome as irrelevant to culpability and simultaneously invoke outcome to justify extra retributive blame. It must state which normative function is being assessed.

The standard examples are also idealized. Real cases rarely establish that both agents had identical capacities, beliefs, precautions, and opportunities. One fatal crash may reveal inattentiveness previously hidden, and one survival may have involved prompt avoidance or luck in another person's actions. The philosopher therefore uses controlled counterfactual pairs to isolate the principle, while actual judges require evidence about what the individual knew and could reasonably have done.

Circumstantial Luck and Unequal Moral Tests

Circumstantial luck arises because the world distributes moral temptations, dangers, and opportunities unevenly. An individual who leaves a country before a repressive movement takes power may never face a decision that another person must confront under threat. A person living in secure conditions may find honesty easy; another under severe coercion may confront costly choices. The relevant feature is variation in what is demanded, not merely variation in which outcomes occur after action.

The challenge to equal desert has two directions. A fortunate person may avoid exposing vicious dispositions because no relevant temptation appears. A less fortunate person may reveal a weakness under exceptional pressure and receive severe condemnation. Conversely, an individual fortunate enough to encounter a rescue may receive public praise for courage, while another with comparable courage has no occasion to display it. The record of observable virtue and vice becomes dependent on the opportunities life distributes.

A response based on actual choice holds that it is reasonable to judge people for decisions they actually face rather than hypothetical decisions they might have faced. Actual wrongdoing creates answerability; imagined wrongdoing often does not. This response respects the epistemic and relational importance of actual action but leaves a residual inequality: an agent is condemned for possessing and exercising a vice when the equally disposed person whose life was easier remains untested. To say that the latter deserves the same condemnation requires a further thesis about what makes counterfactual character morally assessable.

Circumstantial luck also complicates blame in contexts of structural injustice. Poverty, institutional coercion, conflict, medical constraints, and educational opportunity can affect the menu of available actions. These circumstances neither cancel agency in every case nor render the resulting choices morally equivalent. They enter the analysis of capacity, expected sacrifice, reasonable alternatives, and obligations of institutions that created or maintained the constraints. Equal moral concern may demand attention to unequal conditions rather than abstraction from them.

Constitutive Luck: Character, Formation, and Agency

Constitutive luck concerns the partly given features through which moral agency is exercised: temperament, intelligence, habits, emotional dispositions, social learning, and opportunities for cultivating judgment. Human beings do not choose their initial biological constitution, earliest caregivers, language, or cultural environment. By the time they can reflect upon these influences, those influences have already helped form the capacities and standards through which reflection occurs. An evaluation of character therefore faces the problem of attributing to a person qualities whose origin includes uncontrollable factors.

The problem cannot be met merely by saying that adults can change. The capacity to recognize a defect and undertake change is itself unequally developed. Yet the existence of formative influence does not establish that mature reflection has no significance. An agent may become answerable for sustaining, endorsing, correcting, or neglecting a disposition over time. Responsibility can attach to patterns of responsiveness to reasons even when their complete causal origin is not self-created.

Character evaluation also differs from responsibility for specific acts. Someone may be generous without ever encountering extreme need. Another may act generously on one occasion while generally cultivating self-interest. A stable trait is inferred from a pattern of responses, but our available observations are selective. A rigorous theory must distinguish the character someone has, the character actually revealed, and the character it is justified to attribute on the evidence. Constitutive luck is thus connected to moral epistemology and the ethics of judging others.

Nomy Arpaly's inquiry into moral agency and unprincipled virtue clarifies why the quality of motives need not coincide with an agent's explicit theory of her motives (Arpaly, 2003). Her work is relevant to constitutive luck because it locates morally significant responsiveness in psychological life that can exceed deliberate self-description. The constitutive objection presses beyond social explanation. If every feature that makes one morally responsive ultimately depends upon a combination of antecedent endowments and circumstances, can desert survive? The strongest affirmative response locates responsibility in the exercise of mature capacities, not in creating oneself from nothing. The strongest skeptical response asks why the exercise of those capacities should ground unequal fundamental desert when possession of the capacities is itself fortunate. The resulting debate turns on what kind of control responsibility needs.

Causal Luck, Determinism, and the Regress of Control

Causal luck concerns the antecedent events and conditions through which an action comes to occur. Deterministic causation makes the challenge vivid: given the state of the world and the laws, an agent's choice may follow from antecedent conditions the agent never selected. Indeterminism supplies a different difficulty. If the course of willing includes an undetermined occurrence, why should an accidental contribution improve the agent's authorship of the decision? Neither invoking causal necessity nor inserting randomness alone resolves the issue of control.

The moral luck debate is related to, but distinguishable from, the debate about free will. A compatibilist can hold that an action counts as the person's own when it expresses deliberation, responsiveness to reasons, and capacities appropriately free of coercion, even if its causal history is determined. An incompatibilist can demand a stronger form of authorship or alternatives. Both positions must explain how the luck of one's constitution and opportunities bears upon fairness of judgment. A metaphysical account of free choice does not by itself establish what degree of praise, blame, or punishment is appropriate.

The regress is especially severe for theories that demand ultimate responsibility. To be ultimately responsible for an act, one might be required to be ultimately responsible for the kind of person who chooses it. That demand iterates through earlier choices and their character-forming conditions. If every relevant earlier choice presupposes characteristics not yet chosen, the regress threatens to make ultimate self-authorship impossible. This argument motivates skepticism about basic desert, but it does not directly refute other models of responsibility grounded in interpersonal answerability, deliberative competence, or prospective regulation of conduct.

The key conceptual lesson is that causal explanation and normative attribution answer different questions. Discovering causes explains how an act occurred. Attributing an act to an agent determines whether demands for justification, apology, repair, or blame are appropriate within a moral relationship. Normative attribution must be constrained by causal facts about capacity and coercion, but it cannot be reduced to an inventory of causes. The question remains whether those constraints are sufficient for desert or justify only a more modest practical accountability.

What Exactly Can Luck Change?

The central methodological result of this investigation is a separation of evaluative dimensions. First, deontic appraisal asks whether a choice or act was permissible, required, or wrong under the relevant moral reasons. Second, quality-of-will appraisal asks what concern, disregard, hostility, or responsiveness the agent's action manifests. Third, aretaic appraisal concerns enduring virtues and vices. Fourth, outcome appraisal concerns realized benefits, harms, and achievements. Fifth, relational responsibility concerns what an agent owes to those affected. Sixth, desert concerns which reactive or institutional responses the agent merits. These dimensions can come apart in ordinary and exceptional circumstances.

Zimmerman (2002) usefully distinguishes deontic, aretaic, and what he calls hypological evaluation, concerning moral praise- and blameworthiness. His distinction clarifies why the conclusion that a completed killing is worse than an unsuccessful attempt does not determine whether the killer is more blameworthy. The relevant claim could instead concern wrongness, harm, damage, or liability. The analytical importance of such differentiation does not imply that each dimension is independent of the others. In an actual society, moral responses draw upon several dimensions simultaneously.

A practical illustration involves a physician who makes a medically reasonable decision under uncertainty and whose patient nevertheless dies. The tragic result makes the outcome different from that of a successful treatment, and it can create duties of explanation, documentation, and institutional learning. It does not alone establish negligent clinical judgment. If relevant evidence was ignored, culpability may be present; if the adverse event was unforeseeable, the agent's professional responsibility must be assessed differently. The same logical distinctions apply beyond medicine to engineering, public administration, collective decisions, and personal relationships.

The proposal that all conflicts dissolve through distinctions should be resisted. Genuine disagreement persists about whether realized harm can independently intensify blameworthiness and whether counterfactual character can warrant desert. Differentiation instead determines what the disagreement is about. It prevents arguments for outcome-sensitive compensation from being used without further premises as arguments for outcome-sensitive punishment.

Intention, Recklessness, Negligence, and Risk

A defensible control principle must distinguish intention from foresight, reckless disregard from inadvertence, and reasonable risk from culpable endangerment. Intention concerns an end or means adopted by the agent; recklessness commonly involves conscious disregard of an unjustifiable risk; negligence concerns failure to meet a warranted standard of attention or precaution. The degrees of moral and legal blame associated with these categories depend on what the person knew, what could reasonably be known, the importance of the interests at stake, and how difficult the precautions were.

Suppose two people knowingly expose others to the same serious unjustifiable hazard. That choice itself supplies a reason for blame independent of who happens to suffer. If an injury occurs, it adds an actual victim and a new chain of duties. Whether it adds further blameworthiness depends upon whether the moral object of blame includes the realized violation, the risk imposed, or both. The distinction between imposing a risk and causing an injury is fundamental: risks can be wrong even when no one is harmed, and harms can occur without a culpable imposition of risk.

Some theories treat blameworthiness as scalar and allow the successful realization of a wrongful intention to increase its degree. Others regard intention, attention, and the reasonable expectation of outcomes as the proper variables. A compromise can be principled only if it states its criterion. One may hold that outcome contributes to an offense's gravity without contributing to the agent's ill will. One may instead defend an integrated conception of wrongdoing in which completion matters to deserved blame. Either view must show why an unforeseeable accident should, or should not, change the moral standing of the actor.

Negligence creates additional difficulties because control over attention is indirect. A person does not ordinarily choose each momentary failure to notice. Still, practices of training, vigilance, and risk management can be under extended forms of agency. Calling every lapse 'uncontrolled' obscures the normative demand to organize one's conduct safely; calling every accident culpable treats chance as a substitute for evidence. Responsibility for negligence must be evaluated through reasonable capacities and opportunities, including the institutional conditions in which decisions are made.

Moral Blame, Praise, and Reactive Attitudes

Blame is more than a proposition that an act was wrong. It can involve a judgment of culpability, a reactive attitude such as resentment, a demand for acknowledgment, public criticism, withdrawal of trust, or a punitive response. These different activities are often bundled under one word. Some are appropriate in a relationship because an actual violation occurred; others require a justified assessment of the agent's quality of will. The moral luck question becomes clearer when the form and purpose of blame are specified.

A victim can legitimately address a wrongdoer as the cause of actual injury even where another person made an equally culpable attempt but was thwarted. The victim's position is not a mistake about comparative desert. It is an actual relation to an agent who has caused a wrong. Conversely, the victim's understandable anger cannot by itself settle how much state punishment is deserved. An expressive response and a retributive tariff obey different justificatory demands.

Praise shows a parallel asymmetry. A rescuer who saves a life may deserve gratitude from the rescued person, while an equally courageous person whose effort fails may display the same admirable concern. Gratitude naturally attaches to a realized benefit; admiration for courage need not. If we insist upon one undifferentiated measure of praise, this distinction vanishes and favorable luck appears to create virtue. If we acknowledge different modes of appreciation, successful rescue and admirable effort can each receive appropriate recognition.

A further question concerns whether the presence of a blameworthy quality always warrants blaming. Even a person who merits criticism may be treated mercifully for independent moral reasons. The duty to respond with justified attention to vulnerability, relationships, and our own susceptibility to error is not equivalent to the claim that blameworthiness disappears. This distinction has become important in the contemporary literature on mercy and the ethics of blame.

The Egalitarian and Counterfactual Responses

The egalitarian response protects the control principle by making degree of blameworthiness depend on agent-controlled features. In a matched pair of would-be killers, the successful and unsuccessful agents may be equally blameworthy, though their deeds differ in realized damage. The response gains fairness by refusing to allow a random gust of wind to alter the fundamental measure of culpability. The difficulty is to explain the distinctive moral weight of actual harm without covertly reintroducing luck-sensitive desert.

Michael Zimmerman develops an ambitious extension: the responsibility relevant to degrees of blame can depend upon counterfactual truths about what agents would do under circumstances they never encounter (Zimmerman, 1987; 2002). If a fortunate individual would have acted as viciously as an unfortunate wrongdoer under the same morally relevant conditions, the fortunate person may be blameworthy in a way ordinary assessments overlook. The view tries to take the control principle seriously beyond actual opportunities, and thereby reveals how much hidden responsibility an egalitarian morality might contain.

This approach faces the truth-value problem. Some conditionals describing what a person would do in radically different circumstances may lack a determinate truth value. The problem becomes severe when imagined circumstances would have altered the person's character. Which of the person's traits should remain fixed across the comparison? A conditional involving someone raised in a different society might no longer preserve the relevant individual dispositions. It becomes difficult to identify a fair counterfactual benchmark without smuggling back the historical luck under discussion.

A second objection concerns the object of responsibility. Can a person be blameworthy for a never-performed act solely because an unobserved counterfactual indicates its likelihood? The difference between possessing a defect and committing a wrong may have moral importance even where the defect makes a counterfactual act predictable. A third objection is epistemic: courts and moral communities cannot generally know whether remote counterfactual conditions hold. Any theory that extends desert to unperformed actions needs a separate account of the justified practice of accusation.

The more modest egalitarian proposal retains equal culpability for fully matched actual choices while refusing global judgments about hypothetical lives. This limits the ambition of absolute luck immunity but preserves a stable practical constraint. It accepts that a person's evaluative record can depend on opportunities without concluding that every untested person deserves blame for every offense that would have been committed under a different biography.

Defense of Moral Luck and Its Strongest Objections

Defenders of moral luck argue that a complete moral assessment properly takes account of what actually happens to the people involved. Robert Hartman's In Defense of Moral Luck (2017) challenges the premise that uncontrolled consequences can never augment blameworthiness. It draws force from an integrated conception of responsible agency: one may be accountable for a completed deed as a worldly performance, rather than solely for an inner decision followed by morally inert events. The defense captures the significance of actual agency but must identify the principle by which moral weight from an outcome becomes a difference in desert.

The strongest objection remains a matched-agent case. Hold fixed the deliberation, competence, beliefs, circumstances at decision, wrongful aim, degree of risk, and causal contribution up to the point of a chance divergence. If a person whose victim dies deserves more blame than one whose victim survives, the theory must explain why the extra desert attaches to that person rather than only to the gravity of the event and the duties it creates. Calling the outcome 'part of the act' is the beginning of an answer; it cannot alone show why responsibility for the act grows in degree instead of in scope.

A second objection presses the dependence of the victim's presence or vulnerability on coincidence. If completed wrongdoing generates additional deserved blame by sheer occurrence, two identical agents may attract substantially different moral treatment for reasons they could neither know nor avert. This makes the fairness condition especially salient in punitive institutions. A defender can reply that responsibility is for a world-directed undertaking, whose completed nature belongs to the object of responsibility. The disagreement then concerns whether just desert must be invariant under external perturbations.

A third challenge concerns constitutive luck. Some acceptance-based theories endorse outcome luck while retaining familiar standards of control over intentions and action. The critic asks why the same rationale should not extend to the inherited conditions of character and deliberative capacity. If the defender responds that local control is sufficient for responsibility, the theory owes an account of why that local control is normatively significant despite its antecedent fortune. No plausible solution can simply stipulate its preferred threshold without explaining the normative role of agency.

The literature contains a variety of intermediate positions. Some accept resultant luck only with respect to particular objects of responsibility; others accept circumstance-dependent manifestations of character while rejecting bare counterfactual desert. Debate is therefore better mapped by the specific moral dimensions a theory allows luck to affect than by a binary division between people who affirm or reject 'moral luck' in every possible respect.

Moral Luck and Criminal Law

Criminal attempts bring the dilemma into institutional focus. The completed murderer and an equally determined but unsuccessful attempted murderer may differ only because the attempt missed. Nevertheless, legal systems often distinguish completed crimes from attempts, and sentencing commonly reflects realized harm. Is the difference a rational expression of proportionality or an arbitrary reward for criminal bad luck? The question requires a theory of punishment, and such a theory cannot be inferred directly from the control principle.

A strict desert theory makes the degree of culpability central: where culpable choice is equivalent, differences in punishment for the same attempted offense require further justification. Joel Feinberg's treatment of failed attempts argues for taking equal punishment seriously, while distinguishing arguments about deserts from questions about legal administration (Feinberg, 1995; 2003). Another approach treats punishment as appropriately responsive to both culpability and completed harm, because the legal order represents the gravity of actual rights violations. Consequentialist approaches may instead emphasize deterrence, prevention, or incentives and thus need separate evidence about which sentencing arrangements achieve legitimate aims.

David Enoch (2008; 2010) examines relations among moral luck, legal luck, and distributive questions. His arguments make clear that rejecting luck in moral blame does not, by pure logic alone, determine every rule of civil compensation or criminal penalty. The state may have reasons to compensate actual victims and identify actual losses, while limiting punishment for considerations of fairness. Tort law in particular often assigns costs in a framework where negligence, causation, actual damage, and liability each have distinct functions.

There is also an asymmetry between backward-looking desert and prospective social protection. An agent who actually killed may create continuing risks, evidential needs, or repair obligations that an attempted killer did not create. These differences may matter institutionally without proving greater past blameworthiness. Conversely, actual completion may sometimes be evidence of planning or dangerousness. A conscientious legal theory must specify which facts are evidence and which supply independent normative reasons for coercive state power.

The principle emerging from these debates is that the moral luck problem does not settle criminal sentencing by one move. It forces institutions to state their grounds of action and to distinguish desert, deterrence, rehabilitation, incapacitation, expressive condemnation, and restitution. A legitimate system can reject arbitrary disparity in culpability while acknowledging the independent moral claims generated by actual victims.

Agent-Regret, Repair, and Moral Relationships

Williams's account of agent-regret reveals an ethical dimension that survives even when blame is absent. Someone who causes a tragedy through an unforeseeable mishap may reasonably wish to express sorrow, explain what happened, and participate in repair. The agent's regret can be first-personal because the event belongs to the history of what she did. A mere observer can feel compassion, but cannot truthfully say that the same deed was hers. The difference is a fact about lived agency.

Regret must be distinguished from culpable guilt. A person can regret involvement without believing she deserved blame. This distinction protects moral life from the assumption that every serious loss requires a guilty individual. It also protects victims from an overly narrow account in which the absence of fault erases all special obligations of acknowledgment. Not every harm requires compensation by a blameless agent; some may demand institutional arrangements or shared responsibility. The relevant duties depend on relationship, causation, capacity, and distributive justice.

Relational approaches emphasize that wrongs affect the normative positions of persons. A successful assault creates a victim with claims arising from an actual injury. An unsuccessful attempt creates an endangered person who may also have claims, but whose experience and injuries differ. It is possible for offenders to deserve comparable blame and for the injured party to have different rights of repair. That is an intelligible moral structure, not a compromise that denies the reality of harm.

A similar distinction governs praise. A person who risks life in an unsuccessful rescue may display exceptional courage, yet no rescued person can express gratitude for survival that never occurred. Public recognition can acknowledge the admirable risk taken while recognizing the different benefit secured by a successful effort. Gratitude, admiration, desert, and outcome are jointly part of moral life because their grounds differ.

Luck, Evidence, and Moral Epistemology

Most everyday moral assessments are made under serious informational limits. We observe outcomes more easily than motives. The survival of an endangered person may conceal gross negligence, while a catastrophic result may draw disproportionate suspicion toward an agent whose conduct was reasonable. Hindsight changes the perceived predictability of events. These observations are important because some apparent moral luck may consist in biased evidence assessment rather than a genuinely luck-dependent moral fact.

Research on outcome effects and hindsight, including Kneer and Skoczeń (2023), examines how known consequences affect judgments. Such findings require careful philosophical interpretation. An experimentally observed difference between assessments of two vignettes demonstrates a pattern of judgment within the experimental design. It does not by itself determine whether that difference is justified, whether participants are assessing culpability or wrongdoing, or whether features of the scenarios were understood as identical. Empirical psychology supplies a diagnostic resource for normative theory, not a substitute for it.

A fair moral inquiry should follow three evidential disciplines. It should identify the normatively relevant feature being assessed; reconstruct what the agent could reasonably know before action; and separate new information revealed by the outcome from the moral value of the outcome itself. When an injury reveals previously unknown recklessness, a revised blame judgment may be warranted because evidence changed. When the outcome is independent of every relevant feature of culpable choice, unequal blame calls for an additional normative argument.

This produces a further symmetry: philosophical arguments about moral luck often assume perfect knowledge of the agent's mental life, whereas real moral institutions possess fragmentary records. The counterfactual equalization of agents is a powerful tool for isolating principles, but a usable moral practice must also respect epistemic humility. Uncertainty about desert is an ethical reason to avoid confident accusation even when a theory of desert is metaphysically precise.

Social Conditions and the Distribution of Moral Opportunity

Moral luck is also a question about who has the opportunity to become and appear virtuous. Political stability can make peaceful dissent relatively easy, while persecution can make the same choice dangerous. Professional safety systems can support conscientious action, whereas under-resourced institutions may routinely force workers to choose between competing harms. Opportunity shapes the testing, development, and public visibility of character.

The discussion connects to equality of opportunity without equating moral opportunity with material advantage. A person can receive moral advantages from a supportive environment, education, trustworthy institutions, and freedom from coercion. Others may develop admirable strengths in adversity but also face morally destructive constraints. A theory that ignores such asymmetries risks praising the fortunate for the easy availability of good choices and condemning the unfortunate without considering the cost of alternatives.

The social dimension does not imply that all people under oppression act the same way or that no one in difficult conditions can be accountable. Agency is exercised under conditions, and the justified demand on an individual must consider what was available to that individual. This calls for differentiated judgment rather than automatic exoneration. Institutions can also bear independent responsibility for organizing conditions in which preventable moral failures become common.

The relationship between moral luck and distributive luck is informative yet distinct. Distributive justice asks how advantages and burdens should be allocated when people differ in fortune; moral luck asks whether fortune should alter evaluation of what persons deserve as moral agents. The questions intersect where institutions convert unequal opportunity into unequal sanction. They remain distinguishable because compensation for undeserved disadvantage does not resolve whether two agents are equally blameworthy.

The Present State of the Debate

Contemporary scholarship has moved beyond asking whether outcome changes verdicts in simple pairs of cases. Work now tests which kind of control matters, whether responsibility varies by degree or by scope, how blame should be exercised, and whether different theories of agency can survive manipulation and causal history. Recent contributions therefore refine the basic dispute rather than merely restating the 1976 dilemma.

Carlson, Johansson, and Nyman (2026) develop an argument using higher-order control. In their manipulation scenario, agents are placed in conditions where they lack control over their degree of first-order control, although the degree of first-order control itself varies. The example challenges theories that treat control as a simple unproblematic feature, and it asks whether a person can deserve a different evaluation when the possession of control is itself outside the person's command. The point is important because a theory can be lucky at a second level even when an agent exercises control at the first.

Cyr and Hartman (2025) question the common distinction between the degree and the scope of moral responsibility. A theorist may claim that the lucky consequence broadens what one is responsible for without increasing how blameworthy one is. Their discussion tests whether this distinction can explain responsibility for consequences that are themselves shaped by contingency. It presses a central methodological proposal of this monograph: differentiating normative objects is indispensable, but each distinction requires a defensible theory of how the objects belong to the agent.

Hartman (2026) argues that even if moral luck can enhance the blame a person deserves, recognition of shared vulnerability to luck can generate an imperfect duty to forgo retributive blame on some occasions. The distinctive claim separates desert from the all-things-considered permissibility of acting on desert. The moral importance of mercy therefore need not depend entirely on denying an offender's blameworthiness. This position offers a serious challenge to the assumption that a theory of justified blame must automatically authorize its expression whenever blame is deserved.

Research on experimental judgments, social opportunity, equal moral opportunity, and responsibility for omissions likewise continues to reshape the field. These investigations do not establish a universally accepted solution. Their collective significance is that 'control' now names a family of capacities and histories, 'responsibility' names several normative relations, and 'moral assessment' must be specified before the effect of luck can be identified.

An Accountable Pluralism of Moral Assessment

The preceding analyses support an account in which moral evaluation is plural in its objects yet coordinated by common justificatory standards. Its starting proposition is that agents should be assessed according to the respects in which they are answerable. The same episode can involve a culpably chosen risk, an outcome that exceeds control, a victim with a claim of repair, an agent with warranted regret, and an institution with reasons to improve prevention. None of these features can safely serve as a substitute for the others.

The first component is deliberative accountability. At the moment of choice, a person can be asked what reasons she recognized, what interests she considered, and whether the demands upon her were reasonable given her capacities and knowledge. These questions make control a practical concept related to reasons-responsive conduct. They also permit legitimate mitigation for ignorance, coercion, severe disability, and constraints that defeat reasonable alternatives. Deliberative accountability grounds assessments of choice and quality of will without requiring self-creation from nothing.

The second component is realized agency. Actions enter a world in which persons suffer or benefit, relationships change, and projects succeed or fail. Actual consequences can constitute the history and meaning of what someone did, even where they do not change culpability for the prior choice. This component takes Williams's concern with regret and retrospective identity seriously. It recognizes the actuality of victims and beneficiaries and the corresponding moral claims that emerge.

The third component is the fairness constraint on desert. When two agents are genuinely equivalent in all relevant responsible features, the purely accidental difference in outcome requires a separate argument before it can justify more retributive blame. This is a presumption of desert equality, not a claim that consequences are ethically empty. A theory that defeats the presumption must show how realized effects constitute, rather than merely accompany, a greater culpable achievement of agency. Such a demonstration would be substantive; it cannot be borrowed from the severity of harm alone.

The fourth component is epistemic discipline. Moral institutions must judge based on evidence of the relevant criteria, not on vivid outcomes, stereotypes, or retrospective certainty. A strong model of accountability acknowledges the limits of information about character and counterfactual conduct. It therefore distinguishes judgments about an actual wrong, inferences about stable dispositions, and speculative accusations concerning what a person might have done under conditions never encountered.

The fifth component is institutional and interpersonal proportion. The reasons for asking an agent to repair harm differ from the reasons for imposing criminal punishment. An apology can be an acknowledgment of causal involvement; liability can compensate actual injury; punishment requires a public account of justified coercion; personal forgiveness and moral mercy may be appropriate even where culpability persists. Pluralism is accountable when these decisions are traceable to explicit reasons rather than hidden changes of evaluative standard.

This framework does not decree that moral luck never changes blameworthiness. It identifies what a successful argument for that conclusion would need to establish and preserves the possibility of a defended luck-sensitive conception of desert. Its central substantive result is that there is no legitimate inference directly from more terrible consequences to a worse will, or from blamelessness to absence of all special obligations. The moral life of agents and affected persons remains richer than either inference permits.

Open Questions

The most difficult remaining question concerns the origin of agency. If responsibility rests upon reasons-responsiveness, how much of that capacity must be controllable, and what explains why a fortunate capacity supports desert? The constitutive and causal arguments reveal that local accounts of control require a normative defense of their threshold. There is no consensus that an agent's ability to deliberate under actual conditions suffices for ultimate desert.

A second question concerns the morality of outcomes. Can a completed harm amplify blameworthiness when every relevant decision variable is fixed? The defender of outcome luck must articulate a principle of realized agency that does not simply transfer the magnitude of suffering to the moral worth of the wrongdoer. The critic must show how actual wrongs and the claims of actual victims remain fully recognized without that amplification.

A third question concerns counterfactual responsibility. If a fortunate person would have done terrible things under harsher conditions, does the absence of occasion prevent desert or only prevent knowledge? Determining the right comparison class is difficult when character is partly constituted by the very circumstances being changed.

A fourth question is institutional. Which legal differences between attempts and completed crimes are justified by harm, proof, deterrence, and restitution, and which reflect an indefensible lottery of retributive treatment? No single criminal-law answer follows from the general moral luck debate without premises about legitimate punishment.

A fifth question concerns extending the analysis beyond familiar human agents. The Artificial Era raises questions about accountability wherever deliberative systems, human operators, institutions, and nonbiological systems participate in consequential actions. The moral luck framework can help identify control, delegated authority, causal contribution, evidential access, and reparative duties. Attributing full moral desert to any particular Artificial system would require a separately justified account of its moral agency and the applicable canonical definitions. The present dossier makes no new canonical claim about Artificial Sapiens or moral status. Its contribution is to specify a general philosophical research question that any future theory of artificial agency must address.

Philosophical Synthesis

The historical record reveals recurring attempts to locate moral value in an area secure against contingency. Epictetus concentrates ethical discipline on what is up to the agent. Kant places unconditional worth in the good will. Aristotle preserves the primacy of virtuous activity while acknowledging external conditions of flourishing. Smith identifies the powerful influence of fortune on the sentiments by which ordinary moral communities distribute gratitude and blame. Williams shows how projects, retrospective justification, and agent-regret make the life of a moral agent irreducibly historical. Nagel shows how the control required by fair desert threatens to recede when we recognize the causal and constitutive contingencies of agency.

The fourfold taxonomy explains why moral luck is a persistent philosophical problem rather than a local inconsistency about accidents. Consequences, circumstances, character, and causal antecedents affect different dimensions of moral life. The theory that treats all four as simple instances of the same evaluative mechanism lacks the distinctions needed for an adequate judgment. The theory that isolates a pure will from every contingency likewise risks failing to explain moral relationships, worldly projects, and repair.

Three conclusions follow. First, a meaningful control principle must specify the kind of evaluation to which it applies. Second, realized outcomes can change morally important facts without invariably changing the degree of culpability. Third, the fairest practices of moral assessment acknowledge unequal moral opportunities, resist unwarranted confidence about character, and distinguish coercive punishment from interpersonal acknowledgment and repair. These conclusions together constitute the basis of accountable pluralism.

The enduring value of the moral luck problem lies in its capacity to expose hidden transitions among evaluative registers. We begin by asking whether someone is blameworthy, then appeal to the gravity of harm; we begin by praising courage, then measure visible success; we begin by assessing a choice, then attribute a whole inherited character to the chooser. Philosophical analysis makes these transitions explicit and demands justification wherever one kind of assessment is made to determine another. That is a durable contribution of the debate to moral philosophy as a whole.

Conclusion

Moral luck is the problem of justified moral assessment under conditions of contingency. Its classical sources established the vulnerability of flourishing, the ethical significance of choice, and the influence of fortune on human sentiments. Williams and Nagel made the problem systematic by showing how agent-regret, retrospective justification, outcomes, circumstances, character, and causes place pressure on the ideal of luck-independent morality. Contemporary theories dispute whether control-based desert should be preserved, broadened by counterfactuals, or revised to accommodate actual consequences.

The completed analysis yields a disciplined pluralism. Culpable choice, quality of will, realized harm, moral opportunity, character, reparative duty, deserved blame, and legal responsibility are related but distinct objects of evaluation. Contingency can transform what happened, who was harmed, and what the agent now owes. Whether it can change how much blame the agent deserves remains a focused philosophical dispute requiring an account of control and desert. A coherent ethics treats agents as answerable participants in an uncertain world while holding their evaluators accountable for the reasons by which praise, blame, mercy, and punishment are assigned.