Category: ethics
Type: thought experiment
Associated thinker(s): Bernard Williams, Thomas Nagel
Original publication or approximate period: Williams, “Moral Luck” (1976), Proceedings of the Aristotelian Society, Supplementary Volume L, 115–135; Nagel, “Moral Luck” (1979), in Mortal Questions
Difficulty: Intermediate
Themes: moral luck, responsibility, agent-regret
Related experiments: Trolley Problem; Frankfurt Cases
In Williams’s example, a lorry driver runs over a child through no fault of his own. For this retelling, assume that nothing he could reasonably have done would have prevented the death. We can compare him with an equally careful driver whose journey ends without an accident; that comparison is an editorial teaching prompt, not a verbatim paired scenario in Williams’s text.
Williams uses the first driver to point to a feeling that remains even when the driver is not to blame. Bystanders may feel sorry for him, but his own response is different: a distinctive regret about his relation to what happened. The case tests whether moral assessment and first-person feeling can be affected by outcomes beyond an agent’s control.
Can what happens, partly by chance, change how an agent should be morally assessed or how the agent should feel about their own conduct?
A prominent control-based principle says that moral assessment should depend only on what lies within the agent’s control. Yet some ordinary legal and moral practices distinguish between negligent conduct that causes death and similar conduct that narrowly avoids harm. That observation raises a question; it does not establish that outcomes alone determine blame.
The case also raises questions about outcome-sensitive legal responsibility. Broader claims about medicine or engineering require further sources and are not asserted here.
Control-based replies hold that blame and praise should track what is within an agent’s control. Williams’s discussion of agent-regret allows for a distinctive first-person response without necessarily treating the driver as blameworthy. Nagel presses the skeptical challenge that the control principle conflicts with familiar practices of moral assessment. Other writers, including Andrew Latus, question inferences from agent-regret to retrospective justification. These are selected lines of debate, not exhaustive or mutually exclusive camps.
The example does not say the driver is guilty, and Williams’s point is not that outcomes should always determine blame. Nor is “moral luck” a claim that morality is random. It names a tension between control-based principles and practices of assessment. Legal liability, blame, and an agent’s own regret are distinct and can come apart.
Nagel distinguishes constitutive luck, circumstantial luck, luck in causal antecedents, and luck in how actions or projects turn out. He treats a blameless driver’s agent-regret as distinct from moral bad luck; resultant moral luck arises in the altered case where negligence contributes to the death. State those changed premises explicitly before comparing verdicts.
Relevant passages in the complete SEP entry, Williams’s 1981 revised reprint, Nagel’s essay, and Latus’s discussion were read. Williams’s own acknowledgements and the SEP bibliography distinguish Williams’s 1976 paper from Nagel’s separate 1979 essay. The Nagel copy is a hosted reproduction with inconsistent page labels; the Williams text is a revised reprint rather than the original journal printing. The paired-driver comparison and broader applications are identified as this article’s framing, not as established historical variations. The review does not claim that the philosophical dispute is settled or that a human expert reviewed the entry.
Research check: 2026-11-01. Claim-level evidence, access limits, and unresolved issues are recorded in the source-review ledger.