Judging someone's actions as more blameworthy or praiseworthy based on the outcome, even when the outcome was beyond their control.
Imagine two kids both throw a ball in the house. One kid's ball misses everything and bounces harmlessly. The other kid's ball breaks a vase. Even though both kids did the exact same thing, the one who broke the vase gets in way more trouble. We judge people not just by what they chose to do, but by what happened afterward — even when the result was just luck.
Moral luck describes the robust human tendency to let uncontrollable outcomes distort our moral evaluations of people. Two individuals can perform exactly the same action with the same intentions and the same degree of negligence, yet the one whose action happens to produce a bad outcome is judged far more harshly than the one who was simply lucky enough to avoid harmful consequences. This bias extends beyond resultant luck to include circumstantial luck (the situations life places you in), constitutive luck (the character traits you were born with or shaped into), and causal luck (the chain of prior events that determined your choices). The phenomenon reveals a deep tension between our stated belief that people should only be judged for what they can control and our actual practice of heavily weighting outcomes in moral evaluation.
The same glitch looks different depending on the terrain. Finance, medicine, a relationship, a team — same mechanism, different costume.
Investors and fund managers who take identical risks are evaluated very differently depending on market conditions beyond their control. Profitable trades are retrospectively labeled 'smart investing,' while identical losing trades are recharacterized as 'speculation' or 'gambling,' and the fund manager may face reputational or career consequences solely due to outcome.
Surgeons who perform identical procedures with equal skill face vastly different malpractice exposure based on patient outcomes influenced by uncontrollable biological variation. Bad outcomes trigger investigations into the physician's competence, while equally risky decisions that happen to succeed go unquestioned, distorting clinical risk culture.
Teachers whose students perform well on standardized tests are credited with excellent pedagogy, while teachers using identical methods with a cohort that happens to test poorly may be placed under remediation plans. The outcome — driven by student demographics, home environment, and test-day factors — becomes the sole basis for evaluating teacher quality.
Partners who make the same well-intentioned but imperfect decisions are judged very differently based on consequences. Forgetting to lock the door is a minor oversight if nothing happens, but becomes evidence of 'not caring about the family's safety' if a burglary occurs, creating asymmetric blame for identical behavior.
Product teams that launch features with identical risk profiles are evaluated based on user reception and market timing outside their control. A feature that happens to go viral is attributed to the team's 'brilliant product sense,' while an identical feature that flops leads to post-mortems about 'poor decision-making' and leadership changes.
Hiring decisions are evaluated based on how the new employee performs, even though performance depends heavily on team dynamics, onboarding, and market conditions. Hiring managers whose picks succeed are seen as having 'great instincts,' while those whose picks underperform — despite using identical evaluation criteria — are questioned about their judgment.
Political leaders who implement identical policies are judged almost entirely on outcomes shaped by global forces. A president whose term coincides with economic growth is hailed as competent, while one who faces a recession caused by external shocks is blamed for 'mismanagement,' and media narratives reinforce this outcome-dependent evaluation.
The concept was independently formalized by philosopher Bernard Williams (1976) and Thomas Nagel (1976) in their respective essays both titled 'Moral Luck.' Williams introduced the term, and Nagel's response articulated the four varieties (resultant, circumstantial, constitutive, and causal). Psychological investigation followed through researchers like Cushman, Young, Knobe, and others from the 2000s onward.
Punishing based on outcomes, even accidental ones, may have served an adaptive function by incentivizing caution. If harmful accidents lead to social punishment regardless of intent, individuals are motivated to take extra precautions to avoid causing harm. From a pedagogical perspective, outcome-based punishment exploits the offender's capacity to learn from consequences, reinforcing avoidance of dangerous behaviors even when harm was unintended. In ancestral environments where understanding others' true mental states was unreliable, outcomes served as the most observable and verifiable signal of risk.
Machine learning models trained on historical outcome data inherit moral luck bias by learning to associate negative outcomes with the decisions that preceded them, regardless of whether those decisions were sound at the time. Recidivism prediction algorithms, for instance, may penalize individuals whose circumstances led to worse outcomes while treating identically situated individuals who were luckier as lower risk. Reward-based reinforcement learning systems also embed moral luck by optimizing for outcomes rather than decision quality, potentially learning that identical strategies are 'good' or 'bad' based on stochastic environmental variation.
50 cards are free to preview. Buyers unlock the rest of the deck plus the interactive training — Spot-the-Bias Quiz unlimited, Swipe Deck with spaced repetition, My Blindspots, Decision Pre-Flight, the Printable Deck + Cheat Sheets, and the Field Guide e-book. $29.50$59.
