The investor Warren Buffett has for decades put a thought experiment to audiences of successful people. Imagine that twenty-four hours before you are born, a genie offers you the chance to design the economic and social rules of the world you will enter, on one condition: you will then draw a ticket from a barrel containing every person who will be born that day, and you will be that person. You might be born in the United States or in a country with a tenth of its income, healthy or ill, clever or slow, to parents who read to you or to none at all. Buffett calls this the ovarian lottery. His point is that people inclined to credit their position to their own merit have usually drawn a very good ticket without noticing.
The device is older than Buffett's use of it. In 1971 the philosopher John Rawls asked what principles of justice people would choose for a society if they had to choose from behind a veil of ignorance, not knowing which position in that society they would occupy, which talents they would have, or which generation they would belong to. Rawls called the distribution of native talents and family circumstances the natural lottery. He held that the outcome of that lottery is arbitrary from a moral point of view (so no one deserves their starting place) and that a just society is one the losers of the lottery could also have agreed to. I have been circling the first of those claims for eight chapters. Buffett's genie is Rawls's original position (his name for the choice made behind the veil) retold for an audience of winners. I do not know whether Buffett took it from Rawls. The resemblance is exact; both men conclude that the winner's obligations are larger than the winner's instincts suggest.
So far I have written about what advantage is, how it accumulates, where to apply it, how to build it, and when it stops. I have argued that advantage is real, that most of what makes it decisive was built or inherited rather than willed into being in the moment, and that a record of results tells us less than we think about who built what. What does any of that entitle a successful person or firm to claim, and what does it ask of them? I take the question in four parts: what advantage is for, what luck means, what to do about the fact that none of us chose our starting point, and what we owe each other.
What Advantage Is For
A firm's objective is, to a close approximation, return on the capital invested in it; every chapter so far has treated advantage as whatever raises that return. A person's objective is not given. I have put off the question until now because the answer changes everything before it.
I am a capable administrator. I ran a school's innovation activities and a group of thirty people for years, and by the ordinary measures I did it well. I also dislike meetings; administration is mostly meetings. By the chapter 2 tests, administrative capability was an alpha asset for me, since it was rare enough in a faculty and necessary to the job; it also brought the position and the invitations that chapter 4 described. It was an advantage in a contest I did not want to win. Because a capability that would prevail in a setting a person does not want to be in is not an advantage in any sense that should guide them, the inventory in chapter 6 is incomplete until it includes what the person wants the work to feel like.
Psychologists who study motivation have converged on three things that make work satisfying across cultures and occupations: autonomy, competence and relatedness (in plain terms, a person chooses the work, is good at it, and does it among people they are connected to). Competitive success delivers the second and often undermines the first and third. Someone optimizing for distinction takes the role in which they are best and most visible, which is frequently the role with the least autonomy and the most meetings.
The market, for its part, rewards distinction out of proportion to the ability behind it. The economist Sherwin Rosen showed in 1981 that where a performance can be reproduced at low cost and buyers prefer the best, small differences in ability become enormous differences in earnings. The second-best opera singer in the world earns a fraction of what the best one earns. Chapter 3 showed that records of results, income among them, are noisy measures of ability. I think income is also a noisy measure of a worthwhile life.
So the objective comes first. Before asking whether an asset is an advantage, say what the advantage is for: a particular kind of work, a level of autonomy, a set of people, a problem worth solving, and, yes, money, in whatever proportion each person sets for themselves. The worksheet in the appendix begins with that question.
Two Meanings of Luck
I have used the word luck in two senses. When the two are run together, a statistical argument turns into a moral one without anyone noticing.
The first sense is outcome luck: the variation in results, for a given level of ability, that chapter 3 called noise. Think of a venture partner's returns, a founder's exit, or a chief executive's three years of results. Outcome luck is a property of the domain and can be estimated from populations; about the person, it says only that their record should be shrunk toward the average.
The second sense is constitutive luck: the fact that a person's ability, and the circumstances that let them use it, were not chosen by them. Consider the cyclist Greg LeMond's lungs, the swimmer Michael Phelps's wingspan, the year a graduate first looks for a job, or the parents who read to them. Constitutive luck is not noise. A capability can be perfectly real, perfectly measurable and decisive in every contest, and still be unearned, in that its owner did nothing to acquire it. A free-throw shooter who makes 90 percent of their shots is skilled by any statistical test, though they did not choose their hands.
The two senses call for different responses. Outcome luck is a reason for humility about what a record proves; the response to it is the population method of chapter 3. Constitutive luck is a reason for humility about what a real advantage entitles its holder to claim.
Determinism Without Fatalism
Constitutive luck is usually discussed as a matter of degree: how much of a person's success is their own, and how much is their genes, their family and their era. I hold a stronger view, although nothing practical in this chapter depends on it.
I think a person is a physical system, and a physical system does what its state and its inputs determine. Whatever I decide, I decide because of the state of my brain at that moment, a state produced by everything that happened to my brain before, none of which I chose at the start. I have the feeling that there is a separate self behind my eyes, weighing the options and freely choosing. In my view that feeling is an illusion, installed by evolution because it makes a creature deliberate, feel accountable and learn from regret, all of which improved its ancestors' prospects. Experiments in which brain activity predicting a choice is detectable before the subject reports deciding are suggestive of this, though their interpretation is disputed. The majority position among philosophers belongs to the compatibilists, who accept determinism and still call the deliberating self free; I have never been able to see their view as anything but a redefinition of the word. Serious people disagree on this question; the scientific findings constrain the answer without settling it. Someone who rejects my view can still accept everything that follows.
Nothing that follows depends on my view because of a distinction that people routinely miss. Determinism is the claim that what happens is caused by what came before. Fatalism is the claim that what a person does makes no difference to what happens. Fatalism does not follow from determinism; it is also false. Imagine a train driver who sees a child on the track ahead. If they pull the brake, the child lives. Whether their pulling the brake was freely willed or fully determined by their history has no bearing on the child. Now suppose two men enter the cab and force the driver's hand onto the lever. The driver plainly lacks free will; the child lives all the same. Actions are links in the chain of cause and effect whether or not anyone wills them, just as each domino in a row falling in a fully determined sequence still knocks down the next.
A few studies once suggested that people primed to doubt free will behaved worse. A meta-analysis of a hundred and forty-five experiments found that the manipulations changed what people said they believed, and found no reliable evidence of downstream effects on behavior. That result fits the distinction, since people who stop believing in free will do not, in general, become fatalists.
So the practical question for a person who accepts that they did not choose their starting point is unchanged. They still have a brain that evaluates options and acts on the one it expects to value most, the evaluation still improves with information, practice and feedback, and their actions still have effects. Every prescription in chapters 5 through 8 survives; what changes is how they interpret the result.
Credit, Incentives and Desert
Once constitutive luck is admitted, three questions that usually travel together come apart: whether a person caused a good result, whether they should be rewarded for it, and whether they deserve the reward in some deeper sense.
Causal contribution is a fact, often an unknowable one (as chapter 3 said), but a fact. The salesperson did make the calls; the founder did persist when the money was short. Incentives are a separate matter. We justify them by what they will produce rather than by what happened: we pay for results because paying for results produces more of them, whether or not the results were earned in the deep sense. A society that stopped rewarding the free-throw shooter because they did not choose their hands would have fewer good free-throw shooters.
Constitutive luck undermines the third claim, desert: the claim that the successful person is entitled to their reward as a matter of justice. Most people have met the word only in the phrase "just deserts" and have assumed it was misspelled. Desert is the noun from deserve (what a person deserves) and is pronounced exactly like the course after dinner. That is no accident: both come from the French for serving. To deserve was originally to serve well enough to earn something; the dessert was what the servants brought once they had cleared the table. (The desert of the Sahara is a different word that happens to be spelled the same way; it comes from the Latin for abandoned.) Moral desert, then, is what your service has earned you. The practical issue is how much of what you have was service and how much was the ticket. If the capability was not chosen, the circumstances were not chosen, and the outcome was partly noise, then the reward is a fact about the system's incentives rather than a verdict on the person's worth.
The strongest objection to that conclusion came from Robert Nozick, another philosopher, three years after Rawls. Grant that nobody deserves their talents; it does not follow, Nozick argued, that nobody is entitled to what flows from them. A person may hold their native assets undeservedly and still have a legitimate claim to what they make with them, in the way that a person who did not earn their eyesight is still entitled to what they see. I find the objection serious but not decisive. Because it shows that the chain from unchosen talent to undeserved reward is not automatic, I keep causal contribution and entitlement under the rules separate from desert in the deep sense. A person is entitled to what the rules of their society assign to their effort; the rules in turn are justified by what they produce. What the natural lottery removes is the further claim that the rules merely recognize what the person was always owed.
I see three consequences of this view that I find useful rather than merely consoling. The first concerns judging others. A board that fires a chief executive after three years of poor relative returns, which boards routinely do, is acting on a record that chapter 3 says cannot support the judgment. The board is also acting as if the executive deserved the firing, which is a different claim again. The alternative is not to stop making decisions but to decide on the evidence that exists (the population evidence about process) and to drop the language of desert from the decision. Fire the executive if the process predicts bad results; do not pretend the three years proved anything about them. The same holds for the limited partner allocating on two fund records, and for the voter who punishes the government for a recession it did not cause, which voters reliably do.
The second concerns judging oneself. Regret is useful because it is a form of learning: recognizing that a decision produced a bad result, and resolving to decide differently, is how a brain that deliberates gets better. Self-reproach is not useful. Given everything that led to it, a past action could not have been different. Energy spent on blame is energy not spent on the next decision. For an individual, the practical content of determinism is the difference between learning from a mistake and hating oneself for it. I do not think a person needs any more content than that.
The third concerns compassion. If I had drawn a different ticket, I would be a different person with a different record. The person sleeping on the street outside my office is who I would be with a few changes in DNA and neighborhood. None of this means letting dangerous people do as they please. A computer infected with a virus is disconnected from the network, not punished; a society is entitled to protect itself from people whose decision processes produce harm. The argument is for doing so without the satisfaction of moral superiority, and for extending to the unlucky the same recognition the lucky extend to themselves.
Advantage and Everyone Else
So far I have treated advantage as a good thing for its holder. Whether it is good for anyone else depends on its source, which is usually easy to see.
An advantage from scale or learning lowers cost, and competition passes much of the saving to customers; an aluminum smelter that produces at the lowest cost in the world makes teapots and electric cars cheaper for everyone. An advantage from a better product or a proprietary process delivers something customers prefer. The patent that protects it for a period is what society pays to get the next one. An advantage from a brand is a promise about quality that customers pay for willingly. These advantages create value and share it. Most of the alpha assets in this book are of this kind. Other advantages transfer value rather than create it: a permit obtained by bribery, a regulation written by the incumbent, a network effect maintained by blocking interoperability, a scarce drug priced at what desperate patients will pay. The two tests of chapter 2 do not tell these cases apart, because a bribed permit is performance enhancing and hard to acquire. The difference is whether the customer is left better off. I think a firm that wants to be allowed to keep its alpha assets should be able to say which kind it holds.
A society faces a version of this question when it decides what to do about the natural lottery. In chapter 1 I stated two beliefs that nothing since has given me reason to retract. The first is that every person and every organization should have the chance to prove themselves and thereby earn access to further chances; that is equality of opportunity, and no society fully provides it. The second is that because so much of the inequality that determines well-being is unchosen, people acting together are obliged to reduce it where fundamental needs are at stake. Constitutive luck is the reason for both beliefs. Rawls gave the second a precise form that is consistent with paying for results. His difference principle permits inequalities, including large ones, when they work to the benefit of the least advantaged; that is, when the incentive that produces the inequality also produces something the losers of the lottery share in. An advantage from scale that lowers everyone's prices passes; an advantage from a permit obtained by bribery does not. So a society can apply the same test I proposed for a firm's alpha asset, which is whether anyone other than its holder is better off for it. The lucky are not entitled to their luck. They are responsible for what they do with it.
In chapter 1 my mother, a historian, asked what evidence I had that any of this was true. The book has made four kinds of claim, each resting on a different kind of evidence. Some were definitions: advantage, alpha asset, the two tests, the three questions about an early success. Some were mechanisms argued from logic: the flywheels, the four channels of cumulative advantage, the two conditions for disruption. Some were findings from populations: the persistence of returns, the leveling effect of the tools, the Matthew effect in grants, the management experiment in India, the arithmetic of what a record can reveal. And some were stories, including my own, which illustrate the mechanisms and prove nothing. They deserve different weights; the claims I hold with the most confidence are the ones about what cannot be known.
What remains is a short set of questions for the next decision, whether a firm's or a person's. What is the advantage for? What asset would improve expected results in this setting, and would the rivals in this setting find it hard to acquire? Who holds the complements, and who will capture the return? What does the record actually show, given the noise and the number of outcomes, and what does the population say that the record cannot? Which of the assets travel, and which belong to the position? What would make this stop working, and what will I watch for? And when it works, what part of this was mine, and what do I owe for the rest?
A rider born with exceptional lungs did not choose them; they still have to decide how to train, which races to enter, and whom to ask for help. The rest of us face the same decisions with less convenient finish lines. Like that rider, we start with what we were given and decide what to build from it.
Notes
Buffett calls this the ovarian lottery Buffett has told the ovarian-lottery thought experiment on many occasions; a published version appears in Carol J. Loomis, Tap Dancing to Work: Warren Buffett on Practically Everything, 1966–2012 (New York: Portfolio/Penguin, 2012).
In 1971 the philosopher John Rawls asked John Rawls, A Theory of Justice (Cambridge, Mass.: Harvard University Press, 1971; revised edition 1999). The original position and the veil of ignorance are set out in chapter 1, sections 3 and 4; the "natural lottery" and the judgment that its outcome is "arbitrary from a moral point of view" are in chapter 2, section 17; the difference principle in sections 12 and 13.
The strongest objection to that conclusion came from Robert Nozick Robert Nozick, Anarchy, State, and Utopia (New York: Basic Books, 1974), chapter 7, section II, on Rawls's treatment of natural assets; the argument that the undeservedness of a talent does not entail the undeservedness of what flows from it is developed there.
Psychologists who study motivation have converged on three things Richard M. Ryan and Edward L. Deci, "Self-Determination Theory and the Facilitation of Intrinsic Motivation, Social Development, and Well-Being," American Psychologist 55, no. 1 (2000): 68–78.
Sherwin Rosen showed in 1981 Sherwin Rosen, "The Economics of Superstars," American Economic Review 71, no. 5 (1981): 845–858.
Experiments in which brain activity predicting a choice is detectable before the subject reports deciding Benjamin Libet, Curtis A. Gleason, Elwood W. Wright, and Dennis K. Pearl, "Time of Conscious Intention to Act in Relation to Onset of Cerebral Activity (Readiness-Potential)," Brain 106, no. 3 (1983): 623–642. For the case that such findings imply the absence of free will, Robert M. Sapolsky, Determined: A Science of Life Without Free Will (New York: Penguin Press, 2023); for the case that they do not, Alfred R. Mele, Free: Why Science Hasn't Disproved Free Will (New York: Oxford University Press, 2014).
the compatibilists Daniel C. Dennett, Elbow Room: The Varieties of Free Will Worth Wanting (Cambridge, Mass.: MIT Press, 1984), is the classic statement.
A board that fires a chief executive after three years of poor relative returns, which boards routinely do Dirk Jenter and Fadi Kanaan, "CEO Turnover and Relative Performance Evaluation," Journal of Finance 70, no. 5 (2015): 2155–2184, find that chief executives are significantly more likely to be dismissed after poor industry- and market-wide performance that they did not cause.
the voter who punishes the government for a recession it did not cause, which voters reliably do Christopher H. Achen and Larry M. Bartels, Democracy for Realists: Why Elections Do Not Produce Responsive Government (Princeton: Princeton University Press, 2016), chapters 5 and 6, on retrospective voting and blind retrospection.