The Possibility of Social Choice

A Camel, a Committee, and the Survival of a Discipline

Sen opens the lecture with a joke. "A camel," he quotes, "is a horse designed by a committee." It is meant to be a telling example of the terrible deficiencies of collective decisions — and Sen immediately turns it around. The camel is, in fact, a rather excellent solution to a difficult problem. The joke reveals more about our biases than about committees. That reversal — taking a widely accepted piece of intellectual pessimism and showing that it mistakes the failure of a model for the failure of the thing the model is trying to describe — is the central move of the entire lecture.

Amartya Sen was born in 1933 in Santiniketan, Bengal, where the poet Rabindranath Tagore had founded an unusual school and where Sen's maternal grandfather was teaching. He witnessed the Bengal famine of 1943 as a child, watching people die in the streets of Calcutta — people who, he later established, were dying not because there was no food in Bengal but because their ability to command food had evaporated. That experience never left him. It runs as a quiet undercurrent through forty years of technical work in welfare economics, a discipline that at its core asks: what does it mean for things to be better or worse for people, and how do those individual judgements add up to a social judgement?

Social choice theory is the field that takes that question seriously at the level of formal logic. It asks: given the preferences and interests of individual people, is there any consistent, democratic, and reasonable procedure for deriving a collective verdict? Can you build a rational society out of the raw material of rational individuals? For most of the twentieth century, the answer seemed to be a resounding no — because of a result proved in 1951 by a young economist named Kenneth Arrow, which appeared to close the question with mathematical finality. Sen's life work was, in large part, a patient demonstration that this conclusion was wrong — and this lecture is his account of how the field climbed back from that abyss.

Arrow's Theorem, and What It Actually Says

In 1951, Kenneth Arrow proved what is now called the General Impossibility Theorem, and it is one of the most celebrated results in all of social science. The theorem says, roughly, that there is no voting rule or social choice procedure that can simultaneously satisfy four conditions — each of which, taken alone, seems not merely reasonable but almost self-evidently required of any decent democratic system. Arrow showed that the four conditions are jointly inconsistent: you can have any three, but never all four at once.

The conditions are: Unrestricted Domain — the procedure should work for any possible combination of individual preferences, not just tidy, well-behaved ones. Pareto Efficiency — if every single person in society prefers option A to option B, then society should prefer A to B. Independence of Irrelevant Alternatives — the social ranking of A versus B should depend only on how individuals rank A against B, not on how they rank either against some third option C. And Non-Dictatorship — there should be no single individual whose preferences automatically become the social preference, regardless of what everyone else thinks.

Each condition is hard to argue against in isolation. Taken together, they are impossible to satisfy. The result hit the field of welfare economics like a thunderclap. If you cannot rationally aggregate individual preferences into social preferences without violating one of these basic requirements, then the whole project of democratic social choice — the theoretical underpinning of voting, of welfare analysis, of any claim that policy X is better for society than policy Y — seemed to be built on sand. A broad sense of pessimism settled over the discipline. Some economists concluded that welfare economics was simply not viable; others confined themselves to Pareto comparisons alone, the one thin slice of social judgement that Arrow's theorem left untouched.

The reach of Arrow's result extended beyond voting theory. Social welfare functions — economists' tools for evaluating whether one state of the world is better or worse for society than another — turned out to depend on exactly the kind of aggregation Arrow had proved impossible. The mathematical obituaries began. This, Sen argues in the lecture, was a fundamental misreading of what the theorem was actually showing.

Freedom Against Itself: The Liberal Paradox

Before turning to his diagnosis of why Arrow's result is less catastrophic than it seems, Sen has his own impossibility result to present — and in some ways it is more unsettling than Arrow's, because it does not require any complex aggregation machinery at all. It arises from a direct conflict between two things that liberal democracies prize above almost everything else: individual freedom and Pareto efficiency.

Sen's Liberal Paradox, first published in 1970, says that no social choice rule can simultaneously satisfy both a minimal condition of individual liberty and the Pareto principle. The liberty condition requires only that each person should be decisive over at least one pair of options that concern primarily their own life — the classic domain of personal freedom. The Pareto condition requires that if everyone prefers A to B, society should prefer A to B. These two conditions, which separately seem like the minimum requirements of a decent liberal society, cannot both be satisfied at once.

The Liberal Paradox and Arrow's theorem point toward the same underlying problem from different angles. Both reveal that impossibility results arise not from some deep defect in collective rationality, but from a failure to specify clearly enough what information we are permitted to use, and from an over-simple model of what individual preferences actually contain. Sen's diagnosis, developed across twenty years of technical work, is that the root cause is informational poverty — and that recognising this opens a way out.

The Information That Was Thrown Away

Arrow's framework was explicitly designed to exclude interpersonal comparisons of welfare. In the 1930s and 1940s, economists had become increasingly suspicious of such comparisons — the idea that you could say person A is better off than person B by some measurable amount seemed to rest on the ability to get inside other people's heads in a way that science could not justify. Lionel Robbins was particularly influential in arguing that there is no scientific basis for comparing one person's satisfaction against another's. Arrow built this restriction directly into the structure of his theorem: social choice could only use information about the ordinal ranking of outcomes by each individual — A is preferred to B, B is preferred to C — and nothing else. No intensities. No interpersonal comparisons.

Sen's central insight is that this restriction, far from being a neutral scientific precaution, is itself a substantive value judgement — and a deeply impoverishing one. When you are deciding whether a tax cut helps society or whether a public health intervention is worth its cost, you are inevitably making implicit comparisons between different people's gains and losses. The question is not whether such comparisons should enter social judgements, but whether they should enter explicitly and carefully or implicitly and carelessly. Forbidding them does not make social choice purer; it makes it blind.

The technical result Sen built on this insight is that once you allow even partial interpersonal comparability — not the full cardinal precision of nineteenth-century utilitarianism, but something weaker and more defensible — the impossibility results dissolve. You can construct social choice procedures that satisfy generalised versions of Arrow's conditions. The possibility of social choice is restored, not by weakening the democratic requirements but by enriching the information that the procedure is allowed to draw on. This is the move that gives the lecture its title. It is possible to say that a society choosing policy X over policy Y is doing better by its members — not with certainty, not with perfect precision, but with enough basis to act.

There is a quiet echo here of a tension that appears elsewhere in the study of human decision-making. The assumption that what people choose reveals everything relevant about their welfare — that preference and wellbeing are the same thing — is one that Amartya Sen had been challenging from the theoretical direction at the same time that psychologists were beginning to challenge it from the empirical one. What people choose in a given moment, shaped by habit, limited information, and social pressure, does not always reflect what would genuinely make their lives go well. A theory of social choice that relies only on revealed preferences is, as Sen puts it, informationally impoverished in a way that matters.

What People Can Do and Be: The Capability Approach

Once interpersonal comparisons are admitted into the framework, the next question is: comparisons of what? Utility — subjective wellbeing, satisfaction, happiness — is the traditional answer, and the utilitarian tradition is one Sen engages with seriously throughout the lecture. But utility has a deep problem as a basis for social judgement: it is adaptive. People who have lived their entire lives in conditions of deprivation, oppression, or constrained expectations often report relatively high levels of subjective satisfaction, because they have calibrated their hopes and desires to match what their circumstances allow. A woman in a society that has never permitted her to read may not report dissatisfaction with being unable to read, because she has never conceived of reading as a possibility. Measuring utility in such cases tells you about mental states; it does not tell you about the actual texture and range of someone's life.

Sen's alternative is the capability approach, developed across the 1980s and reaching maturity in the work he describes in the lecture. The relevant information for social choice, he argues, is not the utility that people extract from their circumstances but the capabilities they have — the actual freedoms to do and be things that have reason to value. Capability is a kind of practical freedom: not just the formal right to vote, but the real possibility of having one's vote count; not just the absence of a legal prohibition on working, but access to education, health, nutrition, and security that make work genuinely achievable. The distinction matters because two people with identical resources and identical formal freedoms may face radically different capability sets, depending on their physical condition, their social environment, and the history of exclusion or inclusion that has shaped their possibilities.

The capability approach transformed how development was measured. Sen was the intellectual architect of the Human Development Index, adopted by the United Nations Development Programme in 1990, which replaced the crude metric of GDP per capita with a composite measure incorporating life expectancy and education alongside income. The underlying idea — that development is fundamentally about expanding what people are able to do and be, not about the accumulation of goods — is a direct expression of the lecture's framework. Rich in income but short in life, or literate in law but constrained in practice, a society that scores well on one metric may score very poorly when the full capability set is examined.

The Possibility Was Always There

What Sen accomplished in this lecture — and in the body of work it summarises — is a rescue of a discipline from a conclusion it had accepted too readily. Arrow's impossibility theorem was real; the mathematics was beyond dispute. But the interpretation — that collective rationality is logically doomed, that social welfare economics must be abandoned or confined to the thin sliver of Pareto comparisons — was a philosophical error dressed up as a technical result. The impossibility arose because the framework had deliberately impoverished itself, stripping out the very information that makes social judgment possible and then concluding that social judgment is impossible. It was, in a way, a self-inflicted wound.

The broader lesson extends well beyond the technical literature. Every time we are told that some form of collective rationality is impossible — that there is no defensible way to compare different people's wellbeing, that any measure of inequality or poverty is irredeemably subjective, that value judgements cannot be made rigorous — we should ask what information has been excluded from the framework that is generating the impossibility. The answer is almost always: rather a lot. Partial interpersonal comparability, attention to actual capabilities rather than formal rights, sensitivity to how entitlements are distributed across a population — these are not exotic demands. They are the minimum that a serious theory of human welfare requires.

Sen's lecture is also, among other things, a defence of a certain style of doing economics — one that refuses the clean separation between positive science and normative judgment, that takes philosophy seriously as a tool for clarifying what questions can and cannot be asked, and that holds itself accountable to the lives it is ultimately trying to understand. When he moves from Arrow's conditions to the Bengal famine in the space of a single lecture, he is not changing the subject. He is following the argument to where it leads. The question of whether society can make rational collective decisions is not a puzzle for seminar rooms. It is the question that determines, in the most concrete possible sense, who eats and who does not.

Sen was awarded the prize in 1998 and remains, at ninety-one, active and publishing. His 1999 book Development as Freedom, written in the months around the Stockholm lecture, brought the capability approach to a wide general audience and remains one of the most readable works in all of development economics. The famine he witnessed as a nine-year-old boy in Calcutta in 1943 turned out to be not a scar but a compass — pointing, for the next eight decades, toward the questions that mattered most.