The Veil of Ignorance: Justice as a Thought Experiment

The social contract is a thought experiment, not a historical claim. It asks what rational individuals in a pre-political condition, the state of nature, would agree to, and the hypothetical agreement is what establishes political legitimacy and moral obligation. Our research file frames the whole tradition around four questions: what the state of nature is like, why anyone would leave it, what they agree to, and what limits the resulting authority. The tradition supplies the dominant framework of liberal political philosophy, the idea that political authority requires the consent of the governed.

Thomas Hobbes did not claim the state of nature had ever existed. Our file states his position plainly: it is what would obtain if civil authority collapsed, and he pointed to civil wars, failed states and the relations among sovereign nations as approximations of it. This is the correction the rest of the article stands on. The contract tradition was already hypothetical three centuries before the veil, so the veil is not a strange modern liberty taken with an old historical story. It is the tradition saying plainly what our own file says it always was: a thought experiment.

Our file separates the tradition into two strands, and the distinction is what places Rawls. Contractarianism is grounded in self-interest and asks what rational, self-interested agents would agree to; its named proponents are Hobbes and Gauthier. Contractualism is grounded in fairness and justifiability and asks what principles no one could reasonably reject; its named proponents are Kant, Rawls and Scanlon. That is why the veil is a fairness device rather than a bargaining device, and why Rawls stands in the line that runs from Kant rather than the one that runs from Hobbes.

The consent problem is the standing objection to the whole tradition, and our file attributes it to Hume. No government was ever actually founded on consent; most political authority arose from conquest, inheritance or usurpation; the social contract is a fiction, and nobody can be bound by a contract they never signed. The reply our file gives is that the contract is hypothetical: it asks what rational agents would agree to rather than what anyone did agree to, Rawls explicitly uses a hypothetical original position, and the normative force comes from the fairness of the procedure rather than from any historical act.

John Rawls revived social contract theory as a method for determining principles of justice. The original position is a hypothetical situation in which rational agents choose principles for the basic structure of society from behind a veil of ignorance, and behind the veil they do not know their race, sex, class, intelligence, talents, or their conception of the good. That is our own research file's statement of the device, and it is the claim this article is built on.

The veil is a screen, not a blackout, and this is the correction that does the most work in everything below. What it removes: the parties do not know "their place in society, class position, wealth, or social status, nor does anyone know their race, gender, fortune or misfortune in the distribution of natural assets and abilities," nor their own conception of the good, nor their religious and philosophical convictions. What it leaves: the parties possess "general facts about persons and societies, including knowledge of relatively uncontroversial scientific laws and generalizations accepted within the natural and social sciences," meaning economics, psychology and biology. They know about primary social goods, about the circumstances of justice, and about their own higher-order interests in developing the two moral powers. The popular gloss, that the parties know nothing, is wrong.

Primary goods are the currency the parties are choosing over: "Rights and liberties; powers and diverse opportunities; income and wealth; and the social bases of self-respect," described as "all-purpose social means that are necessary to the exercise and development of the moral powers." The circumstances of justice are the background conditions that make justice a question at all: objectively, "rough similarity in mental and physical faculties" and "moderate scarcity of resources"; subjectively, "limited altruism" and "the burdens of judgment" that create reasonable pluralism. Neither the list nor the circumstances appear anywhere in our own research file, and both are load-bearing later, because the list is precisely what Amartya Sen's objection attacks.

Our research file states them as follows. The equal liberty principle: each person has an equal right to the most extensive basic liberties compatible with similar liberties for all, meaning freedom of speech, conscience, association, the vote and so on. The difference principle: social and economic inequalities are arranged so that they are, first, to the greatest benefit of the least advantaged members of society, and second, attached to positions open to all under conditions of fair equality of opportunity.

Rawls states the two principles more than once and the statements are not identical. The first, at section 11 of A Theory of Justice, is one he himself presents as tentative, saying he will approximate the final statement step by step later in the book. A later statement replaces the most extensive basic liberty with a fully adequate scheme of equal basic liberties, and names the least advantaged and fair equality of opportunity explicitly. The final statement, in Justice as Fairness: A Restatement of 2001, is different again. Our file's first principle tracks the earliest phrasing and its second principle tracks a later one, presented together as a single settled pair. No wording anywhere in this article is offered as the two principles full stop. The table says which statement each is.

The two principles are ranked, and the ranking is the structural point. The first principle takes absolute priority over the second, so basic liberties cannot be sacrificed for economic gains, and within the second principle fair equality of opportunity takes priority over the difference principle. That lexical priority is what stops the theory from being a trade-off calculus, and it is the feature that most sharply separates it from the utilitarian aggregation this wing covers in The Greatest Good.

Our file states the argumentative engine in a single line: behind the veil of ignorance, rational agents would adopt maximin, maximizing the minimum outcome, because they might end up as the worst-off member of society, and this is what yields the difference principle. Our file states it flatly and raises no objection to it anywhere.

Rawls does not claim maximin is generally rational. The rule directs choosing "the alternative whose worst outcome leaves us better off than the worst outcome of all other alternatives," and he argues it is rational under three stated conditions. First, there is no secure basis for probability estimates. Second, the maximin choice is acceptable in itself, because "one cares relatively little by comparison for what is to be gained above the minimum conditions." Third, all the alternatives have worst outcomes that would be unacceptable. Rawls argues that those three conditions are met in the original position when justice as fairness is compared with utilitarianism. None of the three appears in our own research file.

Rawls also has an answer to the harder question of why the output of a thought experiment should bind anyone. It is reflective equilibrium: a method of working back and forth between considered moral judgments and general principles until they cohere, in which neither raw intuitions nor abstract principles have absolute priority, and moral theory is a process of mutual adjustment.

Rawls did not invent the device. A veil-of-ignorance style construction was already in the economics literature two decades earlier, and it produced the opposite conclusion. William Vickrey set out the shape in "Measuring Marginal Utility by Reactions to Risk" in Econometrica in 1945, and John C. Harsanyi developed it in the Journal of Political Economy in 1953 and again in 1955. In Harsanyi's construction, an impartial chooser who does not know which member of society she will be, and who assigns herself an equal chance of being any of them, will maximize average utility. Rawls, from a structurally similar starting point, derives the refutation of utilitarianism instead.

Harsanyi pressed the point directly at Rawls. In "Can the Maximin Principle Serve as a Basis for Morality? A Critique of John Rawls's Theory," in the American Political Science Review in 1975, he argued that under complete uncertainty rationality requires the principle of insufficient reason, meaning that equal probability is assigned to every position one might occupy. Do that behind the veil and the rational choice is the arrangement with the highest average utility, not the difference principle. His wider charge is that maximin is not a generally rational decision rule at all: taken seriously it would forbid ordinary risks that any reasonable person accepts, because it attends only to the worst case and ignores both how improbable that case is and how large the forgone gains are. This is the strongest standing technical objection to the argument from the original position, and our own research file does not mention it.

Maximin, on Rawls's reply, is rational in choices that are unique, irreversible and made on behalf of fundamental interests. The parties need not be psychologically risk-averse; rather it is rational to "choose conservatively to protect certain fundamental interests against loss or compromise," and gambling one's basic liberties for the chance of greater gains would be reckless given what is at stake. The original position, he holds, is not a generic decision problem, so a rule that would be foolish in an ordinary gamble can be correct here.

Robert Nozick published Anarchy, State, and Utopia in 1974. Only the minimal state is justified, on his account: protection against force, fraud and theft, and the enforcement of contracts. Any more extensive state violates individual rights, and taxation of earnings for redistributive purposes is "on a par with forced labor."

His entitlement theory of justice holds that holdings are just if they were acquired justly, that is by just acquisition and just transfer, so that historical patterns of acquisition rather than distributional outcomes determine whether a distribution is just. On that view the difference principle is unjust. Our own file renders the reason as taking from the productive to give to the unproductive; that is a polemical compression, it belongs to Nozick's argument, and this article does not state it in its own voice.

The Wilt Chamberlain argument is the counter-thought-experiment. Take a distribution that is just by whatever standard is preferred. Let people freely trade with Wilt Chamberlain, paying to watch him play; our neighbouring research file on wealth and poverty gives the figure concretely as millions of people voluntarily paying 25 cents each. The resulting distribution is unequal, and Nozick's conclusion is that it is nonetheless just, so that patterned theories of justice require continuous interference with freedom, because free exchange constantly breaks any imposed pattern.

Amartya Sen delivered "Equality of What?" as a Tanner Lecture on Human Values at Stanford University on 22 May 1979, published in the first volume of The Tanner Lectures on Human Values in 1980. He examines three candidate notions of equality, utilitarian equality, total utility equality and Rawlsian equality, and argues that none of them provides a sufficient basis for the equality aspect of morality. Against Rawls specifically: primary goods are means, and people differ in how much they can convert means into actual functioning, so equal bundles of primary goods do not deliver equal real freedom. A person with a disability, a chronic illness, or a body with higher metabolic needs requires more of the same primary goods to reach the same place. Sen's constructive proposal is basic capability equality, in which what matters is "a person being able to do certain basic things."

G. A. Cohen's Rescuing Justice and Equality, published by Harvard University Press in 2008, attacks the difference principle from the egalitarian side. Two threads. The incentives argument: Rawls permits inequalities that make the worst-off better off, and the standard justification is that talented people need higher pay as an incentive to be productive. Cohen argues that most versions of that argument fail, because the talented could simply choose to be productive without the premium, in which case the inequality is not necessary but chosen, and the citizens of a genuinely just society would not make that demand. On his reading the difference principle, taken seriously, mandates a far more equal society than Rawlsians suppose. The basic-structure restriction: Cohen argues that Rawls misapplies his own principle by confining it to the design of institutions, when personal choices made inside just institutions also determine how unequal a society ends up.

That second thread has an exact target, which our neighbouring research file on wealth and poverty states more precisely than the file behind this article: Political Liberalism, in 1993, modified the theory so that justice as fairness applies to the basic structure of society, meaning institutions, constitutions and legal systems, rather than to individual acts of charity or personal morality.

Our research file names Sandel and MacIntyre. The objection: the social contract presupposes atomistic, pre-social individuals who choose their community, but humans are constituted by their communities, we do not choose our deepest values and obligations, and so the unencumbered self behind the veil of ignorance is an incoherent abstraction. The reply our file attributes to Rawls himself: the original position is a device of representation and not a description of actual persons, it models the constraints of fairness rather than the nature of persons, and the parties behind the veil do know general facts about human nature and society. Our file gives both sides and crowns no winner.

Carole Pateman's The Sexual Contract, published in 1988, argues that the classical social contract simultaneously established political right and patriarchal right. On her reading the original contract is a fraternal pact among men to share access to women's bodies and labour, and the marriage contract, the employment contract and the prostitution contract are its modern forms. Women were never parties to the original agreement.

Charles W. Mills's The Racial Contract, published in 1997, argues that the social contract is a racial contract: an agreement among white people to categorize and subordinate non-white people, creating a system of white supremacy that structures the political, economic and epistemic order. Our file states that classical contract theorists accepted racial hierarchy, that Locke invested in the slave trade, that Kant held explicitly racist views, and that the state of nature was projected onto non-European peoples to justify colonization. The qualifier is our file's own and it is load-bearing: Mills does not reject contractualism. He argues for a non-ideal, racially aware contract theory that acknowledges actual historical domination rather than pretending the original position was racially neutral.

Our file states the exclusion problem as an objection with a reply. The objection, attributed to Pateman, Mills and feminist critics: the universal contract historically excluded women, non-white populations, Indigenous peoples, the poor and the disabled; it was a contract among propertied white men, and liberal universalism masked particular interests. The reply our file gives: the critique applies to historical contract theorists' application rather than to the method itself, Rawls's veil of ignorance explicitly strips away race, sex and class, and the contract method can be reformulated to be genuinely inclusive.

Charles W. Mills published "'Ideal Theory' as Ideology" in Hypatia in 2005. The argument is separate from The Racial Contract and aimed more directly at Rawls: the dominant post-Rawlsian practice of ideal theory, modelling a well-ordered society under conditions of full compliance, is not merely an incomplete first step but ideologically distorting, because it systematically obscures the actual histories of domination that produced existing distributions. The tension with our file's reply is real, and it sits inside our own source list. Our file answers the exclusion problem by saying the veil strips race and sex away, so the method survives. The file's own cited authority on the racial contract is on record arguing that the idealizing method is itself part of the problem.

Susan Moller Okin's Justice, Gender, and the Family, published in 1989, is a critique from inside the Rawlsian framework rather than against it. Her charge: Rawls treats the family as already just and outside the scope of the principles, and his parties in the original position were at points described as heads of households, which conceals the power relations within a household from precisely the standpoint that was supposed to see everything. Her constructive move is to extend the veil so that the parties do not know their gender either, and to argue that agents so situated would reject a gendered division of labour, authority and opportunity in the family. She called the family the first school of justice, and argued that a theory which exempts it cannot deliver what it promises.

Our corpus's file on political philosophy names four critique families against Rawls where the file behind this article names two: libertarians, that Rawls ignores property rights and self-ownership; communitarians, that the unencumbered self is a fiction; feminists, Okin in particular, that the family is a site of injustice the veil ignores; and global justice theorists, Pogge and Beitz, that justice should be global rather than limited to states. The fourth is named here and not argued. No article in this wing owns it yet.

David Gauthier's Morals by Agreement, published in 1986, derives morality from rational self-interest. Morality is a set of constraints on self-interest that self-interested agents would agree to, because cooperation produces mutual advantage, so it is a rational bargain rather than a dictate of reason or sentiment. His constrained maximization holds that rational agents restrict their self-interest when dealing with others who are similarly constrained, which produces a cooperative surplus.

T. M. Scanlon's What We Owe to Each Other, published in 1998, holds that an act is wrong if its performance would be disallowed by any set of principles for the general regulation of behaviour that no one could reasonably reject as a basis for informed, unforced general agreement. Our file contrasts it with Rawls on two axes: Scanlon's contractualism applies to interpersonal morality rather than to the basic structure of society, and its standpoint is that of individuals with reasons to accept or reject principles rather than of parties behind a veil of ignorance.

In Political Liberalism, published in 1993, Rawls argued that justice requires only an overlapping consensus among citizens holding different comprehensive doctrines, religious, philosophical and moral, and that justice must be freestanding rather than dependent on any particular worldview. Our own research file records the revision in two clauses and does not say what problem drove it.

Our corpus's political philosophy file glosses the overlapping consensus with Rawls's own phrase, political, not metaphysical. The standard account of the problem behind the revision is legitimacy and stability under reasonable pluralism: the burdens of judgment, listed among the subjective circumstances of justice, are exactly the conditions under which a free society produces durable disagreement about the good, and on that account a conception of justice presented as one comprehensive doctrine among others could not be stable under them.

One assessment belongs here and it is quoted rather than adopted. Our corpus's file on political philosophy calls A Theory of Justice "the most important work of political philosophy in the 20th century." That is our own file's judgment. It is carried here as an attributed judgment and never as a finding, and nothing in this article rests on it.

Our file is not thin. It runs to 259 lines with eight populated sections. It is thin on the one subject this article is built from: its section on Rawls is roughly 13 lines, and it contains no critical assessment of the veil or the original position as a device beyond the communitarian objection. Harsanyi is absent. Sen is absent. G. A. Cohen is absent. Okin is absent. Maximin's three stated conditions are absent, the list of primary goods is absent, and so is the fact that the parties behind the veil do know general facts about persons and societies. Everything in sections 02 through 11 that our file does not carry was found and verified outside it, and the external records are listed below.