Robert Paul Wolff, In Defense of Anarchism (1970)

The Solution of Classical Democracy

1. Democracy Is the Only Feasible Solution

It is not necessary to argue at length the merits of all the various types of state which, since Plato, have been the standard fare of political philosophies. John Locke may have found it worthwhile to devote an entire treatise to Sir Robert Filmer's defense of the hereditary rights of kings, but today the belief in all forms of traditional authority is as weak as the arguments which can be given for it. There is only one form of political community which offers any hope of resolving the conflict between authority and autonomy, and that is democracy.

The argument runs thus: men cannot be free so long as they are subject to the will of others, whether one man (a monarch) or several (aristocrats). But if men rule themselves, if they are both law-givers and law-obeyers, then they can combine the benefits of government with the blessings of freedom. Rule for the people is merely benevolent slavery, but rule by the people is true freedom. Insofar as a man participates in the affairs of state, he is ruler as well as ruled. His obligation to submit to the laws stems not from the divine right of the monarch, nor from the hereditary authority of a noble class, but from the fact that he himself is the source of the laws which govern him. Therein lies the peculiar merit and moral claim of a democratic state.

Democracy attempts a natural extension of the duty of autonomy to the realm of collective action. Just as the truly responsible man gives laws to himself, and thereby binds himself to what he conceives to be right, so a society of responsible men can collectively bind themselves to laws collectively made, and thereby bind themselves to what they have together judged to be right. The government of a democratic state is then, strictly speaking, no more than a servant of the people as a whole, charged with the execution of laws which have been commonly agreed upon. In the words of Rousseau, "every person, while uniting himself with all, ... obey[s] only himself and remain[s] as free as before" (Social Contract, Bk. I, Ch. 6).

Let us explore this proposal more closely. We shall begin with the simplest form of democratic state, which may be labeled unanimous direct democracy.

2. Unanimous Direct Democracy

There is, in theory, a solution to the problem which has been posed, and this fact is in itself quite important. However, the solution requires the imposition of impossibly restrictive conditions which make it applicable only to a rather bizarre variety of actual situations. The solution is a direct democracy -- that is, a political community in which every person votes on every issue -- governed by a rule of unanimity. Under unanimous direct democracy, every member of the society wills freely every law which is actually passed. Hence, he is only confronted as a citizen with laws to which he has consented. Since a man who is constrained only by the dictates of his own will is autonomous, it follows that under the directions of unanimous direct democracy, men can harmonize the duty of autonomy with the commands of authority.

It might be argued that even this limiting case is not genuine, since each man is obeying himself, and hence is not submitting to a legitimate authority. However, the case is really different from the prepolitical (or extrapolitical) case of self-determination, for the authority to which each citizen submits is not that of himself simply, but that of the entire community taken collectively. The laws are issued in the name of the sovereign, which is to say the total population of the community. The power which enforces the law (should there be any citizen who, having voted for a law, now resists its application to himself) is the power of all, gathered together into the police power of the state. By this means, the moral conflict between duty and interest which arises from time to time within each man is externalized, and the voice of duty now speaks with the authority of law. Each man, in a manner of speaking, encounters his better self in the form of the state, for its dictates are simply the laws which he has, after due deliberation, willed to be enacted.

Unanimous direct democracy is feasible only so long as there is substantial agreement among all the members of a community on the matters of major importance. Since by the rule of unanimity a single negative vote defeats any motion, the slightest disagreement over significant questions will bring the operations of the society to a halt. It will cease to function as a political community and fall into a condition of anarchy (or at least into a condition of non-legitimacy; a de facto government may of course emerge and take control). However, it should not be thought that unanimous direct democracy requires for its existence a perfect harmony of the interests or desires of the citizens. It is perfectly consistent with such a system that there be sharp, even violent, oppositions within the community, perhaps of an economic kind. The only necessity is that when the citizens come together to deliberate on the means for resolving such conflicts, they agree unanimously on the laws to be adopted.6

For example, a community may agree unanimously on some principles of compulsory arbitration by which economic conflicts are to be settled. An individual who has voted for these principles may then find himself personally disadvantaged by their application in a particular case. Thinking the principles fair, and knowing that he voted for them, he will (hopefully) acknowledge his moral obligation to accept their operation even though he would dearly like not to be subject to them. He will recognize the principles as his own, just as any of us who has committed himself to a moral principle will, uncomfortably to be sure, recognize its binding force upon him even when it is inconvenient. More precisely, this individual will have a moral obligation to obey the commands of the mediation board or arbitration council, whatever it decides, because the principles which guide it issue from his own will. Thus the board will have authority over him (i.e., a right to be obeyed) while he retains his moral autonomy.

Under what circumstances might a unanimous direct democracy actually function for a reasonable period of time without simply coming to a series of negative decisions? The answer, I think, is that there are two sorts of practical unanimous direct democracies. First, a community of persons inspired by some all-absorbing religious or secular ideal might find itself so completely in agreement on the goals of the community and the means for achieving them that decisions could be taken on all major questions by a method of consensus. Utopian communities in the nineteenth century and some of the Israeli kibbutzim in the twentieth are plausible instances of such a functioning unanimity. Eventually, the consensus dissolves and factions appear, but in some cases the unanimity has been preserved for a period of many years.

Second, a community of rationally self-interested individuals may discover that it can only reap the fruits of cooperation by maintaining unanimity. So long as each member of the community remains convinced that the benefits to him from cooperation -- even under the conditions of compromise imposed by the need for unanimity -- outweigh the benefits of severing his connection with the rest, the community will continue to function. For example, a classical laissez-faire economy ruled by the laws of the marketplace is supposedly endorsed by all the participants because each one recognizes both that he is better off in the system than out and that any relaxation of the ban against arrangements in restraint of trade would in the end do him more harm than good. So long as every businessman believes these two propositions, there will be unanimity on the laws of the system despite the cutthroat competition.7

As soon as disagreement arises on important questions, unanimity is destroyed and the state must either cease to be de jure or else discover some means for settling disputed issues which does not deprive any member of his autonomy. Furthermore, when the society grows too large for convenience in calling regular assemblies, some way must be found to conduct the business of the state without condemning most of the citizens to the status of voiceless subjects. The traditional solutions in democratic theory to these familiar problems are of course majority rule and representation. Our next task, therefore, is to discover whether representative majoritarian democracy preserves the autonomy which men achieve under a unanimous direct democracy.

Since unanimous democracy can exist only under such limited conditions, it might be thought that there is very little point in discussing it at all. For two reasons, however, unanimous direct democracy has great theoretical importance. First, it is a genuine solution to the problem of autonomy and authority, and as we shall see, this makes it rather unusual. More important still, unanimous direct democracy is the (frequently unexpressed) ideal which underlies a great deal of classical democratic theory. The devices of majoritarianism and representation are introduced in order to overcome obstacles which stand in the way of unanimity and direct democracy. Unanimity is clearly thought to be the method of making decisions which is most obviously legitimate; other forms are presented as compromises with this ideal, and the arguments in favor of them seek to show that the authority of a unanimous democracy is not fatally weakened by the necessity of using representation or majority rule. One evidence of the theoretical primacy of unanimous direct democracy is the fact that in all social contract theories, the original collective adoption of the social contract is always a unanimous decision made by everyone who can later be held accountable to the new state. Then the various compromise devices are introduced as practical measures, and their legitimacy is derived from the legitimacy of the original contract. The assumption that unanimity creates a de jure state is usually not even argued for with any vigor; it seems to most democratic theorists perfectly obvious.

3. Representative Democracy

Although the problem of disagreement is the more immediate, I shall deal first with the difficulties of assembly which lead -- in democratic theory -- to the device of a representative parliament.8 There are two problems which are overcome by representation: first, the total citizenry may be too numerous to meet together in a chamber or open field; and second, the business of government may require a continuous attention and application which only the idle rich or the career politician can afford to give it.

We may distinguish a number of types of representation, ranging from the mere delegation of the right to vote a proxy to a complete turning over of all decision-making functions. The question to be answered is whether any of these forms of representation adequately preserve the autonomy which men exercise through decisions taken unanimously by the entire community. In short, should a responsible man commit himself to obey the laws made by his representatives?

The simplest sort of representation is strict agency. If I am unable to attend the assembly at which votes are taken, I may turn over my proxy to an agent with instructions as to how to vote. In that case, it is obvious that I am as obligated by the decisions of the assembly as though I had been physically present. The role of legal agent is too narrowly drawn, however, to serve as an adequate model for an elected representative. In practice, it is impossible for representatives to return to their districts before each vote in the assembly and canvass their constituents. The citizens may of course arm their representative with a list of their preferences on future votes, but many of the issues which come before the assembly may not have been raised in the community at the time the representative was chosen. Unless there is to be a recall election on the occasion of each unforeseen deliberation, the citizens will be forced to choose as their representative a man whose general "platform" and political bent suggests that he will, in the future, vote as they imagine they would themselves, on issues which neither the citizens nor the representative yet have in mind.

When matters have reached this degree of removal from direct democracy, we may seriously doubt whether the legitimacy of the original arrangement has been preserved. I have an obligation to obey the laws which I myself enact. I have as well an obligation to obey the laws which are enacted by my agent in strict accord with my instructions. But on what grounds can it be claimed that I have an obligation to obey the laws which are made in my name by a man who has no obligation to vote as I would, who indeed has no effective way of discovering what my preferences are on the measure before him? Even if the parliament is unanimous in its adoption of some new measure, that fact can only bind the deputies and not the general citizenry who are said to be represented by them.

It can be replied that my obligation rests upon my promise to obey, and that may in fact be true. But insofar as a promise of that sort is the sole ground of my duty to obey, I can no longer be said to be autonomous. I have ceased to be the author of the laws to which I submit and have become the (willing) subject of another person. Precisely the same answer must be given to the argument that good effects of some sort will result from my obeying the duly elected parliament. The moral distinction of representative government, if there is any, does not lie in the general good which it does, nor in the fact that its subjects have consented to be ruled by a parliament. Benevolent elective kingship of a sort which has existed in past societies can say as much. The special legitimacy and moral authority of representative government is thought to result from its being an expression of the will of the people whom it rules. Representative democracy is said not simply to be government for the people but also government (indirectly) by the people. I must obey what the parliament enacts, whatever that may be, because its will is my will, its decisions my decisions, and hence its authority merely the collected authority of myself and my fellow citizens. Now, a parliament whose deputies vote without specific mandate from their constituents is no more the expression of their will than is a dictatorship which rules with kindly intent but independently of its subjects. It does not matter that I am pleased with the outcome after the fact, nor even that my representative has voted as he imagines I would have liked him to. So long as I do not, either in person or through my agent, join in the enactment of the laws by which I am governed, I cannot justly claim to be autonomous.

Unfounded as is traditional representative government's claim to the mantle of legitimacy, it seems impeccable in comparison with the claims of the form of "democratic" politics which actually exist in countries like the United States today. Since World War II, governments have increasingly divorced themselves in their decision-making from anything which could be called the will of the people. The complexity of the issues, the necessity of technical knowledge, and most important, the secrecy of everything having to do with national security, have conspired to attenuate the representative function of elected officials until a point has been reached which might be called political stewardship, or, after Plato, "elective guardianship." The President of the United States is merely pledged to serve the unspecified interests of his constituents in unspecified ways.

The right of such a system to the title of democracy is customarily defended by three arguments: first, the rulers are chosen by the people from a slate which includes at least two candidates for each office; second, the rulers are expected to act in what they conceive to be the interest of the people; and third, the people periodically have the opportunity to recall their rulers and select others. More generally, the system allows individuals to have some measurable influence on the ruling elite if they choose. The genealogy of the term "democracy" need not concern us. It suffices to note that the system of elective guardianship falls so far short of the ideal of autonomy and self-rule as not even to seem a distant deviation from it. Men cannot meaningfully be called free if their representatives vote independently of their wishes, or when laws are passed concerning issues which they are not able to, understand. Nor can men be called free who are subject to secret decisions, based on secret data, having unannounced consequences for their well-being and their very lives.

Some while after John Kennedy was assassinated, several memoirs appeared recounting the inside story of the decisions to invade Cuba in 1961 and to risk a nuclear war by blockading Cuba in 1962. More recently, with the advent of the Nixon Administration, we have begun to learn something of the way in which President Johnson and his advisers committed this country to a massive land war in Vietnam. As this book is being prepared for publication, new decisions are being taken in secret which may involve the United States in the Laotian situation.

In none of these instances of major decisions is there the slightest relation between the real reasons determining official policy and the rationale given out for public consumption. In what way, it may be wondered, are Americans better off than those Russian subjects who were allowed, by Khrushchev's decision, to know a bit of the truth about Stalin?

Even those forms of representative government which approximate to genuine agency suffer from a curious and little-noted defect which robs electors of their freedom to determine the laws under which they shall live. The assumption which underlies the practice of representation is that the individual citizen has an opportunity, through his vote, to make his preference known. Leaving aside for the moment the problems connected with majority rule, and ignoring as well the derogations from legitimacy which result when issues are voted on in the parliament which were not canvassed during the election of deputies, the citizen who makes use of his ballot is, as it were, present in the chamber through the agency of his representative. But this assumes that at the time of the election, each man had a genuine opportunity to vote for a candidate who represented his point of view. He may find himself in the minority, of course; his candidate may lose. But at least he has had his chance to advance his preferences at the polls.

But if the number of issues under debate during the campaign is greater than one or two, and if there are -- as there are sure to be -- a number of plausible positions which might be taken on each issue, then the permutations of consistent alternative total "platforms" will be vastly greater than the number of candidates. Suppose, for example, that in an American election there are four issues: a farm bill, medical care for the aged, the extension of the draft, and civil rights. Simplifying the real world considerably, we can suppose that there are three alternative courses of action seriously being considered on the first issue, four on the second, two on the third, and three on the last. There are then 3 X 4 X 2 X 3 = 72 possible stands which a man might take on these four issues. For example, he might favor full parity, Kerr-Mills, discontinuation of the draft, and no civil rights bill; or free market on agricultural produce, no medicare at all, extension of the draft, and a strong civil rights bill; and so on. Now, in order to make sure that every voter has a chance of voting for what he believes, there would have to be 72 candidates, each holding one of the logically possible positions. If a citizen cannot even find a candidate whose views coincide with his own, then there is no possibility at all that he will send to the parliament a genuine representative. In practice, voters are offered a handful of candidates and must make compromises with their beliefs before they ever get to the polls. Under these circumstances, it is difficult to see what content there is to the platitude that elections manifest the will of the people.

The most biting rejection of representative democracy can be found in Rousseau's Social Contract. In opposition to such writers as Locke, Rousseau writes:

Sovereignty cannot be represented for the same reason that it cannot be alienated; its essence is the general will, and that will must speak for itself or it does not exist: it is either itself or not itself: there is no intermediate possibility. The deputies of the people, therefore, are not and cannot be their representatives; they can only be their commissioners, and as such are not qualified to conclude anything definitively. No act of theirs can be a law, unless it has been ratified by the people in person; and without that ratification nothing is a law. The people of England deceive themselves when they fancy they are free; they are so, in fact, only during the election of members of parliament: for, as soon as a new one is elected, they are again in chains, and are nothing. And thus, by the use they make of their brief moments of liberty, they deserve to lose it (Bk. Ill, Ch. 15).

Appendix: A Proposal for Instant Direct Democracy

The practical impossibility of direct democracy is generally taken for granted in contemporary discussions of democratic theory, and it is accounted an unpleasantly Utopian aspect of the philosophy of Rousseau, for example, that it assumes a community in which every citizen can vote directly on all the laws. Actually, the obstacles to direct democracy are merely technical, and we may therefore suppose that in this day of planned technological progress it is possible to solve them. The following proposal sketches one such solution. It is meant a good deal more than half in earnest, and I urge those readers who are prone to reject it out of hand to reflect on what that reaction reveals about their real attitude toward democracy.

I propose that in order to overcome the obstacles to direct democracy, a system of in-the-home voting machines be set up. In each dwelling, a device would be attached to the television set which would electronically record votes and transmit them to a computer in Washington. (Those homes without sets would be supplied by a federal subsidy. In practice this would not be very expensive, since only the very poor and the very intelligent lack sets at present.) In order to avoid fraudulent voting, the device could be rigged to record thumbprints. In that manner, each person would be able to vote only once, since the computer would automatically reject a duplicate vote. Each evening, at the time which is now devoted to news programs, there would be a nationwide all-stations show devoted to debate on the issues before the nation. Whatever bills were "before the Congress" (as we would now describe it) would be debated by representatives of alternative points of view. There would be background briefings on technically complex questions, as well as formal debates, question periods, and so forth. Committees of experts would be commissioned to gather data, make recommendations for new measures, and do the work of drafting legislation. One could institute the position of Public Dissenter in order to guarantee that dissident and unusual points of view were heard. Each Friday, after a week of debate and discussion, a voting session would be held. The measures would be put to the public, one by one, and the nation would record its preference instantaneously by means of the machines. Special arrangements might have to be made for those who could not be at their sets during the voting. (Perhaps voting sessions at various times during the preceding day and night.) Simple majority rule would prevail, as is now the case in the Congress.

The proposal is not perfect, of course, for there is a great difference between the passive role of listener in a debate and the active role of participant. Nevertheless, it should be obvious that a political community which conducted its business by means of "instant direct democracy" would be immeasurably closer to realizing the ideal of genuine democracy than we are in any so-called democratic country today. The major objection which would immediately be raised to the proposal, particularly by American political scientists, is that it would be too democratic! What chaos would ensue! What anarchy would prevail! The feckless masses, swung hither and yon by the winds of opinion, would quickly reduce the great, slow-moving, stable government of the United States to disorganized shambles! Bills would be passed or unpassed with the same casual irresponsibility which now governs the length of a hemline or the popularity of a beer. Meretricious arguments would delude the simple, well-meaning, ignorant folk into voting for pie-in-the-sky giveaways; foreign affairs would swing between jingoist militarism and craven isolationism. Gone would be the restraining hand of wisdom, knowledge, tradition, experience.

The likelihood of responses of this sort indicates the shallowness of most modern belief in democracy. It is obvious that very few individuals really hold with government by the people, though of course we are all willing to obliterate ourselves and our enemies in its name. Nevertheless, the unbelievers are, in my opinion, probably wrong as well as untrue to their professed faith. The initial response to a system of instant direct democracy would be chaotic, to be sure. But very quickly, men would learn -- what is now manifestly not true -- that their votes made a difference in the world, an immediate, visible difference. There is nothing which brings on a sense of responsibility so fast as that awareness. America would see an immediate and invigorating rise in interest in politics. It would hardly be necessary to launch expensive and frustrating campaigns to get out the vote. Politics would be on the lips of every man, woman, and child, day after day. As interest rose, a demand would be created for more and better sources of news. Even under the present system, in which very few Americans have any sense of participation in politics, news is so popular that quarter-hour programs are expanded to half an hour, and news specials preempt prime television time. Can anyone deny that instant direct democracy would generate a degree of interest and participation in political affairs which is now considered impossible to achieve?

Under a system of genuine democracy the voices of the many would drown out those of the few. The poor, the uneducated, the frightened who today are cared for by the state on occasion but never included in the process of government would weigh, man for man, as heavily as the rich, the influential, the well-connected. Much might be endangered that is worthwhile by such a system, but at least social justice would flourish as it has never flourished before.

If we are willing to think daringly, then, the practical obstacles to direct democracy can be overcome. For the moment, we need not discuss any further whether we wish to overcome them; but since our investigation concerns the possibility of establishing a state in which the autonomy of the individual is compatible with the authority of the state, I think we can take it that the difficulties which in the past have led to unsatisfactory forms of representative democracy do not constitute a serious theoretical problem.

4. Majoritarian Democracy

The principal theoretical weakness of unanimous direct democracy is its requirement that decisions be taken unanimously in order for them to acquire the authority of law. As a practical matter, of course, this requirement severely limits the actual situations in which a state can flourish, but it is perhaps an even more serious failing of unanimous democracy that it offers no way at all for men of good will to resolve their differences. Presumably, in order for the concept of a just state to have more than idle interest, it must at least in theory be possible for conflicts to be resolved without a loss of autonomy on the part of the citizens or of authority on the part of the state. The conflicts need not be motivated by divisive self-interest; they may simply be disagreements over the best way to pursue the common good.

The solution which immediately springs to the fore is, of course, majority rule. Where the electorate are divided, take a vote; give to each man one vote, and let the group as a whole be committed by the preponderance of voices. So widespread is the belief in majority rule that there is not a single variant of democratic theory which does not call upon it as the means for composing differences and arriving at decisions. Our task is to discover an argument which demonstrates that the autonomy of unanimous democracy is preserved in a democracy which is guided by the rule of the majority. In other words, we must inquire whether the members of a democratic polity are morally bound to obey the decisions of the majority, and if so, why.

The problem, of course, concerns those who find themselves in the minority on any question. The members of the majority bear the same relation to the law they have passed as do all the citizens in a unanimous democracy. Since the majority have willed the law, they are bound by it, and they remain autonomous in submitting to its authority. A member of the minority, however, has voted against the law, and he appears to be in the position of a man who, deliberating on a moral question, rejects an alternative only to find it forced upon him by a superior power. His readiness to deliberate, and to be committed by his decision, manifests his desire to be autonomous; but insofar as he must submit to the will of the majority, it seems that his desire is frustrated.

One common justification of majority rule is that, on prudential or general moral grounds, it works better than any other system which has been devised. For example, it is said that democratic politics is a substitute for the rule of arms which prevails in lawless societies. Since the majority are, militarily speaking, likely to be the superior body, they must be allowed to rule by the ballot; for otherwise they will resort to force and throw society back into chaos. Or, again, historical observation may reveal that rule by the majority tends to advance the general welfare better than any other system of government (such as rule by the wise or the powerful), since contrary to what Plato and others have supposed, the people know their own interest best. Majoritarian democracy, it is said, is therefore the most effective safeguard against the rule of a hypocritically self-interested elite. From the point of view of the individual, it might be urged that submission to the rule of the majority offers him the best chance, in the long run, for advancing his own interests, since by and large he will find himself in the majority as often as in the minority, and the benefit flowing from collective action will outweigh the losses suffered when his side loses.

All such defenses, and others besides which might be based on considerations of interest or good consequences, are, however, strictly irrelevant to our inquiry. As justifications for an individual's autonomous decision to cooperate with the state, they may be perfectly adequate; but as demonstrations of the authority of the state -- as proofs, that is, of the right of the state to command the individual and of his obligation to obey, whatever may be commanded -- they fail completely. If the individual retains his autonomy by reserving to himself in each instance the final decision whether to cooperate, he thereby denies the authority of the state; if, on the other hand, he submits to the state and accepts its claim to authority, then so far as any of the above arguments indicate, he loses his autonomy.

Indeed, the prudential and casuistical defenses of democracy do not succeed in distinguishing it morally from any other form of political community. A man might find that his affairs flourished in a dictatorship or monarchy, and even that the welfare of the people as a whole was effectively advanced by the policies of such a state. Democracy, then, could claim to be no more than one type of de facto government among many, and its virtues, if any, would be purely relative. Perhaps, as Winston Churchill once remarked, democracy is the worst form of government except for all the others; but if so, then the "citizens" of America are as much subjects of an alien power as the Spaniards under Franco or the Russians under Stalin. They are merely more fortunate in their rulers.

A more serious case for majority rule can be founded on the terms of the contract by which the political order is constituted. According to many theorists of democracy, the transition from unanimous rule, as exemplified by the adoption of the social contract, to majority rule, on which the subsequent functionings of the society depend, is provided for by a clause in the original agreement. Everyone pledges himself henceforth to abide by the rule of the majority, and whenever a citizen objects to being required to obey laws for which he has not voted, he can be recalled to his promise. On that pact, it is asserted, rests the moral authority of a majoritarian state.9

But this argument is no better than the previous one. A promise to abide by the will of the majority creates an obligation, but it does so precisely by giving up one's autonomy. It is perfectly possible to forfeit autonomy, as we have already seen. Whether it is wise, or good, or right to do so is, of course, open to question, but that one can do so is obvious. Hence, if citizens contract to govern themselves by majority rule, they thereby obligate themselves in just the manner that they would be obligated by any promise. The state then has a right to command them, assuming that it is guided only by the majority. But the citizens have created a legitimate state at the price of their own autonomy! They have bound themselves to obey laws which they do not will, and indeed even laws which they vigorously reject. Insofar as democracy originates in such a promise, it is no more than voluntary slavery, and the characterization which Rousseau gives of the English form of representation can as well be applied here.

The force of this point is difficult to grasp, for we are so deeply imbued with the ethic of majoritarianism that it possesses for us the deceptive quality of self-evidence. In the United States, little children are taught to let the majority rule almost before they are old enough to count the votes. Whenever force or wealth threatens to dominate a situation, the voice of the majority is appealed to as the higher call of morality and reason. Not rule by the majority? What else is there, one wants to ask. Perhaps it will help, therefore, to reflect that the justification of majority rule by appeal to an original promise opens the way to justification of virtually any other mode of decision-making, for the contracting citizens could as well have promised to abide by minority rule, or random choice, or the rule of a monarch, or rule by the best educated, or rule by the least educated, or even rule by a daily dictator chosen by lot.

If the only argument for majority rule is its legitimation by unanimous vote at the founding convention, then presumably any method of decision-making at all which was given that sanction would be equally legitimate. If we hold that majority rule has some special validity, then it must be because of the character of majority rule itself, and not because of a promise which we may be thought to have made to abide by it. What is required, therefore, is a direct justification of majority rule itself, that is, a demonstration that under majority rule the minority do not forfeit their autonomy in submitting to the decisions of the collectivity.

John Locke somewhat recognizes the necessity for a proof of the principle of majority rule, and at the very outset of his Second Treatise Concerning Civil Government offers the following:

When any number of men have so consented to make one community or government, they are thereby presently incorporated, and make one body politic, wherein the majority have a right to act and conclude the rest. For when any number of men have, by the consent of every individual, made a community, they have thereby made that community one body, with a power to act as one body, which is only by the will and determination of the majority. For that which acts [i.e., activates] any community being only the consent of the individuals of it, and it being one body must move one way, it is necessary the body should move that way whither the greater force carries it, which is the consent of the majority; or else it is impossible it should act or continue one body, one community, which the consent of every individual that united into it agreed that it should; and so every one is bound by that consent to be concluded by the majority (Ch. VIII).
The key to the argument is the assertion that the body politic must be carried "whither the greater force carries it." If this means that the state must in fact move in the direction of the preponderance of power, it is either trivially true, power being defined by its effects, or else nontrivial and false, since frequently a minority can dominate the conduct of public affairs even though they command far less than a preponderance of the available force in the society. On the other hand, if Locke means that the state ought to move in the direction of the greater moral force, then presumably he believes that the majority will possess that superior moral force because each individual counts for one in the moral calculus. However, even if sense can be made of the notion of a moral force, we are still without a reason why the minority has an obligation to obey the majority.

One possible line of argument is to found the rule of the majority on the higher principle that each person in the society should have an equal chance to make his preferences the law. Assuming for the moment that the principle of equal chance is valid, does majority rule achieve that equality?

It is difficult to decide, since the notion of having an equal chance of making one's preferences law is ambiguous. In one sense, majority rule guarantees to the members of the majority that their preference will become law. Hence if a man knows that he is in the minority, he will realize that he has no chance at all of effecting his will. This is the characteristic of majoritarian democracy which drives permanent minorities into rebellion, and permits what Mill quite justly called the tyranny of the majority. A system of legislation by lot might therefore be more in accord with the principle of equal chance. Each individual could write his preference on a piece of paper, and the winning law could be drawn from a twirling basket. Then, we might suppose, each citizen could have exactly the same chance that his will would become law. But probability is a tricky science, and here again we must pause to reconsider. Each citizen, to be sure, would have the same chance for his piece of paper to be drawn from the basket; but presumably what he desires is simply that the law which he prefers be enacted, not that the enactment take place by means of his personal slip of paper. In other words, he would be equally satisfied by a drawing of any piece of paper on which his preference was written. Now, if there are more slips with alternative A on them than with alternative B, then of course the probability is higher of alternative A being chosen. Thus, legislation by lot would offer some chance to the minority, unlike rule by the majority, but it would not offer to each citizen an equal chance that his preference be enacted. Nevertheless, it does seem to come closer to the ideal of equal chances than majority rule.

We have cited the device of decision by random choice chiefly as a way of exposing the weaknesses of a certain justification of majority rule, but before going on to yet another argument for majoritarianism, it might be well to consider whether random decision is a worthy candidate for adoption in its own right. Is it reasonable to resolve differences of opinion by chance? Does commitment to such a device preserve the autonomy of the individual citizen, even when the die is cast against him?

We must not be too hasty in rejecting the appeal to chance, for in at least some situations of choice it would appear to be the proper method. For example, if I am faced with a choice among alternatives whose probable outcomes I cannot estimate, then it is perfectly sensible to let chance decide my choice. If I am lost in the forest, with not the slightest idea which direction is most promising, and if I am convinced that my best chance is to choose one path and stick to it, then I might as well spin myself around with my eyes closed and start off in any direction. More generally, it is reasonable to choose at random among equally promising alternatives.10 Random decision is also reasonable in another sort of case, where rewards or burdens are to be distributed among equally deserving (or undeserving) citizens, and the nature of the item to be distributed makes it impossible to divide it and parcel out equal shares. Thus, if the armed forces require only one-half of the available men, and cannot adjust matters by halving the service time and doubling the draft, then the fair method of choosing inductees is to put the names in a bowl and pull them out at random.

Since the duty of autonomy dictates only that I use all available information in making my decisions, it is clear that randomization in the face of ignorance is not a derogation of autonomy. This is equally true in the second case, of indivisible payoffs, though we are there obligated to attempt to overcome the inevitable unfairness by incorporating the matter into a broader context and balancing off future rewards and burdens. It follows that the use of random devices in some collective decision will not violate aunomy, assuming for the moment that there has been unaninious agreement on their adoption. But what shall we say of the decision by lot in cases where the obstacle to decision is simple disagreement among the members of the assembly, and not ignorance of future outcomes or the indivisibility of payoffs? Is this, perhaps, a solution to the problem of the subjection of the minority?

In the making of individual decisions, an appeal to chance when the necessary information was at hand would be a willful forfeiture of autonomy. May we then conclude that the same is true for collective decision? Not so, it might be argued. If we are permitted, without loss of autonomy, to bow to the constraints of ignorance, or to the intractability of nature, why may we not with equal justification adjust ourselves to the limitations of collective as opposed to individual decision-making? When the assembly of the people cannot reach a unanimous decision, decision by lot is the only way to avoid the twin evils of governmental inertia and tyrannization of the minority.

This argument seems to me to be wrong, although my reasons for this belief will only be spelled out with any fullness in the last section of this essay. Briefly, there is a fundamental difference between those obstacles to decision which are outside our control, such as ignorance, and those obstacles which are at least theoretically within our control, such as psychological conflict (in the individual) or disagreement (in the society as a whole). Whereas we have no reason to think that we could ever completely overcome natural obstacles, even in an ideal society, we must suppose that some method exists for resolving conflicts among rational men of good will which allows them to concert their activities without forfeiting their autonomy. The gen-eral adoption of decision by lot would violate the an. tonomy of the citizens.

The most ambitious defense of majoritarianism in the literature of democratic theory is that offered by Jean-Jacques Rousseau in Book IV of the Social Contract. The fundamental problem of political philosophy, according to Rousseau, is to discover whether there is "a form of association which will defend and protect with the whole common force the person and the property of each associate, and by which every person, while uniting himself with all, shall obey only himself and remain as free as before."11 The solution to this problem is the social contract by which men first constitute themselves a polity. By means of the contract, the many particular and divisive wills of the prepolitical community are transformed into the general will of the collective body. Each contracting party pledges himself to "place in common his person and all his power under the supreme direction of the general will; and as one body . .. all receive each member as an indivisible part of the whole."

A will is distinguished by Rousseau as general by virtue both of its form and of its content, or aim. Formally, a will is general insofar as it issues in commands having the form of general law rather than particular edict. Thus, Rousseau considers only the laws of the society to be products of the general will; applications of the laws to particular cases are made by the government, which operates under a mandate from the collective will of the people. Materially, a will is general insofar as it aims at the general good rather than at the particular goods of separate individuals. An individual can be said to have a general will, or to strive for a general will, if he aims at the general good rather than his own good, and if he issues commands having the form of law. Similarly, the group as a whole has a general will when it issues laws which aim at the general good. In this way, Rousseau distinguishes a true political community from an association of self-interested individuals who strike bargains among their competing interests, but nowhere strive for the good of the whole. (The same distinction is said to be embodied in the division of function between the Congress, which represents sectional and class interests, and the president, who is supposed to be guided by the national interest.)

It is Rousseau's claim that when a political community deliberates together on the general good and embodies its deliberations in general laws, it thereby acquires legitimate authority over all the members of the deliberating body, or parliament. Thenceforward, each member of the society has a moral obligation to obey the laws which have been willed by the collectivity. That obligation can be suspended only when the general will is destroyed, which is to say only if the parliament of all the people ceases to aim at the general good or to issue laws.

Rousseau, in keeping with the tradition of democratic theory, introduces the device of majority rule into the founding contract. But he recognizes that the legitimacy of laws enacted by a majority of the parliament cannot be traced merely to the binding force of a promise. In Book IV of the Social Contract, therefore, he returns to the problem:

Except in this original contract, a majority of the votes is sufficient to bind all the others. This is a consequence of the contract itself. But it may be asked how a man can be free and yet forced to conform to the will of others. How are the opposers free when they are in submission to laws to which they have never consented?
Rousseau continues:
I answer that the question is not fairly stated. The citizen consents to all the laws, to those which are passed in spite of his opposition, and even to those which sentence him to punishment if he violates any one of them. The constant will of all the members of the State is the general will; it is by that they are citizens and free. When any law is proposed to the assembly of the people, the question is not precisely to enquire whether they approve the proposition or reject it, but if it is conformable or not to the general will, which is their will. Each citizen, in giving his suffrage, states his mind on that question; and the general will is found by counting the votes. When, therefore, the motion which I opposed carries, it only proves to me that I was mistaken, and that what I believed to be the general will was not so. If my particular opinion had prevailed, I should have done what I was not willing to do, and consequently, I should not have been in a state of freedom.

The air of paradox which surrounds this passage has enticed or repelled students of Rousseau ever since the Social Contract appeared. The notion of man being "forced to be free," which was employed by later idealist political philosophers to justify the state's repression of the individual "in the interest of his own true self," can be traced to this argument. Actually, as I shall try to show, there are no sinister implications to Rousseau's argument, although it is not valid.

The foundation of the argument is a distinction, whose lineage runs at least to Plato, between doing what one wills and doing what one wants. An individual may be said to do what he wills so long as he manages to perform the action which he sets out to perform; but he may thereby fail to do what he wants, if the outcome of the action is other than he anticipated. For example, suppose that I arrive at a train station just as my train is scheduled to leave. Not knowing which track I am to leave from, I rush up to a conductor and shout, "Which track for Boston?" He points at track 6, but I misunderstand him and dash off for track 5, where a train for Philadelphia is also on the point of leaving. The conductor, seeing my mistake, has only two choices: he can allow me to board the wrong train, thereby permitting me to do what I will, or bodily hustle me onto the right train, thereby forcing me to do what I want. Rousseau's description seems perfectly apposite. If the conductor makes no move to stop me, I will fail to do what I want to do, and in that sense not be free.

Consider another case, that of an intern who is on duty in the emergency ward of a hospital. A case comes in which lie misdiagnoses as poisoning. He orders a stomach pump, which is about to be applied when the resident in charge happens by, recognizes the case as actually one of appendicitis, for which the stomach pump would be fatal, and countermands the intern's order to the nurse. Here, the intern's aim is of course to cure the patient, and he is assisted in achieving it by the resident's counterorder, which (in a manner of speaking) forces him to treat the patient correctly. Had he been permitted to follow his own diagnosis, he would have accomplished precisely the end which he most wished to avoid.

Plato, it will be recalled, uses this same argument in the Gorgias and Republic in order to demonstrate that the tyrant is not truly powerful. The tyrant, like all men, wants what is good for him. Power, then, is the ability to get what is good for oneself. But the tyrant, through a defect of true moral knowledge, mistakenly thinks that it is good for him to indulge his appetites, deal unjustly with his fellow men, and subordinate his rational faculties to his unchecked desire and will. As a result, he becomes what we would today call a neurotic individual; he compulsively pursues fantasy-goals whose achievement gives him no real happiness, and he thereby shows himself to be truly powerless to get what he wants.

The three cases of the man catching a train, the intern diagnosing a patient, and the tyrant have three common characteristics on which are founded the distinction between getting what one wills and getting what one wants. First, it is supposedly quite easy to distinguish between the goal of the individual's action and the means which he adopts to achieve it. (This is, of course, debatable in the case of the tyrant; it would hardly be denied in the other cases.) Hence, we can speak meaningfully of the agent's willing the means and wanting the end, and therefore of his doing what he wills but failing to get what he wants. Second, the goal in each case is some state of affairs whose existence is objectively ascertainable, and about which one can have knowledge. (Again, Plato's example is open to dispute; this is precisely the point in the development of his ethical theory at which he makes use of the doctrine that there is such a thing as moral knowledge.) It follows that a man may sometimes know less well what he really wants (i.e., what will really accomplish his own goals) than some independent observer. Finally, in all three cases we are to assume that the individual places a purely instrumental value on the means which he adopts, and would be willing to give them up if he believed that they were ill suited to his ends.

Life is full of significant situations in which we strive to achieve some objective state of affairs, and in which we would therefore be sorry if our mistaken views about the means to those ends were to be adopted. For example, if a member of Congress genuinely wishes to reduce unemployment, and if his traditionalistic convictions about the virtues of a balanced budget are overriden by a liberal majority which seeks to spend the nation into prosperity, and if unemployment is thereupon reduced, then (personal pride to one side) we may expect him to be glad that his views were in the minority, for he can now see that "if his particular opinion had prevailed, he should have done what he was not willing to do, and consequently, he should not have been in a state of freedom."

And we can now see what Rousseau intended in the passage quoted above. He assumes that the assembly of the people is attempting to issue commands which have the form of law and aim at the general good. This is a legitimate assumption for Rousseau to make, since he is only interested in discovering whether a community which does aim at the general good thereby confers legitimacy on the laws which it passes. The further question, whether one can often find an assembly which holds to the ideal of the general good instead of pursuing diverse particular interests, concerns the application of Rousseau's theory. Democratic theorists frequently devote great attention to the problem of devising safeguards against the ineradicable partisanship of even the most enlightened men. Although that is indeed a serious matter, their concern tends to mask their unexamined assumption that a majoritarian democracy of thoroughly public-spirited citizens, if it ever could exist, would possess legitimate authority. This is merely one more reflection of the universal conviction that majority rule is self-evidently legitimate. By recognizing the necessity for an independent justification of majority rule, Rousseau plays in political philosophy the role which Hume plays in the theory of knowledge.

Rousseau supposes further that it is an objectively ascertainable fact whether a proposed law has the proper form and aims at the general good. He thinks, finally, that the proper test of these matters is a vote, in which the majority must inevitably be correct. Hence, when a member of the assembly "gives his suffrage," he is not expressing his preference, but rather offering his opinion on the character of the proposed law. He may perfectly well prefer a different measure, which serves his interest better, and nevertheless vote for the proposal because he believes it to aim at the general good. Since the majority are always right, a member of the minority will by that fact be revealed as supporting inappropriate means to his own end; in short, the minority are like the individual who dashes for the wrong train, or the intern who prescribes the wrong treatment.

The flaw in this argument, of course, is the apparently groundless assumption that the majority are always right in their opinion concerning the general good. (Rousseau's appeal to this assumption is contained in the innocuous-looking words "and the general will is found by counting the votes.") What can possibly have led Rousseau to such an implausible conclusion? Experience would seem rather to suggest that truth lies with the minority in most disputes, and certainly that is the case in the early stages of the acceptance of new discoveries. At any rate, if the nature of the general good is a matter of knowledge, then there would appear to be no ground for assuming that the majority opinion on any particular proposal for the general good will inevitably be correct.

I think we can trace Rousseau's error to a pair of complicated confusions. First, Rousseau has not adequately distinguished between an assembly which attempts to aim at the general good, and one which actually succeeds. In a chapter entitled "Whether the General Will Can Err," he writes:

It follows from what has been said that the general will is always right and tends always to the public advantage; but it does not follow that the deliberations of the people have always the same rectitude. Our will always seeks our own good, but we do not always perceive what it is. The people are never corrupted, but they are often deceived, and only then do they seem to will what is bad. (Bk. I, Ch. 3)
The confusion lies in failing to distinguish three possible conditions of the assembly. First, the citizenry may vote on the basis of private interest, in which case they are not even attempting to realize the general good. That is what Rousseau calls an "aggregate will." Second, the people may strive to achieve the general good, but choose poor laws because of their ignorance, or simply the unpredictability of important aspects of the problems which they face. Insofar as everyone does his best to realize the general good, the collectivity is a genuine moral and political community. Finally, the assembly of the people may aim at the general good and hit it. They may deliberately choose to enact laws which do in fact offer the best way to achieve the good of the community.

Now, there may be some ground for claiming that an assembly which is in the second condition has legitimate authority over its members; one might argue that it acquires authority by virtue of the universal commitment of its members to the general good. But Rousseau's proof of the legitimacy of the majority will only work if we assume that the assembly is in the third condition -- that whenever it is guided by the majority it actually succeeds in moving toward the general good. In that case, it really would be true that a member of the minority could get what he willed (the general good) only by failing to get what he voted for.

The confusion between trying to achieve the general good and succeeding is compounded, I would like to suggest, by a second confusion which leads Rousseau to overlook what would otherwise be a rather obvious error. There are three questions which one might suppose the assembly to be presented with. Rousseau mentions two: Which law do you prefer? and Which law tends to the general good? A third question might also be asked: Which alternative will win? Now the peculiarity of this last question is that the majority opinion must be correct. If everyone's vote is a prediction about the outcome, then the members of the minority will hardly desire their choice to prevail, for by so doing they would violate the principle of majority rule to which they are presumably committed. The phrase "general will" is ambiguous in Rousseau's usage, even though he takes great care to define it earlier in his essay. It should mean "will issuing laws which aim at the general good," but it frequently has for him the more ordinary meaning "preponderant opinion" or "consensus of the group." When the assembly is asked "whether (the proposition before them) is conformable or not to the general will," we may view them either as being asked for their opinion of the value of the proposition for the general good, or else as being asked to make a prediction of the outcome of the vote. I suggest that Rousseau himself confused these two senses, and was thereby led into the manifestly false assumption that the majority opinion of the assembly would successfully express what the minority were really striving for, and hence be binding on everyone who voted for or against.

We appear to be left with no plausible reason for believing that a direct democracy governed by majority rule preserves the moral autonomy of the individual while conferring legitimate authority on the sovereign. The problem remains, that those who submit to laws against which they have voted are no longer autonomous, even though they may have submitted voluntarily. The strongest argument for the moral authority of a majoritarian government is that it is founded upon the unanimous promise of obedience of its subjects. If such a promise may be supposed to exist, then the government does indeed have a moral right to command. But we have discovered no moral reason why men should by their promise bring a democratic state into being, and thereby forfeit their autonomy. The implicit claim of all democratic theory, I repeat, is that it offers a solution to the problem of combining moral liberty (autonomy) with political authority. This claim is justified for the special case of unanimous direct democracy. But none of the arguments which we have considered thus far succeed in demonstrating that this claim is also valid for ma-joritarian democracy.

This is not to deny that there are many other reasons for favoring democracy of one sort or another under the conditions which prevail today in advanced industrial societies. For example, one might reply impatiently to all the foregoing argumentation that majority rule seems to work well enough, and that minorities do not show signs of feeling trampled upon, for all that they may be frustrated or disappointed. To which one need only reply that the psychology of politics is not at issue here. Men's feelings of loss of autonomy, like their feelings of loyalty, are determined by such factors as the relative degree of satisfaction and frustration of deeply held desires which they experience. Modern interest-group democracy is, under some circumstances, an effective means of reducing frustrations, or at least of reducing the connection between frustration and political disaffection. But many other forms of political organization might accomplish this result, such as benevolent autocracy or charismatic dictatorship. If democracy is to make good its title as the only morally legitimate form of politics, then it must solve the problem of the heteronomous minority.

Appendix: The Irrationality of Majority Rule

Majority rule can be called into question on grounds of its failure to preserve the liberty of the minority, but it has commonly been thought to be at least a rational method of making decisions, supposing that the members of the community are willing to agree upon its adoption. In fact it turns out that majority rule is fatally flawed by an internal inconsistency which ought to disqualify it from consideration in any political community whatsoever.

Self-consistency is perhaps the simplest sort of rationality which is demanded of, all men in their deliberations and actions. If a man prefers a first state of affairs or action to a second, and prefers the second in turn to a third, then in all consistency he ought to prefer the first to the third. There is of course no psychological law which forces a man to keep his preferences consistent, any more than to adopt only means which he believes are well suited to his ends. But in exploring the theoretical possibility of a legitimate state, we are surely justified in positing a community of citizens who rise to that first level of rationality.

Presumably, also, we desire that the method of group decision which we adopt will lead to collective action having the like virtue of internal consistency. Unanimous democracy achieves this end, for it reproduces in the laws of the state the common preferences of the entire citizenry. If their preferences are consistent, so too will be those of the state. It might be thought that majority rule also preserved consistency of preference, but the facts are otherwise. As a simple example will illustrate, it is perfectly possible for a group of rational individuals with consistent preferences to arrive, by majority rule, at a completely inconsistent order of group preference! Suppose for the sake of simplicity that the community consists of three individuals who are faced with the problem of establishing a social ranking among three alternatives.12 Each member of the voting community is first asked to rank the three possi-bilities in order of his relative preference. He may use any criteria he chooses -- such as social utility, personal interest, or even whim -- but he must be consistent. The group then establishes its collective preference by voting for the alternatives, two at a time. Since there are three alternatives, which we can call A, B, and C, there will be three votes in all: first A against B, then A against C, and finally B against C.

The preference order of the society is completely determined by the preference orders of the individuals, for whenever a pair of alternatives is presented to them, each man consults his private ranking and votes for the higher of the two. Now, there are a great many possible sets of private orderings which, when amalgamated by the device of majority rule, will produce a consistent public ordering. For example, consider the set of orderings in Table 1.

Table 1.
Individual I
Individual II
Individual III

Since Individuals I and II prefer A to B, they outvote Individual III, and the society as a whole prefers A to B. Similarly? Individuals II and III outvote Individual I and commit the society to B over C. Now, if the society prefers A. to B, and B to C, then in all consistency, it ought also to prefer A to C. And so indeed it does, for Individuals I and II vote that preference, and thereby overrule Individual III once more. In this case, majority rule has transformed a consistent set of individual or private preference rankings into an equally consistent social preference ranking. But unfortunately, it is not always so.

Consider the set of individual orderings of the same alternatives in Table 2.

Table 2.
Individual I
Individual II
Individual III

When we pair the alternatives and count the votes, we discover that there is a majority for A over B (Individuals I and II), and a majority for B over C (Individuals I and II), but not therefore a majority for A over C. Quite to the contrary, Individuals II and III prefer C to A, and therefore so does the society. The result is that the group as a whole, starting from perfectly consistent individual preferences, has arrived by majority rule at an absurdly inconsistent group preference.

It might be objected that we have presented a false picture of rule by the majority. Assemblies do not vote on all the pair-wise combinations of possibilities which are under consideration. They either vote for all at once, and allow a plurality to decide, or else they take measures up one at a time, adopting or rejecting them. It makes no difference. The contradictions which we have discovered in majority voting can be reproduced in any of the ordinary variations which might be adopted by an assembly. For example, suppose that the procedure is followed of voting on the alternatives one at a time, until one is adopted, which thereupon becomes law. Each citizen votes against a proposal if there is some alternative still in the running which he prefers. On the other hand, once a proposal has been voted down, it is eliminated from the contest and is ignored by the electorate. Under this system, one can easily show that the winning measure is determined (in the paradoxical case outlined above) solely by the order in which the possibilities are brought before the voters. To see that this is true, consider once more the pattern of preferences exhibited in Table 2. There are three alternatives, A, B, and C. Hence there are six different orders in which the alternatives can be presented to the assembly, namely ABC, ACB, BAC, BCA, CAB, and CBA. Let us see what happens in each case under the system of eliminative voting.

Case 1. A is put before the assembly and loses, since two individuals prefer something else to it.

B is now put before the assembly and wins, for with A eliminated, there are now two individuals who prefer it to anything else (i.e., to C), and only one who still has a prior preference for C.

So B wins.

Case 2. A is put before the assembly and loses; C is put before the assembly and also loses; leaving B, which wins.
Case 3. By the same line of reasoning, when B is put before the assembly it loses; whereupon A also loses, leaving C, which wins.
Case 4.B loses; C wins.
Case 5. Starting with C, which loses, we end up with A, which wins.
Case 6. A wins.

In short, when alternative A is voted on first, alternative B wins; when alternative B is voted on first, alternative C wins; and when alternative C is voted on first, alternative A wins. It is clearly irrational for a society to change its preference among three alternatives whenever it considers them in a different order. That would be like saying that I prefer chocolate ice cream to vanilla when I am offered chocolate first, but prefer vanilla to chocolate when I am offered vanilla first!

Kenneth Arrow, in an important monograph entitled Social Choice and Individual Values, has demonstrated that the inconsistency of the voter's paradox infects virtually every method of social choice which can lay a reasonable claim to being called "democratic." How can it be that when rational men with consistent preferences make collective decisions by the apparently legitimate device of majority rule, they may arrive at inconsistent group preferences? What is it about the process of collective decision which introduces an element of irrationality?

The answer seems to be contained in a very interesting discovery of Duncan Black concerning the conditions under which majority rule can be trusted to yield consistent results. It is obvious that we can guarantee the consistency of majority rule if we are permitted to set limits to the patterns of individual preference which the voters may adopt. In the extreme case, for example, if we require everyone to adopt the same preference order, then of course majority rule will simply reproduce that order as the social preference, which will be consistent. But are there any reasonable restrictions that will do the job? And, further, what is the weakest restriction that will ensure a consistent social preference order? The answer to the latter question is not yet known, but Black has demonstrated that under one interesting and natural restriction, majority rule will work consistently.

Briefly, the restriction is that every individual's preference order must exhibit the characteristic which he calls "single-peakedness" when plotted on a single scale. This means that there is some one-dimensional array of all the alternatives, on which each individual can locate his first choice, and which has the property that for every individual, the farther to the right an alternative is from his first choice, the less he prefers it, and the farther to the left an alternative is from his first choice, the less he prefers it. We are all familiar with such an array, namely the "left-right" spectrum in politics. If we string out the various political positions on the spectrum from extreme left, or radical, to extreme right, or reactionary, then the following is true: First, each individual can locate himself along the spectrum; Second, once he has found his place, which is the position of his first choice, then the farther to the right or left something is, the less he likes it.13 For example, a moderate Republican prefers a conservative to a radical, and he also prefers a liberal Republican to a moderate Democrat. A left-wing Democrat prefers a socialist to a Communist, and also a middle-of-the-road Democrat to an Eisenhower Republican. And so forth. Black has demonstrated mathematically that if every person can satisfactorily fit his preferences onto such a spectrum, then majority rule must give a consistent social preference.

It is not completely clear what the deeper significance is of Black's discovery. One clue seems to be that single-peakedness, or arrangement along a left-right spectrum, occurs when everyone in the society views the alternatives as embodying varying degrees of some one magnitude. This is roughly akin to Aristotle's notion of virtue as a mean between extremes. Each virtue is seen as occupying a position on a scale, midway (roughly) between an excess and a defect. For example, courage is analyzed as a mean between rashness and cowardice. Presumably, the further one errs toward the direction of either extreme, the worse one is. In politics, we might interpret the left-right spectrum as a reflection of varying degrees of government intervention in social questions. At one end are the conservatives, who desire minimum intervention; at the other end are the socialists, who desire maximum intervention; and strung out between the two are various types of moderates who favor a mixture of intervention and nonintervention.14

When a single individual evaluates alternatives, the variable or variables with which he is concerned presumably remain the same throughout his evaluation. This is one of the sources of his internal consistency. But when many individuals evaluate the same objective alternatives, they may do so in terms of a diversity of variables. The result is that when their decisions are collectively amalgamated through voting, the group preference may embody the inconsistency of standards of evaluation which existed, in a disaggregated form, in the voting population. It would seem, therefore, that majority rule has the best chance of yielding consistent results when the entire citizenry views the issues as polarized, in terms of variables which make it natural to prefer alternatives less and less as they diverge, in either direction, from one's first choice.

In order to see how lack of single-peakedness can lead to inconsistency, let us take a look at a simplified society in which there are three voters, a conservative, a welfare-state liberal, and a socialist, who must choose among three alternatives, namely laissez-faire capitalism, welfare-state liberalism, and socialism. The conservative, we may assume, would prefer laissez faire first, welfare-state liberalism second, and socialism last. It is also plausible that the liberal would prefer welfare-state liberalism first, socialism second, and laissez-faire capitalism last. But the socialist, who locates himself at the extreme left of the political spectrum, and prefers socialism first, might not prefer the welfare state second. He might in fact think that the welfare state had the worst features of both laissez-faire capitalism and socialism, with the virtues of neither. The welfare state throttles individual initiative, which does after all have a number of socially desirable consequences under capitalism, while also laying upon the society the burden of bureaucracy devoid of the rational total control possible under socialism. The socialist's preference order might therefore read socialism first, laissez faire second, and the welfare state last. Table 3 summarizes these individual preference orders:

Table 3.
Conservative Liberal Socialist
laissez faire welfare state socialism
welfare state socialism laissez faire
socialism laissez faire welfare state

What would be the result of a vote? The society would prefer laissez faire to the welfare state, two-to-one; it would also prefer the welfare state to socialism, two-to-one. But it would not prefer laissez faire to socialism. Quite to the contrary, by a vote of 2 to 1 it would prefer socialism to laissez faire. Thus even when the members of a voting assembly see the alternatives as embodying varying degrees of a single magnitude (state control), there may still not be a single-peakedness, and hence no consistency in the group preference.


6. In recent years, a number of political philosophers have explored the possibilities of decision by unanimity, and it turns out that much more can be achieved than one would expect. For example, John Rawls, in an influential and widely read essay, "Justice as Fairness," uses certain models taken from bargaining theory to analyze the conditions under which rational men with conflicting interests might arrive at unanimous agreement on the procedural principles for resolving their disputes. See Rawls in Philosophy, Politics, and Society, 2nd series, eds. P. Laslett and W. Runciman.

7. Strictly speaking, this second example of a viable unanimous community is imperfect, since there is a significant difference between committing oneself to a moral principle and calculating one's enlightened self-interest. For an illuminating discussion of the moral importance of committing oneself to a principle, see Rawls, op. cit.

8. Needless to say, the origin of parliaments historically has nothing to do with this problem. It is rather the other way around: first there were parliaments, then there was universal suffrage.

9. A great deal has been written, in mitigation of the manifest historical implausibility of contract theories, about the metaphorical or mythical character of the original "contract." Sometimes, for example, it is said that the contract merely states in convenient form the underlying moral consensus of the society. It should be clear that a sophisticated interpretation of this sort will not do, if one wishes to found majority rule on the promise contained in the contract. A promise is an act, not the mere expression or summation of an existing obligation. It creates a new obligation where none existed before. Whatever may be my general moral obligation to do an act, my promise to do it lays an independent burden of responsibility upon me. Hence, those theorists who trace the legitimacy of majoritarianism to the contract cannot, in all consistency, dissolve the contract into a myth. Needless to say, there can be tacit promises as well as explicit promises, and therefore tacit or quasi-contracts of the sort which are invoked to explain the obligation of succeeding generations.

10. I am deliberately glossing over the much more controversial question, whether it is reasonable to equate a less probable outcome having a high value to me with a more probable outcome having a low value. Somewhat more technically, the question is whether I ought to be guided by my calculation of the expected value, or mathematical expectation, of the alternatives open to me. Von Neumann and Morgenstern, in their development of the pure theory of games, assume the rationality of maximization of expected value, but there is nothing approaching consensus on the issue in the contemporary literature.

11. This is essentially the problem which I have called the deduction of the possibility of political philosophy. Rousseau appears to be the first political philosopher to recognize explicitly the conflict between the demands of moral autonomy and legitimate authority. My treatment of the problem owes a great deal to the Social Contract. (Bk. I, Ch. VI)

12. The paradox, or inconsistency, which is developed in the text may be duplicated in any case involving two or more voters and three or more alternatives, assuming that one is permitted to be indifferent between any pair of alternatives, as well as to prefer one to the other. The "voter's paradox," as it is called, has been known for some time, and was actually the subject of an extended treatise by the nineteenth-century mathematician Charles Dodgson, better known as Lewis Carroll.

13. But notice, nothing can be said about his relative preferences among one position to the right and another to the left. This is because the ordering of his preference is ordinal, not cardinal.

14. Notice that in this case, the conservatives and socialists do not focus their attention upon the same variable, but rather on two different variables which may be supposed to vary together. The conservatives are concerned with intervention per se, but the socialists are presumably concerned with social welfare and social justice, which they believe varies directly with the degree of intervention.