Majoritarianism:
An argument from Rousseau and Condorcet

Trevor Pateman

Abstract. What makes the submission of a minority to a majority reasonable? Can we know that a majority decision is correct? A study of two attempts to answer these questions.

Imagine two societies alike in as many respects as you wish except this: in one society, one person rules over him/ herself and all the rest; in the other, majorities rule over themselves and over minorities. To be a democrat just is to think the latter arrangement better in principle than the former. The democrat who has no other fundamental principles than the majoritarian one is the person who, asked to choose between a benevolent despotism and the tyranny of some majority, is obliged to choose the latter. Considered from the standpoint of any tyrannised minority, it must appear either that democratic majoritarianism is a vicious principle, or else that there is more to democracy than the single, and simple-minded, democrat allows. More precisely, a plausible democratic theory ought to be able to answer two questions to the satisfaction of potential minorities: (1) Why should we prefer to be ruled over by 51 % of the voting population rather than any lesser number? (2) Why should we accept to be ruled over by 51 % rather than any greater number and, in particular, 100% of the voting population - everyone?

Justifying the rule of majorities

One answer sometimes given to the first question is the swings and roundabouts answer: if you live in a democratic polity, sometimes you will be on the majority side and so get your way at least some of the time. In an autocracy, you are either the autocrat or you are nobody. This answer shouldn't satisfy an American Black, a German Jew, a Czech Romany or an Israeli Arab - that is to say, minorities who can find themselves always in a minority on policy decisions which directly, and adversely, affect them. As for the second question, one answer is provided by the Argument from the Polish Diet. In the eighteenth century Polish parliament, voting had to be unanimous before any course of action could be undertaken. Each Polish aristocrat could cast what was known as the liberum veto. And look what happened to Poland! It was carved up between thre other countries (Germany, Russia, Austria-Hungary). That is to say, unanimity or near-unanimity requirements are inefficient; they make societies vulnerable to immobilism and self-destruction.

In the tradition of classical political theory, however, the justification of majoritarianism is approached in a quite different spirit. Majoritarianism is justified in the context of a search for truth, truth concerning the right way to order a society's affairs. In this article, I show how this is true of Rousseau's Social Contract (1762). I do so by disputing a contemporary reading of Rousseau, that of J.C. Hall in his introductory Rousseau (London, Macmillan 1973). I support my reading of Rousseau by linking the interpretation of him to a major but little-studied text by Condorcet, the Essay on the Application of [mathematical] Analysis to the Probability of Majority Decisions (1785). In this way, I hope also to illustrate one advantage of a historically and contextually sensitive reading of the great works of philosophy's past. For Condorcet helps us to make sense of what Rousseau was saying, even though he makes no explicit reference to the Social Contract - hardly surprising when one notes that his book was published in Paris by the Imprimerie Royale just four years before the French Revolution. (The book has, incidentally, never been translated. There is a full discussion and references in K.M. Baker's Condorcet, Chicago University Press, 1975).

Rousseau's daunting problem

Rousseau's central concern with human freedom sets him a daunting task when in the Social Contract(SC) he considers how "to find a form of association which will defend and protect with the whole common force the person and goods of each associate, and in which each, while uniting himself with all, may still obey himself alone and remain as free as before" (SC, p. 174 of the Everyman edition, edited by Cole, Brumfitt and Hall, Dent, London, 1973). For even if freedom consists in obedience to a law which one has prescribed to oneself (SC, p. 178), and not just in the absence of constraint, it is prima facie difficult to see how any association not employing a unanimity requirement for its legislation could be legitimate. Rousseau, however, believes he has an answer to the difficulty, and gives it in the following passage, with which the rest of this article will be concerned:

There is but one law which, from its nature, needs unanimous consent. This is the social compact, for civil association is the most voluntary of all acts. Every man being free and his own master, no one, under any pretext whatsoever, can make any man subject without his consent ... Apart from this primitive contract, the vote of the majority always binds all the rest. This follows from the contract itself ... But it is asked how a man can be both free and forced to conform to wills that are not his own. How are the opponents at once free and subject to laws they have not agreed to?
I retort that the question is wrongly put. The citizen gives his consent to all the laws, including those which are passed in spite of his opposition, and even those which punish him when he dares to break any of them. The constant will of all the members of the state is the general will; by virtue of it they are citizens and free. When in the popular assembly a law is proposed, what the people is asked is not exactly whether it approves or rejects the proposal, but whether it is in conformity with the general will, which is their will. Each man, in giving his vote, states his opinion on that point, and the general will is found by counting votes. When therefore the opinion that is contrary to my own prevails, this proves neither more nor less than that I was mistaken, and that what I thought to be the general will was not so. If my particular opinion had carried the day I should have achieved the opposite of what was my will; and it is in that case that I should not have been free. (SC, pp. 249 - 50)

In his book Rousseau, J.C. Hall suggests that the correct way to read this passage does not involve supposing that outvoted citizens "must change their minds about the merits of the proposal", but that "different answers are given to different questions, the first concerning the policy considered in isolation from its acceptance by the majority, the second taking that acceptance into account" (p. 121). In other words, he says, "to carry out a certain policy may be the best thing one could do to further one's own interest simply because of the policy's wide acceptance, even in the cases where other policies would seem to be more beneficial but for the fact that they were not widely accepted" (p. 121).

Now this interpretation makes Rousseau more of a compromiser than he is. Hall is suggesting that Rousseau is concerned with what Jon Elster has called "politically accessible states" (see his Logic and Society, Chichester, Wiley 1978): the result of a vote shows me that state S2 which I had hoped was accessible from state S1 is not in fact accessible, because of the views of the majority. However, so long as I prefer the state S3 (made available by the majority vote) to a return to a state of nature or personal rebellion, it is rational for me to accept the majority position as a second-best solution. There is, indeed, a tradition of democratic theorising which sees democracy as simply the second-best solution to having your own way on everything.

This approach gives the outvoted citizen no reason to change his [all Rousseau's voters are men] mind on the substantive issue - it is intended to avoid that consequence. But, as a result, it appears only to give the citizen a practical or prudential reason for prescribing the newly-enacted law to himself, and not a moral reason. It is doubtful that this is good enough for Rousseau. For, as all interpreters are agreed, the social contract is supposed to make available to citizens the possibility of moral freedom, or virtue, in place of the natural liberty they lose in entering society. But a man who is motivated by prudence is no more than a foreigner among the citizens.

Further, and more pointedly, Hall's interpretation makes irrelevant Rousseau's stress on the judgmental character of voting, in which voters ideally endeavour to identify an objectively characterisable general will. Voters aim at truth - this is why Rousseau says, in the long passage cited: "If my particular opinion had carried the day I should have achieved the opposite of what was my will", a view only intelligible if voters are seeking the truth and the views of the majority are somehow connected with truth. Voters who, in voting, are merely expressing their approval or disapproval of a proposal could respond in the way Hall suggests, but not Rousseauean voters.

In other words, to achieve consistency with the notion that the general will has objective characteristics, having to do with what is in the common interest, and about which voters ought to make judgements in which they abstract from their particular interests and prejudices, we do have to read Rousseau as arguing that minorities actually have to change their minds, after a proper vote, on the substantive question of what is in the common interest. But it does not follow from this that Rousseau does, or ought to, believe that the majority could not be mistaken, which Hall suggests does follow from the interpretation I am urging (p. 77). Members of a minority have only to conclude that they are more likely to be mistaken than the majority in any instance for them to change their minds, quite rationally. They do not have to believe in the infallibility of the majority: which is as well, since majorities are not infallible - a point of which Rousseau is aware.

Condorcet and the correctness of the majority

Of course, on my reading, Rousseau's 'solution' to the problem of the freedom of outvoted minorities actually begs the question to which it is addressed. For if both majority and minority voters give their votes in the approved fashion, as judgements of truth on a question of what is in the common interest, the outcome of a non-unanimous vote simply yields two sets of views of the general will or common interest, the majority's and the minority's. There is, as yet, no indication of why the majority's view should be thought more correct than the minority's. On this question, Rousseau is silent; yet if his theory of legitimacy is to hold up, he badly needs a principled solution to the puzzle. And so does anyone seeking to sustain or elaborate his central doctrines.

That this is so is attested by the fact that subsequent to the Social Contract there is an identifiable literature in French democratic theory which treats the 'puzzle' I have pointed to as a serious problem requiring resolution. This literature is headed by Condorcet's monumental Essay of 1785 and it continues as late as 1837 (the year of publication of S-D Poisson, Recherches sur la probabilité des judgements en matière criminelle et en matière civile). Though Condorcet's work is often thought of as principally concerned with the so-called paradoxes of voting, periodically rediscovered ever since (definitively by K.J. Arrow in Social Choice and Individual Values, 1951), it is, in fact, largely concerned with the problem of justifying majoritarianism. Though this article is not the place for a full exposition of Condorcet's work (my own M. Phil. thesis, How is Political Knowledge Possible?, Sussex, 1978, contains an extensive bibliography), I will discuss it briefly in its bearing on Rousseau's puzzle.

Condorcet takes the view that since a majority decision submits people (Condorcet unlike Rousseau was a feminist, so I now use 'people' rather than 'man') to an opinion which is not theirs or which they believe contrary to their interest, a very great probability of the correctness of the decision is the only reasonable and just basis on which one can demand the minority's submission. Using a theorem in the calculus of probabilities, earlier proved by Jacques Bernouilli, Condorcet goes on to show that provided the probability of the correctness of each voter's vote (given as a judgement) is greater than a half, then the greater the absolute majority in a vote, the greater the probability of the truth of the majority decision, the limit of this probability being certainty (p=1).

Provided, then, that we have enlightened voters (whose votes have a better than even probability of being correct), who give their votes as independent judgements on the question asked, then all that remains at issue is to decide, for each particular decision, what is the probability of its correctness required if submission to it by the minority is to be reasonable. Here 'reasonable' is to be taken in the strong sense which makes it rational for the minority to accept the decision both intellectually (as more correct than their own judgement) and practically (as fit to govern their own conduct). For example, Condorcet argues that if in the normal course of things there are risks to our life which we would habitually ignore, then we could not reasonably refuse assent to a law which has an equal or lesser probability of exposing our lives to risk. He does not flinch from the calculations and comes up with a risk of 1/144,768 as one which we would habitually ignore He argues - as one immediate consequence - that since no death penalty could be consistently imposed with such a low probability of claiming innocent victims, then the death penalty is unjust. In saying this, he adds his voice to the abolitionist cause launched some years earlier by the Italian jurist, Cesare Beccaria.

The pursuit of political truth

In relation to Rousseau's justification of majoritarianism, what is especially interesting about Condorcet's probabilistic justification is that the necessary conditions Condorcet specifies for majority votes to yield judgements or verdicts which are most probably correct parallel Rousseau's own account of the features of an uncorrupted sovereign body. To what extent he does this intentionally we do not know.

First, majoritarianism is an irrational practice where voters are unenlightened and more desirable the more enlightened they are. Where Condorcet came to see the enlightenment of future citizens as the central task of public education, Rousseau's citizens are competent to discharge their tasks because they live in a simple, egalitarian society where only good sense (Descartes' equally distributed bon sens) is needed to perceive the common good (SC p. 247).

Second, Condorcet's probability calculus can only be applied if voters are giving their votes as judgements on the questions asked, namely "Is x in the common interest?" Rousseau is well aware that this need not be so (see SC, p. 247 where he writes of voters "changing the state of the question"), but he looks to the publicity of proceedings in the assembly, the civil religion, the independence of citizens, and their general education to combat the pull towards partiality and prejudice.

Third, only those votes count in the probability calculus which are given independently of each other, just as the probability of throwing two sixes with two dice is 1/36 only if the dice fall independently of each other. Condorcet is consequently as hostile to partial associations, factions, intrigues and eloquence as Rousseau (SC, p. 185 and 247), who additionally stipulates that "no citizen shall ever be wealthy enough to buy another, and none poor enough to be forced to sell himself' (SC, p. 204).

This kind of analysis could be developed at considerable length. What this short article should have illustrated, in the course of its discussion of Rousseau and Condorcet, is a quite different way of thinking about political decision-making than those now most familiar to us. It is the high point of Enlightenment thought to suppose that government could be organised just like the republic of science, and that democratic government should involve the pursuit of political truth by all citizens. Among contemporary philosophers, it is Jürgen Habermas who most obviously continues this Enlightenment, intellectualist tradition of political thinking, though as he ruefully remarks in his Legitimation Crisis (1976) "Anyone who still discusses the admissibility of truth in practical questions is, at best, old-fashioned".