Karl Popper on Democracy: The Open Society and its Enemies Revisited


Editor’s Note: In 1988 The Economist invited the philosopher Karl Popper to write an article on democracy. It appeared in the issue of April 23rd that year and made the case for a two-party system. As America’s presidential race begins, we are republishing it, below. The first book in English by Professor Sir Karl Popper was accepted for publication in London while Hitler’s bombs were falling, and was published in 1945 under the title The Open Society and its Enemies. The book was well received, but in this article Sir Karl questions whether his central theory of democracy (which he does not characterise as “the rule of the people”) has been understood.


Sir Karl Popper

By Karl Popper

MY THEORY of democracy is very simple and easy for everybody to understand. But its fundamental problem is so different from the age-old theory of democracy which everybody takes for granted that it seems that this difference has not been grasped, just because of the simplicity of the theory. It avoids high-sounding, abstract words like “rule”, “freedom” and “reason”. I do believe in freedom and reason, but I do not think that one can construct a simple, practical and fruitful theory in these terms. They are too abstract, and too prone to be misused; and, of course, nothing whatever can be gained by their definition.

This article is divided into three main parts. The first sets out, briefly, what may be called the classical theory of democracy: the theory of the rule of the people. The second is a brief sketch of my more realistic theory. The third is, in the main, an outline of some practical applications of my theory, in reply to the question: “What practical difference does this new theory make?”

The classical theory

The classical theory is, in brief, the theory that democracy is the rule of the people, and that the people have a right to rule. For the claim that the people have this right, many and various reasons have been given; however, it will not be necessary for me to enter into these reasons here. Instead, I will briefly examine some of the historical background of the theory, and of the terminology.

Plato was the first theoretician to make a system out of the distinctions between what he regarded as the main forms of the city-state. According to the number of the rulers, he distinguished between: (1) monarchy, the rule of one good man, and tyranny, the distorted form of monarchy; (2) aristocracy, the rule of a few good men, and oligarchy, its distorted form; (3) democracy, the rule of the many, of all the people. Democracy did not have two forms. For the many always formed a rabble, and so democracy was distorted in itself.

If one looks more closely at this classification, and if one asks oneself what problem was at the back of Plato’s mind, then one finds that the following underlay not only Plato’s classification and theory, but also those of everybody else. From Plato to Karl Marx and beyond, the fundamental problem has always been: who should rule the state? (One of my main points will be that this problem must be replaced by a totally different one.) Plato’s answer was simple and naive: “the best” should rule. If possible, “the best of all”, alone. Next choice: the best few, the aristocrats. But certainly not the many, the rabble, the demos.

The Athenian practice had been, even before Plato’s birth, precisely the opposite: the people, the demos, should rule. All important. political decisions—such as war and peace—were made by the assembly of all full citizens. This is now called “direct democracy”; but we must never forget that the citizens formed a minority of the inhabitants—even of the natives. From the point of view here adopted, the important thing is that, in practice, the Athenian democrats regarded their democracy as the alternative to tyranny—to arbitrary rule: in fact, they knew well that a popular leader might be invested with tyrannical powers by a popular vote.

So they knew that a popular vote may be wrongheaded, even in the most important matters. (The institution of ostracism recognised this: the ostracised person was banned as a matter of precaution only; he was neither tried nor regarded as guilty.) The Athenians were right: decisions arrived at democratically, and even the powers conveyed upon a government by a democratic vote, may be wrong. It is hard, if not impossible, to construct a constitution that safeguards against mistakes. This is one of the strongest reasons for founding the idea of democracy upon the practical principle of avoiding tyranny rather than upon a divine, or a morally legitimate, right of the people to rule.

The (in my opinion, vicious) principle of legitimacy plays a great part in European history. While the Roman legions were strong, the Caesars based their power upon the principle: the army legitimises the ruler (by acclamation). But with the decline of the Empire, the problem of legitimacy became urgent; and this was strongly felt by Diocletian, who tried to support the new structure of the Imperium of the God-Caesars ideologically with traditional and religious distinctions and the corresponding attribution of different titles: Caesar, Augustus, Herculius, and Jovius (i.e., related to Jupiter).

Yet it seems that there was a need for a more authoritative, deeper religious legitimation. In the next generation, monotheism in the form of Christianity (which, of the available monotheisms, has spread most widely) offered itself to Constantine as the solution to the problem. From then on, the ruler ruled by the Grace of God—of the one and the only universal God. The complete success of this new ideology of legitimacy explains both the ties and the tensions between the spiritual and the worldly powers which thus became mutual dependants, and therefore rivals, throughout the Middle Ages.

So, in the Middle Ages, the answer to the question “Who should rule?” became the principle: God is the ruler, and He rules through His legitimate human representatives. It was this principle of legitimacy which was first seriously challenged by the Reformation and then by the English revolution of 1648-49 which proclaimed the divine right of the people to rule. But in this revolution the divine right of the people was immediately used to establish the dictatorship of Oliver Cromwell.

After the dictator’s death, there was a return to the older form of legitimacy; and it was the violation of protestant legitimacy by James II, by the legitimate monarch himself, which led to the “Glorious Revolution” of 1688, and to the development of British democracy through a gradual strengthening of the power of Parliament, which had legitimised William and Mary. The unique character of this development was precisely due to the experience that fundamental theological and ideological quarrels about who should rule lead only to catastrophe. Royal legitimacy was no longer a reliable principle, nor was the rule of the people. In practice, there was a monarchy of somewhat dubious legitimacy created by the will of Parliament, and a fairly steady increase of parliamentary power. The British became dubious about abstract principles; and the Platonic problem “Who should rule?” was no longer seriously raised in Britain until our own days.

Karl Marx, who was not a British politician, was still dominated by the old Platonic problem which he saw as: “Who should rule? The good or the bad—the workers or the capitalists?” And even those who rejected the state altogether, in the name of freedom, could not free themselves from the fetters of a misconceived old problem; for they called themselves anarchists—that is, opponents of all forms of rule. One can sympathise with their unsuccessful attempt to get away from the old problem “Who should rule?”

A more realistic theory

In “The Open Society and its Enemies” I suggested that an entirely new problem should be recognised as the fundamental problem of a rational political theory. The new problem, as distinct from the old “Who should rule?”, can be formulated as follows: how is the state to be constituted so that bad rulers can be got rid of without bloodshed, without violence?

This, in contrast to the old question, is a thoroughly practical, almost technical, problem. And the modern so-called democracies are all good examples of practical solutions to this problem, even though they were not consciously designed with this problem in mind. For they all adopt what is the simplest solution to the new problem—that is, the principle that the government can be dismissed by a majority vote.

In theory, however, these modern democracies are still based on the old problem, and on the completely impractical ideology that it is the people, the whole adult population, who are, or should by rights be, the real and ultimate and the only legitimate rulers. But, of course, nowhere do the people actually rule. It is governments that rule (and, unfortunately, also bureaucrats, our civil servants—or our uncivil masters, as Winston Churchill called them—whom it is difficult, if not impossible, to make accountable for their actions).

What are the consequences of this simple and practical theory of government? My way of putting the problem and my simple solution do not, of course, clash with the practice of western democracies, such as the unwritten constitution of Britain, and the many written constitutions which took the British Parliament more or less as their model. It is this practice (and not their theory) which my theory—my problem and its solution—tries to describe. And, for this reason, I think that I may call it a theory of “democracy”, even though it is emphatically not a theory of the “rule of the people”, but rather the rule of law that postulates the bloodless dismissal of the government by a majority vote.

My theory easily avoids the paradoxes and difficulties of the old theory—for instance, such problems as “What has to be done if ever the people vote to establish a dictatorship?” Of course, this is not likely to happen if the vote is free. But it has happened. And what if it does happen? Most constitutions in fact require far more than a majority vote to amend or change constitutional provisions, and thus would demand perhaps a two-thirds or even a three-quarters (“qualified”) majority for a vote against democracy. But this demand shows that they provide for such a change; and at the same time they do not conform to the principle that the (“unqualified”) majority will is the ultimate source of power—that the people, through a majority vote, are entitled to rule.

All these theoretical difficulties are avoided if one abandons the question “Who should rule?” and replaces it by the new and practical problem: how can we best avoid situations in which a bad ruler causes too much harm? When we say that the best solution known to us is a constitution that allows a majority vote to dismiss the government, then we do not say the majority vote will always be right. We do not even say that it will usually be right. We say only that this very imperfect procedure is the best so far invented. Winston Churchill once said, jokingly, that democracy is the worst form of government—with the exception of all other known forms of government.

And this is the point: anybody who has ever lived under another form of government—that is, under a dictatorship which cannot be removed without bloodshed—will know that a democracy, imperfect though it is, is worth fighting for and, I believe, worth dying for. This, however, is only my personal conviction. I should regard it as wrong to try to persuade others of it (emphasis added).

We could base our whole theory on this, that there are only two alternatives known to us: either a dictatorship or some form of democracy. And we do not base our choice on the goodness of democracy, which may be doubtful, but solely on the evilness of a dictatorship, which is certain. Not only because the dictator is bound to make bad use of his power, but because a dictator, even if he were benevolent, would rob all others of their responsibility, and thus of their human rights and duties. This is a sufficient basis for deciding in favour of democracy—that is, a rule of law that enables us to get rid of the government. No majority, however large, ought to be qualified to abandon this rule of law.

Proportional representation

Such are the theoretical differences between the old and the new theories. As an example of the practical difference between the theories, I propose to examine the issue of proportional representation.

The old theory and the belief that the rule of the people, by the people, and for the people constitutes a natural right, or a divine right, form the background of the usual argument in favour of proportional representation. For if people rule through their representatives, and by majority votes, then it is essential that the numerical distribution of opinion among the representatives mirrors as closely as possible that which prevails among those who are the real source of legitimate power: the people themselves. Everything else will he not only grossly unfair but against all the principles of justice.

This argument collapses if the old theory is given up, so that we can look, more dispassionately, and perhaps without much prejudice, at the inescapable (and possibly unintended) practical consequences of proportional representation. And these are devastating.

First of all, proportional representation confers, even if only indirectly, a constitutional status on political parties which they would otherwise not attain. For I can no longer choose a person whom I trust to represent me: I can choose only a party. And the people who may represent the party are chosen only by the party. And while people and their opinions always deserve the greatest respect, the opinions adopted by parties (which are typically instruments of personal advancement and of power, with all the chances for intrigue which this implies) are not to be identified with ordinary human opinions: they are ideologies.

In a constitution that does not provide for proportional representation, parties need not be mentioned at all. They need not be given official status. The electorate of each constituency sends its personal representative to the chamber. Whether he stands alone, or whether he combines with some others to form a party, is left to him. It is an affair he may have to explain and defend to his electorate.

His duty is to represent the interests of all those people whom he represents to the best of his ability. These interests will in almost all cases be identical with those of all the citizens of the country, of the nation. These are the interests he must pursue to the best of his knowledge. He is personally responsible to persons.

This is the only duty and the only responsibility of the representative that must be recognised by the constitution. If he considers that he has also a duty to a political party, then this must be due solely to the fact that he believes that through his connection with that party he can do his primary duty better than without the party. Consequently it is his duty to leave the party whenever he realises that he can do his primary duty better without that party, or perhaps with a different party.

All this is done away with if the constitution of the state incorporates proportional representation. For under proportional representation the candidate seeks election solely as the representative of a party, whatever the wording of the constitution may be. If he is elected, he is elected mainly, if not solely, because he belongs to, and represents, a certain party. Thus, his main loyalty must be to his party, and to the party’s ideology; not to people (except, perhaps, the leaders of the party).

It can therefore never be his duty to vote against his party. On the contrary, he is morally bound to the party as whose representative he was voted into parliament. And in the event that he can no longer square this with his conscience, it would, in my opinion, be his moral duty to resign not only from his party but from parliament, even though the country’s constitution may place no such obligation upon him.

In fact, the system under which he was elected robs him of personal responsibility; it makes of him a voting machine rather than a thinking and feeling person. In my view, this is by itself a sufficient argument against proportional representation. For what we need in politics are individuals who can judge on their own and who are prepared to carry personal responsibility.

Such individuals are difficult to find under any party system, even without proportional representation—and it must be admitted that we have not yet found a way of doing without parties. But if we have to have parties, we had better not, by our constitution, add deliberately to the enslavement of our representatives to the party machine and to the party ideology by introducing proportional representation.

The immediate consequence of proportional representation is that it will tend to increase the number of parties. This, at first glance, may seem desirable: more parties means more choice, more opportunities, less rigidity, more criticism. It also means a greater distribution of influence and of power.

However, this first impression is totally mistaken. The existence of many parties means, essentially, that a coalition government becomes inevitable. It means difficulties in the formation of any new government, and in keeping a government together for any length of time.

Minority rule

While proportional representation is based on the idea that the influence of a party should be proportional to its voting power, a coalition government means, more often than not, that small parties can exercise a disproportionately great—and often decisive—influence, both on the formation of a government and on its resignation, and so on all its decisions. Most important of all, it means the decay of responsibility. For in a coalition government there is reduced responsibility for all the partners in the coalition.

Proportional representation—and the greater number of parties as a result thereof—may therefore have a detrimental effect on the decisive issue of how to get rid of a government by voting it out of office, for instance in a parliamentary election. The voters are led to expect that perhaps none of the parties will obtain an absolute majority. With this expectation in their minds, the people hardly vote against any of the parties. As a result, on election day none of the parties is dismissed, none is convicted. Accordingly, nobody looks on election day as a Day of Judgment; as a day when a responsible government stands to account for its deeds and omissions, for its successes and failures, and a responsible opposition criticises this record and explains what steps the government ought to have taken, and why.

The loss of 5% or 10% of votes by one or other of the parties is not seen by the voters as a verdict of “guilty”. They look at it, rather, as a temporary fluctuation in popularity. In time, the people become used to the idea that none of the political parties or their leaders can really be made accountable for their decisions which may have been forced on them by the necessity to form a coalition.

From the point of view of the new theory, election day ought to be a Day of Judgment. As Pericles of Athens said in about 430 BC, “although only a few may originate a policy, we are all able to judge it.” Of course, we may misjudge it; in fact, we often do. But if we have lived through a party’s period of power and have felt its repercussions, we have at least some qualifications for judgment.

This presupposes that the party in power and its leaders were fully responsible for what they were doing. This, in turn, presupposes a majority government. But with proportional representation, even in the case of a single party governing with an absolute majority and thrown out by a majority of disenchanted citizens, the government may not be turned out of office. It would rather look for the smallest party strong enough to go on ruling with its help.

Hence the censured leader of the larger party would still continue to lead the government—in direct opposition to the majority vote and on the basis of help received from one of the small parties whose policies, in theory, may be far removed from “representing the will of the people”. Of course, the small party may not be strongly represented in the new government. But its power will be very great since it may topple the government at any time. All this grossly violates the idea that lies at the root of proportional representation: the idea that the influence exercised by any party must correspond to the number of votes it can muster.

The two-party system

In order to make a majority government probable, we need something approaching a two-party system, as in Britain and in the United States. Since the existence of the practice of proportional representation makes such a possibility hard to attain, I suggest that, in the interest of parliamentary responsibility, we should resist the perhaps-tempting idea that democracy demands proportional representation. Instead, we should strive for a two-party system, or at least for an approximation to it, for such a system encourages a continual process of self-criticism by the two parties.

Such a view will, however, provoke frequently voiced objections to the two-party system that merit examination: “A two-party system represses the formation of other parties.” This is correct. But considerable changes are apparent within the two major parties in Britain as well as in the United States. So the repression need not be a denial of flexibility.

The point is that in a two-party system the defeated party is liable to take an electoral defeat seriously. So it may look for an internal reform of its aims, which is an ideological reform. If the party is defeated twice in succession, or even three times, the search for new ideas may become frantic, which obviously is a healthy development. This is likely to happen, even if the loss of votes was not very great.

Under a system with many parties, and with coalitions, this is not likely to happen. Especially when the loss of votes is small, both the party bosses and the electorate are inclined to take the change quietly. They regard it as part of the game—since none of the parties had clear responsibilities. A democracy needs parties that are more sensitive than that and, if possible, constantly on the alert. Only in this way can they be induced to be self-critical. As things stand, an inclination to self-criticism after an electoral defeat is far more pronounced in countries with a two-party system than in those where there are several parties. In practice, then, a two-party system is likely to be more flexible than a multi-party system, contrary to first impressions.

It is said: “Proportional representation gives a new party a chance to rise. Without it, the chance is much diminished. And the mere existence of a third party may greatly improve the performance of the two big parties.” This may well be the case. But what if five or six such new parties emerge? As we have seen, even one small party may wield quite disproportionate power if it is in the position to decide which of the two big parties it will join to form a coalition government.

It is also said: “A two-party system is incompatible with the idea of an open society—with the openness for new ideas, and with the idea of pluralism.” Reply: both Britain and the United States are very open to new ideas. Complete openness would, of course, be self-defeating, as would be complete freedom. Also cultural openness and political openness are two different things. And more important even than opening wider and wider the political debate may be a proper attitude towards the political Day of Judgment.

This article was originally published in The Economist. It is republished here for academic fair use purposes to help educate people who do not read The Economist.

© 2016, Glynn Wilson. All rights reserved.


  3 comments for “Karl Popper on Democracy: The Open Society and its Enemies Revisited

    February 14, 2016 at 10:56 am

    The only problem with majority rule is that it is much more imperfect than the system we now have. We do not live in a democracy. We live in a Constitutional Republic, which just happens to be the best form of government the world has ever seen…up until now. Not perfect, but not too shabby, either.

    As far as our two-party system is concerned, we are always open to a third party–or even a fourth party–if the platforms and the candidates are awesome.

  2. February 14, 2016 at 11:01 am

    A democratic republic, if you can keep it…

      February 14, 2016 at 11:17 am

      Thank you, Ben. Touche!

Comments are closed.