The dramatic developments in Eastern Europe that culminated in the fall of the Berlin Wall in the late autumn of 1989 set off boisterous comments on the final victory of democracy over dictatorship. When the revolutionary euphoria had settled down, however, we were all forced to return to political normality, where reality shows itself to be more complicated than simplified slogans. A common trait was the collapse of repressive, authoritarian regimes. But the emerging democracies - to the extent that it was possible to talk about democracies - were far from being stabilized. It soon became clear that the transformation of Eastern Europe would be a long and for some countries (such as perhaps Rumania) even a painful process. And why should we be puzzled? The democratic form of government seem obvious for us in West Europe; yet, this has not always been the case.

The fall of 1989 was not only euphoric, it was also surprising - not the least for us political scientists. It's true that we had a premonition during the summer of 1989 when Solidarity won a landslide victory in the election to the newly created upper house in the Polish Siem. Yet 'the autumn of the people' was also preceded by the massacre at Tiananmen Square, where the Chinese regime clearly showed that it was not ready to make any

Resumé

The dramatic and surprising process of democratization in Eastern Europe poses a challenge to political science. There is a demand for theories which may help us to understand these transitions from authoritarian rule to democracy. This article is primarily focused on that set of hypotheses which are found in Adam Przeworski's writings on liberalization and democratization. Its main purpose is to develop some proposals for a game theoretical interpretation of Przeworski's ideas. At the outset this seems to be foredoomed to failure, since in some cases - i.e. the collapse of communism in East Germany and Czechoslovakia - the process of democratization was turbulent to the extent that even the characteristics of a game were the subject of dramatic changes. One may then ask if it is at all possible to model these processes as a game, i.e. a situation where the actors, their opportunity sets and their payoffs are well defined? In lieu of a conclusion the article ends with a suggestion that the snowball effect, as observed at the repeated demonstrations in such places as Leipzig and Wenzler Square, can be understood in terms of Granovetter's threshold model of collective action.

concessions to demonstrating students. Since the developments in Eastern Europe came as such a surprise, it is not obvious that political science can come up with analytical instruments that are needed to explain what happened. The question is whether even now, looking back, we can really understand how and why democracy was established in such countries as Sweden at the end of the 1910s.

The purpose of this article is to provide a commentary on this general problem. At the outset, I must admit that my knowledge about Eastern Europe is limited, nor have I been able to avail myself of thorough analyses of the democratization processes in these different countries. So far, no such analyses have been made, thus forcing me to rely on the same information that is offered to every interested consumer of the mass media. Certainly, more empirical information would have been preferable, but every quest for scientific knowledge also assumes some sort of theoretical starting-point. And in this article, I am concerned with those general ideas, concepts and theories, that can help us understand these processes of democratization.

To begin with, I devote some attention to previous theorizing about democratization. My intent, however, is not to offer a thorough review of different theoretical perspectives; the influential work by Samuel P. Huntington, for example, is entirely neglected. Instead, I focus on a specific tradition according to which 'conflict is the engine of change' (Rustow 1970). My primary focus is Adam Przeworski's recent writings on liberalization and democratization. The main purpose is to develop some proposals for a game theoretical interpretation of his ideas. The article ends, therefore, with consideration of whether some of the events observed in Eastern Europe may be understood in terms of Granovetter's threshold model of collective action.

An Idea about the Different Stages of Democratization

Problems connected with the democratic form of government have been central matters of concern for political science during the entire post-war period. Still, the question of how to explain democratization is a relatively new problem. Traditionally, the establishment of democracy was not treated as a special issue. According to the modernization school which predominated among researchers in the 19605, democratization was considered natural consequence of the improvement of a country's living standards. Economic and social development (measured in terms of urbanization literacy) was accompanied by a change in the thinking of citizens. Modernization of a society brought with it broad-mindedness and

greater openness for other's ideas, which in turn created greater possibilities for democratic decision-making (Lipset 1959, 1983; cf. Hadenius 1992). Consequently, democracy was seen as a sort of spinoff of luxury. It was the kind of system that does not arise until a society has reached a certain level of modernization; and it was a form of government that presumed a certain standard of living among the population.

The basic idea behind modernization theory, in short, was that democracy could be accounted for in terms of the modernization of society. But having arrived at this point, representatives of modernization theory could subconsciously slide between different decisions about what should be explained. Questions about the establishment, degree and stability of democracy were consistently mixed together, as if the causal relations were identical.

Eventually, researchers realized that these issues should be treated as different problems. A pioneering work was Barrington Moore's Social Origins of Dictatorship and Democracy (1965). In addition, there is reason to mention Dunkwart Rustow's article from 1970, 'Transitions to Democracy'. noted that the question of why democracy had a stronger position in some countries than in others must be answered apart from the question of why only some countries had succeeded in introducing democracy. While earlier research could at most only give satisfactory answers to the question of how democracy functioned, Rustow intended to offer a suggestion for finding its origins. His contribution consisted of a sort of generalized, ideal type for the processes of democratization.

Rustow's idea on democratization was based on a comparative study of the processes of democratization in Sweden and Turkey. In summary, the argument can be characterized as a kind of stage theory with four distinct phases. In the introductory phase, according to Rustow, it is necessary to build some sort of national unity. This is the model's only true background condition. The establishment of popular representation requires wide agreement on what constitutes a body politic. In the second, preparatory, phase, a competitive political alternative emerges. A dilemma in this connection is that too much confrontation can make democratization more difficult. According to Rustow, however, the model's first two components together should guarantee a suitable balance between conflict and agreement. third phase is decisive. Here the decision is made over the type of constitution. This is followed by a habituation phase, where the democratic form of government is consolidated.

Rustow's model of democratization can, of course, be criticized. It is clearly insufficient as an explanatory theory, since it is much too general and vague. It is not even well developed as an ideal-type, although I do not do it justice here. At the same time, the model's general character has contributed to its popularity. The model has provided a satisfactory

framework for many other, more precise explanations of democracy's breakthrough. Rustow does not stand out as an eminent theorist on democracy, nevertheless he has secured a place in the history of research on democracy. An important reason for this is that in his empirically oriented work, he tries to capture a prerequisite for democracy which is fundamental: democratic government is founded on the idea that we all agree to give each other the right to disagree.

Democratization as a Result of Power Shifts

In 1980, the Swedish sociologist Goran Therborn published a study of capitalism's significance for the emergence of democracy. The basic methodological of the study is that the question of the emergence of democracy should be answered by using chronological analyses of its breakthrough in OECD's 17 member countries. Therborn's main purpose was to achieve a more precise understanding of the relationship between democracy and capitalism (cf. Hadenius 1988), but his investigation consisted of a systematic account of what preceded the decision to instate universal suffrage in these 17 advanced capitalist democracies. Therborn's conclusion is that democracy has been established in three main ways:

(1) In five of the countries (Austria, Finland, Germany, Italy and Japan), the old authoritarian regimes were replaced by democracies after losing wars. In the case of Austria, Finland and Germany, this occurred after both World Wars. According to Therborn, even Sweden falls indirectly into this category: opposition to democracy had already declined greatly in 1918, but the fall of the German Kaiser and the revolutionary transformations in Sweden's surroundings certainly speeded up the development.

(2) War has also played an important role for the processes of democratization some other countries (Belgium, Canada and Norway). It might have been the decisive factor, although in a very different manner than discussed above. In some cases, the development preceding and during a war has strengthened the political forces working for democratization. pattern comes under the more general category, where democratization is carried out in connection with national mobilization.

(3) Among the remaining countries, Therborn claims that democracy was won through internal developments. In the cases of Australia, Denmark, New Zealand and Switzerland, democratization was propelled an independent middle class, while division among the ruling classes, according to Therborn, was the main cause in France, Great Britain, the Netherlands and the USA.

In all of these patterns Therborn tries to identify the decisive cause for the shifting of advantage in power relations from the opponents to the supporters of democracy. He distinguishes between external and internal factors, as well as between those factors which weaken opponents and those which strengthen supporters. As a result, four main types of explanations emerge, since Therborn's third category can be divided into two parts. These distinctions give us a useful two-by-two table as depicted in Fig. 1.

What happens then if Therborn's typology is applied to the democratization Eastern Europe? Most of the former Soviet satellites should undoubtedly be classified as democratization caused by external factors (the weakening of the Communist Party in the USSR) that drastically the opponents of democracy (the ruling Communist Parties). Furthermore, once the process had started, first in Poland and then in Hungary, a domino effect emerged. As a result, external factors spread their influence to the regimes in East Germany, Czechoslovakia, Rumania and Bulgaria, affecting them with increasing force. The democratization within the Soviet Union is more difficult to interpret, although the weakening of the communist regime was probably decisive. The problem is whether this, in turn, should be explained in terms of external pressure (such as Reagan's military buildup) or internal tendencies toward disintegration and crises.

So far, this brief review has consisted of rather general assertions. It is, of course, possible to develop this analysis further by adding a large quantity of empirical material. It is doubtful, however, if Therborn's typology can be helpful in achieving much more than general assertions of the type suggested. To be fair, Therborn's main objective is to demonstrate the

absence of a direct causal relation from capitalism to democracy. (After 1989, such an investigation is no longer necessary.) This is also the only argument that his methodological design allows. Yet this methodological limitation does not prevent him from developing an extensive argument regarding the fundamental causes of the emergence of democracy. And it is further argumentation that I am interested in.

Therborn's typology over how democracy has been established can be seen as a way of developing Rustow's idea of the different phases of democratization. It is an attempt at distinguishing between a few main factors of the development that takes place within the framework of Rustow's second and third phases (i.e. the preparatory and decision phases). It is apparent, however, that we are still using a rather unsatisfactory of analysis. An immediate objection which can be made against Therborn's approach is that he oversimplifies by reducing the explanation of the establishment of democracy to only one of the four factors. From a methodological viewpoint, it is preferable to have an explanation of democratization processes that allows for different possibilities combining the four explanatory factors. The case of Sweden illustrates this.

A similar position can be taken in explaining the democratization of Eastern Europe. As Timothy Garton Ash (1990) has pointed out, there is good reason to distinguish between Poland and Hungary on the one hand, and East Germany and Czechoslovakia (as well as Rumania and Bulgaria) on the other. The last-mentioned cases can be described as revolutions that were set off by the development of external events. The democratizations of Poland and Hungary, however, were longer, more drawn-out processes. Ash, therefore, calls them 'refolutions'.

... in Poland and Hungary, what was happening could . . . hardly be described as revolution. It was in fact, a mixture of reform and revolution. At the time, I called it 'refolution'. There was a strong and essential element of change 'from above', led by an enlightened minority in the still ruling communist parties. But there was also a vital element of popular pressure 'from below'. In Hungary, there was rather more of the former, in Poland of the latter, yet in both countries the story was that of an interaction between the two. The interaction was, however, largely mediated by negotiations between ruling and opposition elites (Ash 1990, 14).

Thus, there is a need to discriminate between processes of democratization
by combining Therborn's different explanatory factors.

Another problem with Therborn's methodology is that his sample of countries is insufficient for drawing conclusions on the causes of democratization. analysis of the democratization of the capitalist West is, therefore, faulty in more than one point. It does not give much guidance for analyzing the democratization of Eastern Europe. One of Therborn's ideas, however, may serve as the basis for further discussion. This is his

idea of democracy as a possible result of a power shift. Building upon this
idea, the next step is to focus on the power struggle between the supporters
and opponents of democracy.

The Political Prerequisites for Democracy

A recent suggestion for analyzing democracy that has achieved much attention comes from the rational choice theorist Adam Przeworski. He has presented his ideas in two articles from the 1980s: 'Some Problems in the Study of the Transition to Democracy' (1986) and 'Democracy as a Contingent Outcome of Conflicts' (1988). Later, they have been developed in his book Democracy and the Market. Political and Economic Reforms in Eastern Europe and Latin America (1991). The empirical components of Przeworski's investigations usually take the form of illustrations - with particular emphasis on the Polish case - which implies that so far, his ideas may be characterized as untested, theoretical models. My own contribution, therefore, will consist of a game theoretical interpretation of Przeworski's hypotheses.

Przeworski's presentation of the transition from authoritarian to democratic rule is distinctly rationalistic. There is no reason, however, to overemphasize the differences between Przeworski and Therborn. Przeworski's analysis of the processes of democratization actually resembles Therborn's, in the sense that he sees it as a power struggle. Nevertheless, while Therborn is mostly interested in the underlying causal factors, Przeworski focuses his attention on the power struggle that precedes the emergence of democracy. Przeworski uses a more limited scenario for analyzing the power struggle that leads to democratization. His analysis does not exclude influence from war or other external factors, but it concentrates on the course of events within a country, and focuses on the game that takes place between those who remain loyal to an existing authoritarian regime and those who constitute the potential opposition (which might also be democratic).

Przeworski assumes that the regime's representatives want to maintain their authoritarian rule, preferably without need to take repressive measures. opposition would like to overthrow the regime, but it need not have democratic goals. The struggle against authoritarian regimes, for example, has often aimed at ousting the old regime and replacing it with a new authoritarian one. It is likely, of course, that a democracy is most easily established in a country where convinced democrats dominate both the political elite and the remaining population. Yet from a theoretical viewpoint, it is interesting to ask first whether democracy can be established, even in cases where nobody or only a few have it as their foremost goal.

Can democracy be established without convinced democrats? Is it likely that a democratic form of government can come about as the result of a tactical compromise between actors who as a matter of principle support some form of authoritarian rule?

If some level of simplification is allowed, this question can be analyzed in terms of a simple two-person game consisting of common as well as conflicting interests. Both the regime and the opposition are assumed to be unitary actors with two alternative courses of action: they can strive to reach a compromise (C) with the opponent or they can choose a confrontation Depending on their choices, four different outcomes are possible. As indicated by Fig. 2, the stated premises imply that only one of these outcomes makes possible the establishment of a democracy.

As a starting-point, it is assumed that the regime maintains an authoritarian that excludes the opposition from influence over public decision-making (outcome DC). A democratization requires that the regime at the very least is capable of changing its strategy to one of accepting compromise. Przeworski's own idea is rather simple: in order for the regime to increase its willingness to compromise, it is necessary for the opposition's inclination for confrontation to rise. This can be seen, and has also been shown by the development of events in the people's democracies, when it became clear for both the opposition and the existing regime that the Brezhnev Doctrine was no longer in force.

The above conflict model can also be used to illustrate some historical cases. As an example, a couple of extremes can be named. The democratic breakthrough in Sweden in 1917-18 can be seen as a relatively undramatic shift from DC to CC. By comparison, the transformation of Russia during the same period was much more dramatic and complicated within the framework of the model. The development went from the Czar's monarchistic (DC) via the February Revolution (CC) and civil war (DD) to a communist, one-party state (CD).

The next step in using the analytical model assumes that we determine
how the actors rank the various outcomes. A successful democratization

can generally be understood as a change in at least one of the actors' preferences. At the outset, it is assumed that the actors take part in a game where DC is the equilibrium point and changes in preferences cause CC to be the solution to the new game. In terms of the stated problem, this implies that we raise the question of whether this is possible if both actors prefer authoritarian rule over democracy. In other words. is it all possible to establish a stable democracy without convinced democrats? To begin with, we will investigate what types of actors are conceivable and interesting in this context.

In Przeworski's (especially 1988) discussion, he takes it for granted that both the regime and opposition want to avoid a civil war (DD), as well as subordinating themselves to authoritarian rules (CD, respectively DC). Compared with these outcomes, democracy (CC) is seen as an acceptable compromise, but it is not necessarily the best outcome - not even for the opposition. From the regime's viewpoint, this means that CC is better than CD and DD, and that DC is better than CD and DD (the corresponding is true for the opposition). With these restrictions, only a few possible rankings of outcomes are possible (ef. Hermansson 1990, 194 f.). The regime can be characterized as belonging to one of the following four types of actors (the rankings for the opposition are obtained by mirroring those of the regime), where the last two indicate principle support for democracy (i.e. democracy is better than being able to decide alone):

(1) Militant actor: DO CC > DD > CD
(2) Bargaining actor: DO CO CD > DD
(3) Moderate actor: CO DO DD > CD
(4) Pacifistic actor: CO DO CD > DD

If these types of actors are combined, 16 different two-person games become possible (with some being each other's mirrors). The most studied of these are the famous Prisoners' Dilemma (both actors are militant), 'Chicken' (both are bargaining) and the Assurance game (both moderate). The so-called Control game (where one is bargaining and the other moderate) another interesting situation, which is extremely difficult from the viewpoint of cooperation.

A systematic evaluation of these games shows that a cooperative solution (where individual rationality leads to the equilibrium point CC) is not theoretically possible unless both actors have CC as their first preference. None of the games which include a militant or a bargaining actor has in normal form outcome CC as an equilibrium. Thus, the above model is most suitable for illustrating why certain patterns of conflict hinder the emergence of a stable democracy. If the model is taken for granted we can also draw another important conclusion: a democratic form of government presupposes convinced democrats.

The immediate objection to this conclusion is that the model above is too simple. The argument would go something like this: One ought to consider that democratization is a dynamic process. Hence, the actors are facing the choice between C and D not only once, but again and again. It would then be more suitable to model democratization as a supergame in which the simple two-person game is incessantly repeated (Taylor 1987; Axelrod 1984). And the result from the mathematics of supergames is that even a repeated Prisoners' Dilemma has a lot of cooperative solutions, i.e. there are several combinations of superstrategies corresponding to Pareto optimal (and interchangeable) equilibria. One such equilibrium consists of both actors choosing the well-known superstrategy 'tit-for-tat'. However, this requires that we interpret the situation in a different way. The Prisoners' Dilemma supergame is in fact more similar to an Assurance game than to an ordinary Prisoners' Dilemma. In a supergame the actors choose between superstrategies. Tit-for-tat and other similar superstrategies may then model the attitudes of someone who prefers democracy over authoritarian rule and at the same time is willing to defend democracy against its enemies, if necessary by temporarily setting democracy aside. Thus, the same conclusion holds, but we may add that in order to protect democracy, the convinced democrats have to be ready to use undemocratic means.

Liberalization and Democratization

According to conventional wisdom, democracy implies a form of government on the universal and equal right to vote, where adult citizens choose their rulers in free and open elections. Such a definition emphasizes that democracy is based on popular participation and political competition. In Przeworski's case, it is not so essential for him to formulate a noncontroversial of democracy. His main objective is to show what constitutes the decisive break point in the transition from authoritarian rule to democracy. In this case, Przeworski states that the emergence of competition is the major factor. Such a process typically contains two different aspects: liberalization and democratization.

Under liberalization, an authoritarian regime approves of or is forced to accept the emergence of independent organizations. They need not be political organizations, but under an authoritarian system every type of independent popular organization is seen as a potential alternative to the regime. Thus, liberalization implies the emergence of competition of opinions. Przeworski's main point concerning liberalization is that it necessarily implies an unstable situation. He argues that an authoritarian regime cannot approve of a liberalization without it collapsing. The conflict

situation Przeworski has in mind can be described by analyzing the matrix
in Fig. 3.

Przeworski's hypothesis tells us that attempts by the regime to obtain popular support by liberalizing the system (causing a change from DC to CC) can either lead to a collapse of the authoritarian system (CD) or bring about a new era with greater repression (DD). A reasonable interpretation of Przeworski implies that the liberalization game should be specified as a Control game, where the regime's and opposition's ranking of outcomes corresponds to the moderate and bargaining actors noted above. The preference structure of this situation is depicted in Fig. 4 (cf. Nordlund 1990).

In normal form, this game lacks an equilibrium. The so-called cooperative outcome, CC, can only be a solution if one of the actors is allowed to act before the other (i.e. in extensive form). Both have an interest in letting the opposition choose first, because the opposite order gives the solution DC. A tacit or openly proclaimed agreement of this sort, however, is not very realistic. Instead, the only conceivable final result is that the regime forces the opposition into passivity by maintaining its monopoly position and thereby removes all attempts at liberalization from the political agenda.

The liberalization game gives an incentive structure that shows why all
attempts at liberalization in communist countries before glasnost sooner or
later ended up in repression resembling the pre-reform era. The un-

controlled liberalization during the 1980s must therefore be explained in terms of a change in preference within the regime or the opposition. The resulting collapse of the system that spread throughout Eastern Europe during later years could be better understood by taking into consideration the increasing costs of repressing dissidents after the Helsinki accords and the greater media coverage it encouraged. As a result, DD would have been the regime's worst result, C the dominant strategy and CD the solution of the game. An alternative, and perhaps even more interesting solution could have been to concentrate on the opposition's increasing militancy and instead explain the regime's change of strategy as an unsuccessful attempt for the regimes individual actors to maintain cooperation (a Prisoners" Dilemma in which cooperation means that the opposition will be treated harshly). The details of these hypotheses will not be developed any further here, however. Rather, attention will be focused on a slightly different problem, i.e. the problem of accounting for a shift in conditions defining the game.

The situation which liberalization brings about can not only be described in terms of competition of opinions. Sooner or later a situation arises where the old regime's power monopoly is threatened. Once the political arena also contains competition over power, the authoritarian system has ceased to exist. Then it is no longer a question merely of liberalization, but rather of democratization.

Hence the critical moment in any passage from authoritarian to democratic rule is not necessarily the withdrawal of the army into the barracks or the opening of the elected parliament but the crossing of the threshold beyond which no one can intervene to reverse outcomes of the formal democratic process. . . . Democratization is a process of subjecting all interests to competition, of institutionalizing uncertainty. It is thus this very devolution of power over outcomes which constitutes the decisive step toward democracy. There is a moment before which the authoritarian power apparatus controls outcomes and after which no one does. Power is devolved from a group of people to a set of rules (Przeworski 1988, 62 f.).

In several of the previous communist regimes the collapse of the authoritarian happened very quickly. Even in those countries where the shift in power between the regime to the opposition was a more drawn-out process - this is clearly the case of Poland, but also of Hungary - it is possible to discern a certain point when the old power apparatus completely lost its authority. This is illustrated in Fig. 5 showing possible outcomes of the democratization game, where the equilibrium point is quickly moved from DC to CD. The compromise case CC implies here that the old regime's successors are guaranteed a certain degree of power even after the authoritarian regime has fallen.

But why does such a shift occur? What causes the changes in strategy?
My hypothesis is that the change is brought about by a decisive change in
how the actors perceive the outcome DD. Before the collapse of the

authoritarian system, both the regime and the opposition perceive that an open conflict would result in a defeat of the opposition. After a certain point, all the actors believe that an open conflict will lead to the overthrow of the regime. The democratization process can thus be described as a change from one game to another. If we make a few reasonable assumption on how the regime and opposition rank the different outcomes, it can technically be seen as the same type of game before and after the shift of power. In both cases it is a game where a militant actor meets a bargaining one; the only difference is that the actors have changed roles (cf. Fig. 6).

The Snowball Effect

The above interpretation of the democratization process might appear superficial. The proposed hypothesis does not really say more than that the changes occurred because the actors have changed their preferences. In this respect, I follow the typical pattern for rationalist analysis: the actors' preferences are taken for granted and changes in preferences are caused by exogenous factors. Since both partners in the democratization game consist of collective actors, it should be possible to take the analysis at least one step further. The given changes of both the regime and the opposition can be analyzed as a game between actors at a lower level.

Farther down the line, of course, it is the individual's actions which should be described. But in the first step, the regime's actions can be analyzed as a three person game (with hawks, doves and centrists), where the middle group's behavior is decisive for the development. Similarly, the opposition's actions should be interpreted as an outcome of a game between several actors. A complete model over the power struggle that precedes the establishment of democracy must then include several games that simultaneously interact with each other (cf. Tsebelis 1990). To conclude this article, however, I will limit myself to presenting a possible interpretation of the opposition's actions seen as a typical collective action problem.

The opposition's collective action can be described as a function of the citizens' collected stance against the regime. In this case, cooperation implies active protesting against the authoritarian regime, while the opposite implies passivity. Confrontation is reached when a considerable of the population chooses to protest. There is no reason to assume that all citizens have the same relative evaluation of protesting compared to passivity. A large part of the population, however, can be assumed to have utility functions of the sort illustrated in the Schelling diagrams contained in Fig. 7. In this case, the utility functions correspond to a n-person's Assurance game where protesting (C) is valued greater than passivity (D), under the condition that at least t percent of the other actors also choose to protest. The difference between the diagrams are seen by the different threshold levels t.

In a symmetrical Assurance game - i.e. everyone has the same threshold t - there are two different equilibrium points: everyone chooses C or everyone chooses D. Although everyone prefers cooperation, there is thus a risk that everyone will remain passive if this is the original situation. The same holds true in the asymmetric case, unless t = 0 for at least some of

the actors. Depending on the preference or threshold distribution, however, there may be other equilibria as well - e.g. a bimodal distribution with strongly separated peaks may under certain circumstances contain an equilibrium which only a fraction of the actors choose C.

However, besides the citizens who are conditionally prepared to protest, we can of course admit the possibility of other groups. All of the Eastern European states had civil rights organizations which continuously decided to protest despite hard repression. In addition, there was probably also a group of the population which for different reasons was never prepared to participate in demonstrations, regardless of how many others chose to protest. Both of these groups are characterized by the respective threshold levels t = 0 and t = 100. As long as most of the citizens have significantly higher threshold levels than those of the civil rights movement's core group, the majority of the citizens will most likely remain passive, keeping the opposition rather small and weak.

This sort of collective action problem is illustrated by the diagrams on the left in Fig. 8 (ef. Granovetter 1978). The function f(t) depicts the distribution of thresholds among the citizens and the function F(t) gives the cumulative portion of the citizens having threshold levels less than or equal to t. The functions thus describe how the inclination to protest is distributed among the population. Both of the conceivable outcomes (equilibrium points) are reached by observing in the diagram below where the curve intersects the 45° axis from above. The lowest point of intersection indicates a situation where members belonging to the core of civil rights groups are almost the only ones willing to protest, while the higher point of intersection represents a situation where everyone with Assurance game preferences (moderate actors) join the opposition and only loyalists to the regime and purely free-riders remain passive.

The diagrams to the right in Fig. 8 describe a quite different collective action problem. This difference comes from a rather small change in the preference or threshold distribution. However, this corresponds to a shift of the function F(t) slightly upward with the result that there is now only one intersection point with the 45° axis. The costs for participating in protest actions have presumably decreased, thus causing the threshold to decrease for the majority of citizens. The difference between the two pairs of diagrams thus might be small, in the sense that the inclination to protest might be similarly distributed. Yet, the differences between the outcomes of collective action can be considerable. In the diagrams on the left, the opposition mostly consists of a hardcore group of protesters, while the diagrams on the right indicates that the opposition includes a large majority of the population.

Changes which are small, even marginal at the individual level, therefore,
can cause radical changes at the collective level. My proposal is that the

dramatic changes in Eastern Europe can be understood in just that manner. The threshold model may also explain the snowball effect that we witnessed in several of the 'satellite states' - for example, the repeated demonstrations in such places as Leipzig and Wenzler Square that grew larger and larger with every new demonstration. Not only did this catch the members of the regime off guard, it was equally startling for many among the protesters. It is no wonder, therefore, that we, who were only observers, were also surprised.

REFERENCES

Ash, T. G. 1990. We the People. The Revolution of 89 Witnessed in Warsaw, Budapest, Berlin
and Prague. London: Granta Books.

Axelrod, R. 1984. The Evolution of Cooperation. New York: Basic Books.

Granovetter, M. 1978. Threshold Models of Collective Behavior', American Journal of
Sociology 83, 1420-1443.

Hadenius, A. 1988. 'Democracy and Capitalism: Collective Action Theory and Structural
Analysis', Scandinavian Political Studies 11, 21-43.

Hadenius, A. 1992. Democracy and Development. Cambridge: Cambridge University Press.

Hermansson, J. 1990. Spelteorins nytta. Om rationalitet i vetenskap och politik. Stockholm:
Almqvist & Wiksell International.

Lipset, S. M. 1959. 'Some Social Requisites of Democracy: Economic Development and
Political Legitimacy', American Political Science Review 53, 69-105.

Lipset, S. M. 1983 (1959). Political Man. London: Heinemann, expanded edition.

Moore, B. Jr. 1965. Social Origins of Dictatorship and Democracy. Lord and Peasant in the
Making of the Modern World. Boston: Beacon Press.

Nordlund, P. 1990. 'Demokrati och kapitalism. Organisationernas roll i demokratiseringsprocessen',
Department of Government, University of Uppsala, mimeo.

Przeworski, A. 1986. 'Some Problems in the Study of the Transition to Democracy', in O'Donell, G., Schmitter. P. C. & Whitehead, L., eds., Transitions from Authoritarian Rule. Prospects for Democracy. Part III: Comparative Perspectives. Baltimore: The Johns Hopkins University Press.

Przeworski, A. 1988. 'Democracy as a Contingent Outcome of Conflicts', in Elster, J. &
Slagstad, R., eds.. Constitutionalism and Democracy. Cambridge: Cambridge University
Press.

Przeworski, A. 1991. Democracy and the Market. Political and Economic Reforms in Eastern
Europe and Latin America. Cambridge: Cambridge University Press.

Rustow, D. A. 1970. 'Transitions to Democracy", Comparative Politics 2, 337-363.

Therborn, G. 1980. Kapitalism och rösträtt. Om den borgerliga demokratins uppkomst. Lund:
Arkiv.

Tsebelis, G. 1990. Nested Games. Rational Choice in Comparative Politics. Berkeley: University
California Press.