Columbia University

November, 1997

The challenge set by the broad historical and regional comparisons made in this volume is to discern how the growth of associations encourages or inhibits democratization. A strong civil society, and the 'social capital' embedded in its many associations, is often thought of as necessary to a strong democracy, and even capable of transforming an authoritarian regime to a democratic one. But the relationship is not simple. The pathways are many and complicated, as befits a complex and changing phenomenon such as civil society. In this note, I would like to sketch some of the relevant dimensions in order to sort out some of the different sorts of statements that can be made about the impact of associations on democracy.

By associations, we are referring to a relatively small set of the very large number of organized groups in society. The set is commonly defined negatively: it defined by being the residual category along each of four dimensions. They are organized groups which are non-governmental, non-profit, non-party and not families or clans. That excludes most groupings and organizations, but leaves a wide variety of cultural, political, professional, and volunteer organizations. Into this mix go many types, acknowledged to be quite different among themselves, but among which there is no commonly accepted categorization.

To some, associations are marginal precisely because they are not organizations forming the basic structures of society - government, the economy, the electoral and legislative process, families and communities. But, an increasing number of observers are seeing them as playing a crucial role in building and holding together polities - especially modern democracies. They invoke an image of free forming organizations built on a measure of voluntary participation around identities of the most varied kind. Associations appear to be the uncontrolled, yet purposeful and dynamic parts of modern democracies. Despite their status outside of the major political institutions, they seem to be able to provide a way of harnessing energies into the whole in a way that seems peculiarly democratic.

But, even if we accept their importance, their actions obviously do not always promote democracy. Many associations advocate non-democratic programs, exist comfortably in authoritarian regimes - professional associations in Nazi Germany, for example - and they often behave in ways which are disruptive or downright hostile to democracies - for example, armed militias in the United States today. The organizational tools of associations can be used by leaders to stifle expression as well as encourage it. Laws or personal influence may severely limit their free formation and adaptability. In short, their impact on democracy cannot be assumed, since different sets of associations have different consequences for democracy.

Further, similar sets may vary in their impact from time to time or place to place. Many seem to have positive consequences in some circumstances, and negative ones in others. Secret societies of liberal minded aristocrats may be a powerful force for democracy in an authoritarian monarchy, but be a reactionary anti-democratic force in a more open system. At another level, whether associations form blocs in solidarity with political parties may, on the one hand, be associated with democratization when forces were being mobilized to assault elitist-clientelist systems, as in 19th century Italy. Then, mobilizing the opposition required massing civil society support behind reformist or revolutionary political leadership. But such partisan alignments have led to institution-destroying polarization in other situations, such as Weimar Germany or Cold War-era Latin America. It depends on the circumstances, and the circumstances are exceedingly difficult to reduce to a simple set of contextual variables.

The theoretical challenge, then, is to specify when associations help bring about or support democracy. The project on which this note comments compares cases across space (Western and Eastern Europe) and time (the 19th and 20th centuries). In my own work, I add a concern for contemporary Latin America. Ideally, such a broad scope gives us the opportunity to perceive fundamental causal relations by seeing beyond the local 'noise'. But so far, this has been very difficult. It suffers a common malady in areas of social science theory where policy outcomes are important. Increasing generality is accompanied by diminishing specificity. Finding a particular constellation of factors with predictable effects in all cases seems to produce rather empty, inoperable generalizations such as, ‘democracy is more likely if the emerging associations are democratic’.

Perhaps it is impossible to reduce this confusion of organizations and situations to a simple model of relationships answering the core question of when civil society is democratizing. After all, the rediscovery of civil society in our understanding of politics is not so much rediscovering an interesting aspect of society, but rather is opening up analysis to a rich and complex arena of social life. Whereas once we might have been content to see only classes and interests, now we see multiple identities and structures of ideas and manifold participation, intertwined with markets and rapidly changing information networks. There is no reason to assume that the important aspects of this complex set of interrelationships can be reduced to some simple set of theoretical variables.

But we cannot close the door on trying to find simpler explanations and interpretations before we seriously consider the possibility that it is not the material, but our present form of conceptualization, evolved for other purposes, that makes it difficult to form simple propositions. Just as with the 'partisan alignment' variable mentioned above, common ways of classifying associations, for example, by whether they are 'left-wing ' or 'right-wing', professional or volunteer, 'modern' or 'traditional', or elite or grass-roots, may not be directly related to whether they are democratizing or not. We have to explore whether we can order patterns of associations in ways more useful to understanding their impact on democracy.

In this brief note, I want to make a small step in this direction by calling attention to one possible source of confusion, the sharp differences in the aspects of the associational patterns in civil society which have been identified as important for democratization. Although various observers argue that 'the rise of associations promotes democracy', they often have very different things in mind.

A useful assessment of the conditions promoting a democratizing impact of associations will have to include all of these dimensions, going beyond asking whether associations are consciously and effectively working for democracy to figuring out the impact of complex patterns of different and constantly changing associations. To assess the justice, stability and effectiveness of competitive democratic institutions requires looking at the broad configuration of civil society associations.

This means not becoming limited to a focus on the projects of NGOs and non-profits, and shifting to an emphasis to the broader operation of the system. The project focus is, of course, especially common among donors and activists in the NGO world. Although many NGO projects have goals distinct from democracy, such as disease control or infrastructure development, some are clearly related to politics, and, at least in the current environment, their goals are often to promote some democratic outcome. Human rights activism by NGOs has been impressive in the recent democratic transitions, and many efforts have been made to overcome drastic inequality by aid the very poor, or to improve equality of access to the legal process through legal reform. The discussion of the contribution of associations to democracy often confines itself to whether these projects are well conceived and effectively executed. But however important the immediate results are for donors, the impact on the regime of a collection of micro-projects - from freeing political prisoners to empowering the poor - we have to look not only at how the work of these groups 'scales-up' to the national level, but the impact of the associations in civil society taken together, in all their complexity and multiplicity. Even if the associations are focused on promoting the arts, caring for health, or advocating religious causes, the aggregate of their activities must also be considered for its effect on democratic institutions. Particular patterns of associations may promote polarization, expand or restrict the range of representation in society, or advance transparency and legitimacy in government actions. It is the aggregate impact which is the subject here.

Let us now turn to the various aspects of associations which may impact on the health of democratic institutions.

Despite what has just been said, the goals of associations, of course, can be important to competitive democratic institutions if their objectives are cast at the broadest level of creating, perfecting or defending such institutions, and they are effective at doing so. Many cases in this project document a common European pattern of ideologically liberal or democratic associations consciously founded to counteract aristocratic power and to discuss, propagandize for and conspire to promote democracy. In Poland under Communism, associations such as Solidarity and organizations associated with the Church rallied anti-totalitarian sentiment and provided a key part of the struggle that ultimately promoted democracy. Many organizations sought, sometimes successfully, to subvert military governments in Latin America in the 1970s and promote democracy, often using a human rights rhetoric.

Rather than spell out this rather obvious point, however, I would like to draw attention to the question of timing. Associations, as opposed to political parties or government commissions and agencies, are particularly important when the 'normal' vehicles for such activity cannot act, most obviously when the regime is non-democratic. Because of their flexibility, including their ability to disappear and reappear in new forms, associations are a crucial vehicle for mobilizing support for democracy in authoritarian, and in 'traditional' systems. Another way to put it is that when political parties - which seem more naturally the bearers of this 'constitutional' function - do not or cannot defend democracy, associations can play a key role.

The line between an association and a governmental agency or a political party is clearly porous. Political parties and other agencies may be organizations which have moved from being associations in the sense we are using the term, to the being officially sanctioned organizational parts of the regime, leaving the residual category defining our set of organizations. (Going the other direction, from official entity to non-governmental organization seems less likely, although privatization may be making that possible, too.) It may be that we should think of civil society associations not only the agent for mobilizing opposition, but also as a testing space for governmental or partisan organizations. I will return to this in discussing the dynamic patterns of associations.

In already democratic regimes, or even in periods of democratic consolidation, as opposed to the initial transition, associations with the goals of building democratic institutions decline in importance. Some associations may play an important 'watchdog' role, particularly for things like human rights violations and election fraud, and others may work on ‘perfecting democracy’. But the once very important role of associations in establishing and defending democratic institutions is less important, since they are joined and overshadowed by the institutions of government, the press, the courts and political parties.

The second approach to finding insights is to look at the aspects of the internal structure of associations. Democratization or democratic consolidation might occur not so much because of what associations tried to do, but because of the model they present or the experience they provide to members.

One form of such an argument says that it is the particular type of association prevalent at any one time that is important. An obvious, but too simple argument is that associations are democratizing when they are, in the main, structured democratically. People may be said to learn democracy, and democratic associations can be good schools, when other opportunities are unavailable. This seems to have been a factor in 19th Century English development analyzed by Morris, for example. The paternalistic forms of civil society associations in early Southern Italy or the authoritarian forms in 19th century German society may have had the opposite effect.

Whether this effect is important or not probably depends on the absence of other democratic practices in elections, public debate, local government and other arenas of politics in the society. Associations in this respect can again be seen as filling in for, or making a beginning in stimulating democracy, when the 'real' institutions of democracy are incapable of doing so. They are likely to have an importance in circumstances when democratic practices are either repressed, or have not supplanted more communal or clientelistic forms of political organization.

The desirability of democratically structured associations might even change significantly as democracy becomes established in elections and public debates. It may be more important for associations to be effective in accomplishing their goals than to be democratically run to convince citizens that they are being effectively represented. Effectiveness may demand some sort of responsibility to the constituencies, of course, in order to be able to mobilize resources, and sometimes the more democratic the organization the more efficient it is. But it may also be that internal decision making should be - and often is rather sharply hierarchical for the sake of a more solid democracy. Guarantees of ‘exit’ may be much more important as a device to protect participants from abuses that assurances of ‘voice’. We cannot specify that ‘the more democratic the internal structures of associations, the stronger the democracy’. Rather we need some criteria of the distribution of types of associations, which I will address briefly below.

An alternative approach is not to look for the currently dominant form of internal structure of associations, but to explore what all associations have in common. Whether or not they are internally democratic or not, perhaps the simple fact of association has an impact on democracy. Because the associations we are looking at are defined negatively, that is, those organizations which are not governmental, partisan, for-profit or familial, the common qualities found are likely to be rather vague or general. This approach is, however, the basis for the newly revived argument, which goes back at least to Tocqueville, that democracy is crucially supported by associational life, almost any associational life.

Rather than any special quality of the associations in a particular place, simply the practice of participation and working together according to rules inculcates values and teaches procedures that promote democracy - more or less independently of whether associations have participatory decision making styles or distribute benefits equitably (or, of course, whether or not their goals are to promote democracy). 'Social capital' is accumulated in the process of working together. Without the obligations of the relationships established with the state, without the restricted partisan commitments of political parties, without the 'bottom-line' pressures of business and with a voluntary quality missing in families and ethnic groups, people learn to make society. And, the argument implies, that society is likely to support democratic institutions.

This argument, as presented by Robert Putnam in Making Democracy Work, (Princeton, 1993), has been amply discussed, and I will not carry it further here, except to note that the plausibility of the argument probably works for suggesting what makes an already democratic system work, and not what makes authoritarian systems become democratic. Associations qua associations probably help any kind of regime 'work'. In the Italy that Putnam analyzed, the northern areas where he found that democracy worked also seem to be the areas where fascism flourished.

Let us turn to the third strategy for formulating the democratizing impacts of associations, and that is to ask what aspects of the way associations are linked to society can be democratizing? What networks of relationships with other associations, with national institutions and with their constituencies promote democracy and which inhibit it?

This strategy is promising because it takes a point of view common to attentive observers of non-governmental and voluntary organizations. To look at the goals or internal structure of the organizations, as we have been doing, does not focus on the things about associations uppermost in the minds of their active members and those studying them at close hand. That is not their ideology nor their organization, but their connections. To whom are they responsible? Where do they get their funding? How do they recruit new members? Whom do they see when they want something done? To see the roots of their effectiveness (or lack of it), one very often has to look at how they mobilize others, not how they organize their own resources. The answers to questions like these suggest that it might be important to describe associations in terms of the structure of their external relationships.

One strategy for analyzing this external structuring would be to proceed as we did in the case in discussing 'social capital', and to identify characteristics of links which are true of all associations. Here, the negative definition of the set of associations again makes the task problematic, but in a sense, the negative quality itself suggests something about associations. We do not have any notion as general as 'social capital', but not being part of the structures of government, the party system, or business can be thought of in terms of a reservoir of autonomy. Although it would be foolish to argue that all associations are independent, their ability to negotiate their connections may be a factor in promoting democracy, at least in contrast to extremes of totalitarian or clientelistic ties. As I said in the opening paragraphs, there is something about the self-determining, free wheeling character of associations, that is, their lack of confining links to other centers of power, which gives associations their attractiveness as bringing - or foretelling - democracy. But the impact would no doubt be very general, and given the ease with which any particular organizations may be coopted or created by economic firms, local power holders or the state, however, it seems risky to place too much emphasis on this general character of associational life as a guarantor of democracy.

A second approach would be to look at the patterns of links characteristic of a particular time and place. This is where one finds a rich variety of concepts in use, for example, the concepts 'clientelism' and 'corporatism', as well as the notion of ‘solidarity’ among blocs of associations, each linked to a political party. To be sure, the first two are sometimes used to describe situations in which associations are severely limited. Corporatism is applied to situations in which any association-like organizations are assimilated into the state, that is, they become governmental. Clientelism could be limited to configurations where associations are dominated by a more personal, family-like ties. But both may be, and have been extended to situations where associations clearly exist, but have particular linkage patterns. We often call a situation clientelistic when an association, through its leadership, engages in the kind of unequal exchange of services and favors with a protector or patron in the government or the private sector that parallels the more conventional ‘clientelism’. This extended sense of clientelism could cover a wide range of situations, from early Anglo-American philanthropic institutions to Mexican trade unions in the recent past.

An extended definition of corporatism, too, is used to highlight the variation in the degree to which associations, even if they remain nominally autonomous, are dependent on, or are 'colonized' by government bureaucracies, or through regulation or self-selection, form privileged, yet dependent links with government agencies. The terms 'iron triangles' and 'bureaucratic rings' have been used to suggest something of this sort.

But it is not clear that either concept helps us in our task. The relation of clientelism or corporatism to democracy, in either its narrow or extended sense, is ambiguous. With regard to corporatism, for example, it might be the antithesis of democracy, when it refers to entrenched privilege. But if one focuses on representation, corporatism in Europe and perhaps even more in later industrializing countries, has signified the incorporation of emerging groups - principally industrial workers. It has provided a kind of 'affirmative action', giving influence to groups which are less able to mobilize the resources necessary to have strong associations.

From the angle of institutionalization, corporatism seems to have been a way to stabilize democratic, or semi-democratic institutions and neutralize polarization. Corporate institutions were often meant to build representatives of competing class-based associational blocs into a set of official organizations and procedures in order to convert potentially destructive conflict into peaceful negotiations. On the other hand, corporatist relationships may be very disruptive, if they are seen as preventing change and adaptation. This is the contemporary take on corporatism, as the bogey man of liberal reformers seeking to privatize and free the market. Although the ruling value in the new discourse is more economic efficiency than democracy, the inflexibility of associational links with the state has the implication that it will undermine democracy. If the corporatist relationships become exclusionary of, say, new religious, ethnic or local identities and interests, it could become destabilizing, rather than institutionalizing. In the end, corporatism - and similar arguments could be made about clientelism - is a relevant way of understanding the links of associations to the state, but hardly a simple variable linked to democratization.

As suggested earlier, similar kinds of messily contingent - and thus not very conclusive - statements can be made about the patterns of stable, solidary alliances among associations and with political parties. Bloc formation, particularly when associated with ideological Lager can be a sign of polarization. In the classic European cases, when the Church and the Left are constituted by a dense network of associations which create separate life environments for their members consciously designed to associate them with a 'movement' or a 'community', it promotes, even constitutes social and political polarization. Whether this supports or undermines democracy, of course, depends on a) whether there is a viable transformational project such a bloc might support, or b) in on-going democracies, whether the blocs can cooperate effectively in a ‘consociational democracy’. On the other hand, flexible and changing coalitions among associations and political parties may limit the threats of polarization, but open up questions of governability and stability.

We may think of the links involved as those between the associations and the legislative/policy process - including not only parties but also legislatures and those offices of the executive which share in policy making. Pluralists have long focused on the way in which 'interest groups' are built into the 'lobbying' process, for example. Both the range of representation and the successful institutionalization of the democratic process have seemed to many to rely on the capacity of parties to serve some kind of organizing function ‘between’ the associations which form in civil society and the legislative process. The old clientelism image of political parties serving as ‘interest aggregators’ suggested a very strong link to strong parties which would ‘process’ demands.

But with the - temporary or permanent - decline of the programmatically strong political parties, the vision of political party role has been revised. In the United States it has long been clear that ‘interest groups’ do in fact by-pass political parties to lobby directly with individual legislators and government agencies. Whether there is any effective way of regularizing the process of aggregating these interests so that democratic values, and institutional legitimacy are preserved, remains an open question, demonstrated by the endless argumentation in the United States about regulating lobbying and shaping campaign financing. In other countries where traditions are more statist, such as in Latin America, building links between NGOs and the policy process is a practical, immediate necessity. Aside from allowing free access and competition, there does not yet seem to be much in the way of norms to guide institution builders. Smoothly functioning modern democracies do seem to require that parties coordinate elections and the legislative process. Just what role they have, and therefore, how associations should be linked to them, in interest bargaining and policy debates is less clear. I know of no convincing norm of just what these relationships should be to maintain the representativeness or the stability of democratic institutions. What seems clear, given the character of associations, is that any stable democratic pattern will have to go beyond building a few representative organizations into the state in the manner of corporatism or tying a few organizations into political parties in the model of party mobilization. It would have to find ways of handling a multiplicity of ties that are constantly changing.

One aspect which may play a role in modern democracies is the 'political service sector' that develops around efforts to influence policy - media consultants, professional lobbyists, public relations experts, poll takers, planners, consultants and the like. In an earlier era, professionalization of the political process was associated by Weber with the professionalization of party organizations with skills parallel to these, but now associational forms - often for profit, of course - are a significant factor. It seems plausible to argue, even though it makes the argument just that much more complicated, that the impact on associational links with democracy depends on issues such as equal access to these ‘political services’ and mechanisms and incentives which lead service providers to promote the general good, and to maintain legitimacy. The United States, which seems to lead the world in the development of this political service sector, apparently is also far from resolving these dilemmas, judging from the contemporary discourse condemning ‘lobbyists’ and ‘special interests’.

This brings us to the fourth aspect of associations that might be considered as having an impact on democratization, the distribution of associational types. What combination of varied types of associations, taken as a whole, may be democratizing? So far we have assumed that some active, dominant associations will be crucial, or that those things that all associations share will be important. But this approach misses the most dramatic characteristic of associations in any particular society - their multiplicity and diversity. The associations in any one actual configuration may vary on all the three dimensions noted so far. Associations in any one country have varied goals to be accomplished in different arenas (political and social, national and local, etc.), have different internal structures (democratic and authoritarian, mass membership and professional, large and small bureaucracies, etc.), and varied external links (with and without patrons, with and without policy access, etc.) The question here is whether it is possible to describe particular distributions of these characteristics which would make a difference for democracy.

One place to start would be the arguments that some distributions of associations promote democracy to the extent that the range of goals they articulate represents as large as possible range of all the interests of groups in society. But ‘all the interests of groups in society ‘ is not a meaningful standard without some heroic assumptions about the real nature of interests, such as provided in a simple class model of society. Even taking ‘interests’ to be defined as whatever people say their interests are would set up a moving target, and would leave out all those interests which people ‘don't know they have’, but might learn, for example concerning complex ecological threats.

Even assuming that one can overcome these formidable obstacles, it may not make sense to establish a norm that the largest possible number of interests ought to be represented by associations, since it is not clear whether associations ought to be considered - as the old functionalist model would seem to have it - as always the most appropriate 'articulator' of interests in the political system. It might be a form adapted to some interests and not others, and thus democracy would be promoted by combination with other forms. Working class interests should perhaps be represented in political parties, general public interests through the watchdogs of the press, etc., business interests through the working of the market, and so forth. Finally, many associations are not political in the sense of advocating or defending positions in the political process. No simple link between democratization and the number of goals and interests of associations seems possible.

Rather, one would need to look for some kind of match between the sorts of interests and the forms of associations and other intermediaries. The explosion of both national and international non-governmental organizations in the field of 'development' is, in theory, at least, an example of this kind of matching. The poor, the isolated and the marginalized in developing countries are thought to need a set of associations which combine grassroots mobilization and consciousness-raising, together with an organization that combines sophisticated, problem solving experts and with links to both national and international sources of assistance. A group of small businessmen in the inner city would need a different combination, and a religious group seeking both salvation and a peaceful relation with society quite a different kind of association. Democracy might be seen as being hindered when associations dealing with the poor are paternalistic and not empowering, or when associations mobilizing religious groups reinforce their isolation from others in the community. Although the principle of having a variety of associations which somehow ‘fit’ the needs of the interests is easily stated, it is apparent that it would be very hard to formulate a general hypothesis about what would produce results and what would not. It is the kind of challenge which suggests some sort of ‘market’ in associational structures and linkages, with the task then becoming to specify what constitutes a ‘perfect market’.

Another way of thinking about the distribution of varied associations relates to the cognitive side of politics. Politics can be thought of as a negotiation/struggle to decide who get what. It is also, however , also a problem-solving and reality-defining process. There have been a variety of suggestions that associations such as Universities, policy think-tanks and research agencies play a significant role in this process through shaping public perceptions, setting agendas, bringing new approaches to old problems to the point where they can be discussed and adopted, and many other ways. It is plausible to suggest that a healthy democracy may profit from particular configurations of these, along with grassroots organizations and interest-lobbying organizations.

The mix varies according to the situation, as with so many of these variables. In authoritarian systems beginning the change, non-governmental groups unconnected to parties, may be the main locus of such debate. The struggle to establish the possibility of alternatives to authoritarian systems, or to explore the connection between, say, Communist rule and economic stagnation in Eastern Europe, took place in clubs and forums and roundtables. The array of associations most likely to promote democracy included not just advocacy groups and opposition-mobilizing groups, but also ones that make space for discussion and finds ways of making it consequential - in influencing policy, parties or popular mobilization.

In systems already democratic in some sense, the regular political institutions and the party systems are no doubt the most central arenas of consequential debate, but in a complex policy process study centers, issue groups, specialized libraries and data bases and many other entities begin to supplement government agencies, party meetings, and parliamentary debates. Policy networks create ties between associations others, and by some accounts take over much of the cognitive side of politics.

What mix of associations will be democratizing? What combinations of think tanks and policy fora and interest associations will promote a debate which will give due respect to the broadest representation and avoid paralyzing polarization and delegitimizing ineffectiveness?

The final dimension of the patterns of associations I wish to discuss concerns not the static distribution of associational types, but the way that distribution forms, changes and decays over time. One of the notable characteristics of associations, in contrast to political parties, government agencies, clans or families is the frequency with which they come and go. They can be created in a short time, and can dissolve rapidly. Of course, some associations endure for a very long time and, some parties and government agencies change frequently. But the standards for judging when parties or government agencies are unstable are much higher than those applied to the thousands of associations that rise and fall. People expect more change. A fluid and changeable set of associations is less disruptive to the polity than, say, a fluid and changeable party system.

Conversely, permanence is a virtue for parties and government agencies, but may not be for associations. We commonly think of the durability of the American two-party system as a strength. The staying power of charitable associations, sports clubs and the like are seen as signs of strength of that particular group, not usually of the system itself. The durability of interest and lobbying groups - and especially their links with those in power - is likely to be thought of in negative terms as 'entrenched'.

If they do change, then, how do they change, and what patterns of change support democratization by inhibiting polarization, expanding representation and promoting the stability of democratic institutions. One pattern that can be derived from the cases in this book, involves associations acting as, in effect, experiments and 'pre-tests' for politically relevant agencies and organizations. For example, there was the emergence in Europe of associations foreshadowing modern political parties. Cases noted in this book look back behind the major parties and find organizations, sometimes clandestine, which evolved and ultimately were given recognition as political parties. Presumably many such associations withered when they failed to find the right formulas and strategies to draw support or avoid repression.

Associations might thus be thought of as playing a role in the shaping of democratic institutions by constituting a kind of testing ground in some evolutionary scheme of institutional formation. The most democratizing role of associations would be one in which the 'selection' process favored the development of effective and democratically oriented parties. An anti-democratic process would be one in which the outcome was parties which were violent or authoritarian or simply ineffective. This would not require that all associations that sought political status were democratic. In fact the reverse might be useful for democracy. Democratic outcomes might emerge in part because violent or authoritarian tendencies in some groupings were exposed and isolated. Forming an association is usually seen as a way of mobilizing support and resources behind a cause, but even for that reason, unless it is a successfully secret organization, it is also a way of publicly identifying its goals and aims which may be the basis for rejection, condemning the group to a marginal status. It is possible that neo-Nazis in Germany and anti-government militia groups in the US, especially when they publicly aligning themselves with racial hate, are in fact caught in such a process of exposure and isolation. The groups are clearly not associations promoting democracy, but the process maybe an important one that safeguards democratic institutions.

Scholars have spent much energy in trying to predict when such movements become significant parties - as in the case of the original National Socialist Party in Germany. There is no simple generalization here. It is not just a question of how associations form, but many complex aspects of the context into which they emerge. But it is worth thinking about this process not just as a question of social movements, but also as potentially a complicated safety valve for democracies.

Another well known example of this foreshadowing function is the way charitable organizations anticipated state welfare agencies. Private philanthropic organizations developed schemes for rendering assistance which were, in many cases, later taken over and refined by the state. Others languished or disappeared. It is possible to see associations, again, as a testing ground from an evolutionary point of view. If one assumes that successful democratization requires an effective way of delivering services to the poor, then the role may be important by virtue of the way they test new ideas or new situations. Either the absence of a collective effort to provide for welfare, or a system which destroys legitimacy through corruption or biased distribution would provide a model of an anti-democratizing process. It is possible to see the current crop of poverty-alleviating national and international NGOs as experiments to be absorbed into, or be imitated by, the government. Possible negative outcomes could be that being international, these groups do not develop means of accountability, or that erratic funding patterns fail to establish a solid basis for a transition in the long run (whatever good they might do in the short run.)

The 'foreshadowing' function may not always lead to the establishment of a 'normal' formal governmental welfare organization. The privatization of welfare and the importance of development of social programs through nongovernmental organizations suggests the development of a different form of welfare network, involving private, public, domestic, international and local entities. This accumulation of organizations may consolidate itself into some new kind of complex semi-governmental network, rather than into government agencies. If it did so, associations would once again be playing ‘midwives’, but this time to new forms of organization of welfare of which they would be one part.

These sorts of 'foreshadowings' imply a pattern of change that progresses through a significant period of historical time. But associations form and reform constantly. Beyond potential steps towards the future, does this daily dynamic of associational change have a significance for democracy? Does it matter what the principles are that describe, or, if these are consciously chosen and applied, regulate what sorts of organizations are formed and which rise to influence and which are marginalized or are abandoned? It does if we understand there to be some significance in that organizational marketplace.

In a way parallel to the long term historical 'experiments' in party organization or social welfare we have just discussed, the everyday rise and fall of groups has something to do with problem-solving and innovation in policy. An organization created to distribute condoms to fight AIDS, or another to rebuild and distribute housing in a city after with a major earthquake, or another to exploit a new technology of gathering information, may constitute a probing and testing of ideas which may affect the way in which public and private officials think about and act on social problems. Some of these ideas succeed in establishing a practice, or in changing the public perception of the problem, many others fail. Some of the freewheeling experimentation that takes place with start up companies in the business world, testing out new products, new strategies and new technologies, can be thought of as having a parallel in policy making.

An aspect of this pattern of group formation which may more obviously relate to democratic norms is the way associations can form to act as highly specific feedback mechanisms. Citizen's groups forming in response to failure and inefficiencies of government provides a straightforward example. In the sweep of history, working class based associations have been seen as being created to balance the 'natural' organization of the established elites. In a complex world in which interests are not so simple, a system where associations form rapidly in response to hundred of new 'imbalances' depending on constantly redefined conflicts - and dissolve when the balance is no longer relevant - can be essential for guaranteeing a representative system.

What patterns of associational dynamics promote innovation and feedback, and thus, democracy, and what are not? This is a challenging question, not only theoretically, but for the practical task of legislation and building regulatory models which will recognize and promote patterns of associational formation which are positive in their consequences. In many eras, governments have tried to suppress them in the cause of unity. In the modern world, multiple and changing associations seem inevitable. Asking what sorts of institutions would encourage the best for the sake of democracy, requires knowing the answers to what patterns produce democracy.

In conclusion, the dimensions of association which might have an impact on democratization are many. It is clear that in each aspect, associations do not always promote democracy, and that the conditions under which they do are complex. Theory construction in this field is challenging. As I said in the introduction to this note, there may not be any parsimonious theories that cover this field. But some order can be brought to the task, at least, by identifying the various aspects, and distinguishing the different sorts of questions they raise.

The proliferation of civil society associations that has so taken the imagination of policy makers and social theorists is here to stay, short of some new difficult-to-imagine totalitarianism. Associations are no panacea for bringing about democracy, nor for rendering complex societies fully democratic. But they seem inevitably to accompany the growth of democracy and to be part of any modern system. We will need as much theoretical imagination as we can muster to understand them.