公 法 评 论

et revelabitur quasi aqua iudicium et iustitia quasi torrens fortis



Joshua Cohen and Charles Sabel[1]


In this essay we defend a form of democracy that we will call "directly-deliberative polyarchy." We argue that it is an attractive kind of radical, participatory democracy with problem-solving capacities useful under current conditions and unavailable to representative systems. In directly deliberative polyarchy, collective decisions are made through public deliberation in arenas open to citizens who use public services, or who are otherwise regulated by public decisions. But in deciding, those citizens must examine their own choices in the light of the relevant deliberations and experiences of others facing similar problems in comparable jurisdictions or subdivisions of government. Ideally, then, directly deliberative polyarchy combines the advantages of local learning and self-government with the advantages (and discipline) of wider social learning and heightened political accountability that result when the outcomes of many concurrent experiments are pooled to permit public scrutiny of the effectiveness of strategies and leaders.

One starting point for our argument is a commonplace of contemporary political debate: that current economic and political institutions are not solving problems they are supposed to solve, in areas of employment, economic growth, income security, education, training, environmental regulation, poverty, housing, social service delivery, or even basic personal safety. A second point of departure is the intrinsic appeal of collective decision-making that proceeds through direct participation by and reason-giving between and among free and equal citizens. Directly-deliberative polyarchy is the natural consequence of both beginnings: desirable both in itself and as a problem-solver. That is what we hope to show, or at least make plausible.

But obstacles lie along both paths. However commonplace the recognition of institutional failures in problem-solving, the conventional categories used to explain those failures and defend strategies of repair obscure important developments that suggest the plausibility of a directly-deliberative alternative. Moreover, gestures at radical democracy invite skeptical observations about the intrusive and opressive aspects of localism, or simply about the burdens of endless evening meetings. And the force of such observations will only be deepened by adding in improbable claims about the problem-solving powers of participatory self-government in vast, heterogeneous societies. To take the chill of manifest implausibility from our project, therefore, we start by discussing the limits of current debate as revealed in the promising developments it overlooks (or misrepresents), and specifying the criticisms of radical democracy to which we must respond if we are to offer more than a consoling prospect for democrats in hard times.[2]

Consider first the conventional interpretations of institutional failure, and the projects of reconstruction associated with them. On one interpretation these failures reveal the limits of state regulation, and suggest the promise of social life freed of the political fetters on exchanges amnong individuals and so fully ordered by market principles. To fulfill this promise, we need only remove the regulatory detritus of twentieth century political failure. And that means constraining government from doing anything wrong by constraining it from doing much at all: by fracturing political power both vertically and horizontally, setting stricter constitutional limits on government, and interpreting the rule of law as a law of rigid rules.[3]

A counter-interpretation sees unrestricted market ordering as a threat to political arrangements carefully crafted earlier in this century to provide goods collectively that will not be provided individually, protect the weak from the strong, and ensure that our destinies in life are not determined by the vicissitudes of market success. The correlative political project is to protect the increasingly fugitive state from attack, and hope that a turn in the political cycle will restore public confidence in collective political action.[4]

Yet a third interpretation condemns the false dichotomy of state and market. Well-functioning markets and well-ordered political institutions can, it observes, be mutually reinforcing. Both, however, require prior bonds of trust that can be undermined, but not created or sustained, by self-interested market exchange or selfishly exercised political influence. Those bonds depend, rather, on protecting family, church, and voluntary association--the pre-contractual, pre-political background responsible for accumulating the social capital we need to preserve our economic and political artifice.[5] But because such social solidarities are understood as anterior to both economy and state--preconditions for the proper functioning of both (on any conception of such proper functioning)--the implications of such rebuilding for economic or political institutions are entirely indeterminate.

We are skeptical about these contending diagnoses and remedies. A number of emergent solutions to problems as varied as public safety and public education seem not to result from either a shift in the balance between "state" and "market" forms of coordination, or a shift in the balance between these taken together and civil society. Instead of the state's retreat or the market's resurgence, or even the transfer of functions from government to non-governmental organizations, secondary associations, civil society more broadly, or some other third something alongside state and market, these phenomena suggest a set of changes that disrupt those categories, the social-political boundaries they express, and the associated idea that an effective polity is one that balances responsibilities optimally among the arrangements that fall within those boundaries.

Consider, for example, community policing: a strategy for enhancing public security that features a return of police officers to particular beats, regular discussions between them and organized bodies in the communities they are policing, and regular coordination between those bodies and agencies providing other services that bear on controlling crime.[6] Or consider forms of school decentralization that--while shrinking school size and permitting parents to choose schools--also replace close controls by central bureaucracies with governance mechanisms in which teachers and parents play a central role. Or arrangements for local and regional economic development, that include strong components of training and service provision, and whose governance includes local community interests, service providers, representatives of more encompassing organizations, as well as local representatives of regional or national government. Or, closely related to these arrangements, consider firm-supplier relations that transcend episodic exchange to establish long-term collaboration coordinated through regular discussions, disciplined by reference to officially recognized standards--which standards themselves commonly emerge in regular discussions between and among groups of firms and suppliers, and may include public research, technical assistance, or training facilities as well. [7]

These new arrangements suggest troubles for the standard categories of analysis and remedy. The arrangements are not conventionally public because, in solving problems, they operate autonomously from the dictates of legislatures or public agencies; they are not conventionally private in that they do exercise collective problem-solving powers, and their governance works through discussion among citizens rather than the assignment of ownership rights; at the same time, they do not presuppose a successful, densely-organized, trust-inspiring network of associations. Indeed, they often emerge precisely against a background of associative distress. Nor are these new arrangements mere intellectual curiosities. They are attractive because they appear to foster two fundamental democratic values--deliberation and direct citizen participation--while potentially offering advantages as problem-solvers that programs conceived within the limits of conventional representative democracies do not. Indeed, if the same properties make them both democratically and pragmatically attractive, we would have a compelling case for the novel form of public governance that we call directly-deliberative polyarchy.

Because these new governance arrangements resonate so strongly with the (often implicit) programmatic suggestions associated with radical democratic criticisms of the modern state, a straightforward and appealing generalization of them seems at hand. Congenitally hostile to the market inequalities and economic subordination, but always suspicious of an overweening state as the best defense against them, radical democracy emphasized the deficiencies of centralized power, virtues of decentralization, expressive and instrumental values of participation, and values of citizen discussion both as an intrinsically attractive form of politics and as a good method of problem-solving.[8]

But evoking the core features of the radical democratic tradition--its emphasis on direct participation and deliberation--immediately suggests three lines of criticism: First, that in a large scale political system widespread participation in decision-making is organizationally or administratively impossible, so the ideal of radical democracy is vacuous. Second, if participation could be ensured, the mutual reason-giving that constitutes deliberation depends on a higher degree of homogeneity among citizens than can reasonably be assumed in a large-scale, pluralistic democracy. And third, direct decision-making requires a localism incompatible with the constitutional safeguards needed to ensure equal treatment for citizens.

Here, the threads of our argument come together: Guided by the experience of emerging problem-solving institutions and mindful of the values associated with radical democracy, our aim is to sketch the alternative social-political world of directly-deliberative polyarchy in sufficient detail to meet these objections. We start (section 2) by presenting an account of the ideal of democracy and explaining why the properties of directness and deliberativeness make highly participatory forms of direct democracy especially compelling realizations of that ideal. To be sure, the classical institutions of direct, assembly democracy are unavailable as realizations of directness and deliberativeness. But by separating those properties from their familiar institutional expressions, we suggest that they might still guide current institutional reform. In section 3 we describe the current practical impasse in problem-solving, and propose that the roots of that impasse lie in part in the mismatch between current arrangements of constitutional democracy and fundamental properties of unsolved problems. In section 4 we describe the new form of state that would result from the generalization of deliberative problem-solving arrangements and foster their successful operation. By way of conclusion, we contrast, in section 5, the idea of the democratization of everyday life at the heart of directly deliberative polyarchy with the idea of radical democracy as a rear-guard protest against the subversion of collective autonomy by technical constraint articulated in the work of Habermas, Arendt, and others influenced by them.

Our approach is conjectural. We are guided by political values, a view of current failures, and some hunches about promising developments. But our aim is neither to articulate a set of normative principles and deduce institutional conclusions from them, nor to predict the course of current institutional evolution. Still less is it to explain fully the causes of the failures of representative democracy or the origins of the new arrangements. Instead, we take the very existence of these arrangements as a sign of the insufficiency of theories that would explain what democracy can do, and try to imagine what democracy could be from the vantage point of the possibilities suggested by their presence.


Democracy is a political ideal that applies in the first instance to arrangements for making binding collective decisions.[9] Generally speaking, such arrangements are democratic just in case they ensure that the authorization to exercise public power--and that exercise itself--arises from collective decisions by the citizens over whom that power is exercised.

The ideal of democracy comes in several variants, which are associated with different interpretations of "authorization" and "collective decision." Our principal aim in this section is to sketch and defend a directly-deliberative interpretation of the democratic ideal. We begin by exploring the virtues associated with democracy quite generally, and then consider the special advantages of directly-deliberative as against representative-aggregative democracy. We conclude by returning to the conventional criticisms of directly-deliberative democracy, thus setting the stage for our later efforts to describe a form of radical democracy that can answer these criticisms.

Before pursuing these competing interpretations, however, we want to clarify the relationship between those democratic ideals and conventional institutions of electoral democracy. Following Robert Dahl, we use the term "polyarchy" to cover political systems in which virtually all adults have rights of suffrage, political expression, association, and office-holding, as well as access to diverse sources of information; in which elected officials control public policy; and citizens choose those officials through free and fair elections.[10] Continuing to follow Dahl (and subsequent writers), we note that polyarchy has considerable value, both for its intrinsic fairness and instrumental success in keeping the peace and protecting certain basic rights. It is not of value simply because it establishes conditions required for achieving some greater ideal. Furthermore, under modern circumstances of political scale and social pluralism, polyarchal institutions are necessary for realizing fully an ideal of democracy, however that ideal is specified. Though polyarchies can be more or less democratic, making them more so does not require negating, sublating, or otherwise transcending the political institutions definitive of polyarchy. This said, however, polyarchy is insufficient for full democracy--or full political equality--because, for example, it is compatible with inequalities in opportunities for effective political influence that would be condemned by any plausible statement of the ideal.

Building on these three considerations, then, we use the term "directly-deliberative polyarchy" for a form of polyarchy distinguished by the presence of a substantial degree of directly-deliberative problem-solving. (As we will see later, this presence transforms the role and functioning of conventional polyarchic institutions.) And we use the term "directly-deliberative democracy" for our account of the democratic ideal: fully democratic arrangements that feature a substantial degree of directly-deliberative problem-solving. Directly-deliberative polyarchies, then, more closely approximate the ideal of directly-deliberative democracy than do existing forms of polyarchy, but--like polyarchies sans phrase--need not have the entire range of qualities necessary for full democracy.

Three Virtues

Consider an association that needs to make binding collective decisions. Assume, too, that the members of the association, whose conduct is to be regulated by those decisions, are regarded as free and equal. Very roughly, they are equal in all possessing, to a minimally sufficient degree, the capacities that make persons free.[11] They are free in that they have the capacity to regulate their conduct by reference to a conception of justice and set of ends with which they identify; to use practical reason to bring them to bear on individual and collective conduct; to reflect on the plausibility of both; and to adjust their aims to the requirements of justice. Why should binding collective conditions for such an association be made democratically?[12]

Recall that democracies, abstractly conceived, are systems in which decisions to exercise collective power are made in institutions that treat those subject to such power as its ultimate authors. To that end, democracies need at least to satisfy the conditions of polyarchy--to protect constitutive liberties of participation, association, and political expression, establish direct or indirect electoral control of public policy, and ensure adequate information. That said, the reasons for democratic authorization divide naturally into goods intrinsic to the process and goods that arguably result from it.[13]

First, democratic arrangements have the intrinsic virtue of treating those who are subject to binding collective decisions with respect, as free and equal: "the person of the humblest citizen is as sacred and inviolable as that of the first magistrate."[14] Thus, the judgments of citizens, who are expected to govern their conduct in accordance with collective decisions, are treated by the processes of collective decision as equally authoritative. Though decisions will rarely, if ever, be unanimous, no one's judgment of the proper rules of cooperation is treated as having greater weight. Given the background conception of citizens as free and equal, any assignment of differential weights to the views of different citizens is a form of disrespect (unless it can be provided with a suitable justification).[15] Furthermore, the protection of the basic expressive and associative liberties establishes favorable conditions for reflecting on the plausibility of alternative views about justice, and which ends are worth pursuing. And the assurance of adequate and diverse information contributes to the exercise of practical reason, in working out the implications of conceptions of justice and of suitable ends.

Second, democratic arrangements are instrumentally important: they help protect the basic rights of citizens and in advancing their interests, as defined by the ends and projects with which they identify. Thus, democracies provide mechanisms for regular, popular authorization of exercises of public power: in representative democracies, that means (at a minimum) regular elections of legislators, in a direct democracy it means regular opportunities to review past decisions and evaluate the performance of officials responsible for implementing those decisions. Such regular renewal serves to make the exercise of collective power accountable to the governed in the formal sense that the governed can impose sanctions of removal from office on government. More fundamentally, an accountable system for the exercise of collective power, in which citizens are treated as equals, arguably helps ensure peaceful transitions of power, restrain the exercise of power by protecting majorities from minority rule, avoid at least some egregious violations of minority rights, and foster greater responsiveness of government to the governed.[16]

Both arguments--intrinsic and instrumental--are strengthened when we consider the educative aspects of democracy. Thus, by establishing the position of equal citizen, with associated entitlements to participate in determining the terms of association, democratic arrangements not only respect but also provide instruction in fundamental political values--in particular, the value of equality itself, and the conception of citizens as free and equal. By participating, citizens acquire political ideas in light of which democracy itself is justified. Furthermore, by opening debate to all, and addressing problems through public discussion--rather than through market exchange or bureaucratic command--democracy not only assumes adequate information, but helps to ensure it. Democracy provides a way to pool dispersed information relevant to problem-solving, and explore the range of possible solutions to practical problems: in short, a framework for collective learning. As Rawls puts it, within a democracy: "Discussion is a way of combining information and enlarging the range of arguments. At least in the course of time, the effects of common deliberation seem bound to improve matters."[17]

Two Dimensions of Democracy

We said that political institutions are democratic just in case they link the authorization to exercise public power--and that exercise itself--to collective decisions of citizens, understood as free and equal. There are, of course, very different ways to interpret this abstract ideal of democracy, corresponding to different interpretations of the notions of collective and authorization.

Democratic collective decision-making can be either aggregative or deliberative, depending on how we interpret the requirement that collective decisions treat citizens as equals. Understood aggregatively, a democratic decision is collective just in case the procedure gives equal consideration to the interests of each person: it treats people as equals by giving their interests equal weight in making a binding decision. Conventional rationales for majority-rule as a method of collective decision rest on the idea that it gives direct expression to this requirement of equal consideration.[18]

Understood deliberatively, democratic decisions are collective just in case they proceed on the basis of free public reasoning among equals: Interests unsupported by considerations that convince others carry no weight. Put otherwise, in deliberative decision-making, decisions are to be supported by reasons acceptable to others in the polity of decision makers; the mere fact that decisions are supported by a majority of citizens, deciding on the basis of their interests, does not suffice to show that the decisions are democratically authorized. On the deliberative interpretation, then, democracy is a framework of social and institutional conditions that both facilitates free discussion among equal citizens by providing favorable conditions for expression, association, discussion; and ties the authorization to exercise public power--and the exercise itself--to such discussion, by establishing a framework ensuring the responsiveness and accountability of political power to it.

As to authorization to make collective decisions, we have again two distinct understandings: In representative democracy, popular authorization proceeds through a choice by citizens of representatives who decide on content of public decisions. Citizens vote as individuals for persons who will participate in making binding collective choices in an aggregative or deliberative legislature. In direct democracy, citizens authorize public action by making deciding on the substance of public policy. Again, those direct decisions can be made either aggregatively, as some argue is true in referenda, because of their yes/no structure,[19] or deliberatively, as in an idealized town meeting, in which decisions on policy take place after debate on the merits. The essential distinction between direct and representative is not the level of participation, but the topic on the agenda: direct requires decisions on substance, whereas representative involves choice on legislators who decide on substance.[20]


Forms of democracy that are deliberative-direct seem especially attractive in view of the three reasons for endorsing a democracy as a way to make binding collective decisions. While those reasons support democracy generally, they provide especially strong support for a deliberative-direct democracy.

Consider, for example, the idea that democratic procedures are desirable because they treat citizens with respect, as free and equal. The deliberative conception offers a particularly forceful rendering of this condition. For suppose all participants support their views with considerations that others regard as relevant and appropriate. Nevertheless, because of differences in views about the weight of those considerations, there is disagreement about the right outcome. Still, the minority can scarcely contest the fundamental legitimacy of the decision. After all, not only the procedures but the arguments themselves treat each as well as they can reasonably demand.[21] Thus the deliberative conception of collective decision extends the idea of treating people with respect from rights and procedures to justifications themselves. A similarly strong case can be made for directly-deliberative decision-making on the basis of the arguments about instrumental benefits and learning. But we postpone consideration of these until we have said more about the operations of directly-deliberative polyarchy.

Despite these virtues as an expression of democratic values, radical democracy--a system with high degrees of directness and deliberativeness--is subject, we noted earlier, to a series of closely related criticisms: that under modern conditions of political scale, it is not feasible, except as local pockets of direct citizen engagement; that even within those pockets--and certainly as scale increases--cultural heterogeneity thwarts the mutual reason-giving that defines public deliberation; and that the localism characteristic of radical-democratic schemes leaves local minorities at the mercy of their locality.

The starting point of these criticisms is the identification of radical democracy with direct assembly democracy, and especially with the Greek polis as both the ideal and practical inspiration for modern critics of centralized, representative democracy. In a direct assembly democracy, legislative power--and the power to review conduct of all officials--is vested in a body which all citizens may attend. In the case of the Athenian ecclesia, that often meant meetings of 5,000 (with women and slaves excluded from participation). In the polis, the unit of collective decision-making was small, and the members homogeneous in general outlook and sufficiently disconnected from banausic activities (because sufficiently secure in their social and economic positions) to devote their passions and energies to common affairs. If the combination of directness and deliberativeness can only be achieved under these conditions, then the conventional criticisms of radical democracy are individually damaging and collectively overwhelming.

To vindicate the virtues of deliberativeness and directness, then, we must distinguish these values themselves from familiar ways of institutionalizing them--for example, citizen assemblies, or such modern analogs as workers' councils or economic parliaments--and then describe a modern set of arrangements of collective decision-making suited to these values and to modern conditions of scale and heterogeneity. To guide this elaboration of a workable direct and deliberative alternative to assembly democracy we need first to establish criteria for "workable" democratic solutions by characterizing the problems democracies now face and the limits of representative, aggregative arrangements in addressing them.


Conventional explanations of current institutional failure range, we said earlier, from too much state (and associated rent-seeking), to too much market (private control of investment under conditions of globalization), to many civic deficits (decline of trust-building associations). And we indicated, too, that emergent problem-solving institutions suggest the limits of those explanations. But what could an alternative be?

Our own proposal is that existing forms of constitutional democracy--and the associated boundaries between state, market, and civil society that inspire the limited categories of current debate--block democratic and effective strategies of problem-solving in the current environment: where existing political institutions favor uniform solutions throughout a territory, the problems require locally specific ones; moreover, the environment is volatile, so the terms of those local solutions are themselves unstable. In short, because of high diversity and volatility, important problem-solving possibilities are not being exploited by existing institutions. To the extent that this is so--to be sure, it is not the whole story--the problems of modern democracy arise quite apart from the clash of antagonistic interests or any guileful exploitation by individuals of blockages created by constitutional arrangements: they are (in the game-theoretic sense) problems of failed coordination, in which mutual gains are available, but different parties are unable to come to terms in a way that captures those gains. If the right arrangements of collective choice were in place, the parties could come to terms on one of the available alternatives. In contrast, recognition of the mismatch between solutions and available structures of decision-making leads, by itself, to paralysis, as it reasonably suggests that it is better to do nothing than something that will almost certainly fail.

Put another way, we assume that for some substantial range of current problems, citizens agree sufficiently much about the urgency of the problems and the broad desiderata on solutions that, had they the means to translate this general agreement into a more concrete, practical program, they would improve their common situation, and possibly discover further arenas of cooperation. This is not to make the foolish claim that everyone endorses the same ranking of solutions, only that they prefer a wide range of alternatives to the status quo. No surprise, then, that the new problem-solving institutions have begun to emerge just in those areas--public safety, public education, economic restructuring--where established institutions have most conspicuously broken down, and problems are agreed to be urgent. For breakdown opens space for new initiatives, and where, as we are assuming, actors are urgently motivated to look for a solution and prefer many alternatives to the status quo, that space is likely to be occupied.

But even in thus qualifying the extent of agreement, we may still be accused of an extravagant confidence in consensus. In its stronger form, this accusation rejects the idea of deliberative problem-solving altogether by criticizing the assumptions about consensus on which it depends. It asserts that the fundamental problem of politics is the pervasiveness of deep disagreement, the consequent fragility of political order, and the immanence of its disintegration into violence. So any assumptions about agreement--and not simply the set just noted--miss the point.[22] The criticism is right in recognizing disastrous possibilities, but wrong in the lessons draws from them. Assume the setting of a consolidated polyarchy: one in which there is no organized alternative to democracy, in which democracy is "the only game in town."[23] And assume--as is suggested by such consolidation--that citizens, who know that they disagree on moral, religious, and political issues, nevertheless accept the importance of conducting political argument on common ground. Those assumptions suffice to make deliberative politics possible.

In its more limited form, the objection is straightforwardly empirical: We assert and the critic denies that there is currently substantial agreement on a list of public problems and on desiderata as to their solution. We point to the diffusion of new problem-solving arrangements; the critic points to congressional gridlock; we think our diagnosis explains the gridlock; the critic thinks that the new arrangements are too marginal to require explanation. We don't propose to adjudicate this disagreement here, but only to reconfirm that our proposal, like all others, has its empirical commitments.

To return to the diagnosis: At the root of this mismatch between problems and problem-solving institutions is, we assume further, a fundamental and familiar characteristic of contemporary political problems: diversity. A commonplace of discussion of regulation and administration is that rules and services aimed at achieving any broad end--protection of the environment or training for economic activity--must be tailored to (constantly changing) local circumstances to be effective. Moreover, because the pursuit of such ends often requires the integration of many means--a regime of incentives and fines may have to be combined with monitoring and clean-up programs as well as research and development efforts to achieve acceptable levels of environmental protection--local combination of locally specific solutions are required as well.

But fundamental considerations of democracy apparently favor, if they do not mandate, uniform solutions. Thus a basic democratic idea is that citizens are to be treated as equals, which might be thought to imply that state regulations are to be cast in the form of general rules. Why constrain the free play of interest through aggregation or deliberation only to allow the powerful to favor themselves by writing laws that accord them benefits directly? It might be thought, too, to imply a requirement of precision or lack of ambiguity in those regulations. For why prevent directly self-serving regulations, but then permit indirect self-service through exploiting vagueness at the stage of interpretation and application of laws?

Other, related devices of constitutional democracy have the same effects. Thus, a basic institutional expression of the requirement of the rule of law--in particular, of the ban on self-serving interpretation--is the separation of powers, understood as the requirement that rule-making authority be vested in a body that includes representatives of diverse particular interests, but that does not itself apply the rules it makes to individual cases. The conventional rationale for this separation of rule-making and rule-applying is that it permits diverse interests to be incorporated into rules, even as it decreases incentives for rule-makers to design rules that favor themselves (either as representatives or as officials). But in obstructing corruption the separation of powers so understood reinforces the substantive uniformity requirement, and thereby tightens the constraint on tailoring solutions to special circumstances.

Hence a familiar and inconclusive tug of war: When problems need to be solved pressure mounts to violate the constitutional constraints of the rule of law and separation of powers--to overturn the Tudor polity--precisely because of the restrictions these impose on problem-solving. Then, as the dangers of violations mount, as politics threatens to degenerate into a patchwork of particularistic deals and local privileges, as constitutional democracy approximates pre-Tudor feudalism--pressure mounts to reimpose a system of strict rules. Thus in the United States, the standard criticism leveled against administrative agencies--created precisely to adopt law to particular circumstances--is that they pave the road back to serfdom. And standard proposals for reform--ranging from Lowi's juridical democracy and Sunstein's post-New Deal constitutionalism, to Hayek's neo-liberal constitutionalism--would redeploy rule-making authority to legislatures in order to ensure such substantive uniformity, thus reimposing the very constraints that had prompted earlier constitutional reform.[24]

This to-and-fro cannot be resolved simply by cutting the Gordian knot of constitutional constraint. Absent the most stringent civic sensibilities, a constitutionally unconstrained representative system--in which decisions by a representative body suffice to make the regulation legitimate, irrespective of concerns about substantive uniformity--produces the Hayekian nightmare: a pure bargaining democracy in which legislative decision-making is under no pressure to be deliberative, and hence under no pressure to explore improved solutions, or even to meet minimal conditions of coherence and efficiency. Outcomes will simply reflect the balance of political forces, with no obligation to consider how legislative choices will cumulatively solve the problem.[25] (According to public choice views, this is all that democracy is, or could be. But this supposes, improbably, that the real purpose of democracy is to achieve political equilibrium, not to solve problems or establish the legitimacy of solutions).

Nor can the mismatch of institutions and problems produced by current understandings of the rule of law and the separation of powers be finessed by a strategy of federalist decentralization that would permit local tailoring within a regime of strict rules. Federalism, generically conceived, is a system with multiple centers of decision-making, including central and local decision-makers, and separate spheres of responsibility for different units. In such a system problems requiring local solutions could be delegated to local centers of decision making, while problems admitting of general solutions could be addressed centrally. If logrolling was Hayek's nightmare of democracy, a radical version of federalism, in which the center did little more than register the generalizable results of local units, was his democratic arcadia.

But federalism, thus understood, creates troubles of its own precisely because it does not require the units of decision-making to communicate and pool their information. To underscore the force of this point, we extend our original characterization of the problem situation of modern democracies beyond the assertion that uniform solutions are not optimal, to the further proposition that particular locations, operating in isolation, lack the capacity to explore the full range of possible solutions. For this reason, optimal problem-solving requires a scheme with local problem-solvers who, through institutionalized discussion, learn from the successes and failures of problem-solving efforts in locales like their own. Through such exchanges each problem-solving unit would be better situated to capture the benefits of all relevant, locally tailored solutions, thus transcending the limits of localism without paying the price of uniformity such transcendence would otherwise require.

Federalism as currently understood does not foster such mutual learning from local experience; the scheme of a "directly-deliberative polyarchy" does. Indeed, abstractly conceived, it simply marries the virtues of deliberation and directness to an ideal of learning by explicitly pooling experience drawn from separate experiments. Whether this marriage can be made to work is our next subject.


The intuitive idea of directly-deliberative polyarchy is to foster democracy in its most attractive--direct and deliberative--form, and thereby increase our collective capacity to address unsolved social problems by overcoming current dilemmas of coordination. As background, to remind, we assume that the institutions of polyarchy are in place. More immediately, we assume that citizens--despite conflicts of interest and political outlook--agree very broadly on priorities and goals, but cannot translate this preliminary agreement into solutions fitted to the diversity and volatility of their circumstances because of constitutional uniformity constraints. So we look for institutions that are friendly to local experimentation, and able to pool the results of those experiments in ways that permits outsiders to monitor and learn from those efforts.

Consider first the implications for individual decision-making units. Diversity implies that reasoned decision-making in each will need to draw on local knowledge and values; volatility means it will need regularly to update such information. As each unit is distinct, none does best by simply copying solutions adopted by others, though they may do well to treat those solutions as baselines from which to move; as each faces changing conditions, practical reasoning requires a system of collective decision-making that fosters regular readjustment of solutions to those changes. Local problem solving through directly-deliberative participation is well-suited to bringing the relevant local knowledge and values to bear in making decisions. Direct participation helps because participants can be assumed to have relevant information about the local contours of the problem, and can relatively easily detect both deception by others and unintended consequences of past decisions. Deliberative participation helps because it encourages the expression of differences in outlook, and the provision of information more generally: The respect expressed through the mutual reason-giving that defines deliberation reinforces a commitment to such conversational norms as sincerity and to solving problems, rather than simply strategically angling for advantage (perhaps by providing misleading information); furthermore, if preferences over outcomes themselves are shaped and even formed by discussion, and mutual reason-giving reduces disagreements among such preferences, then being truthful will also be good strategy.

But the same concern for a form of decision-making that it is attentive to unexplored possibilities and unintended consequences requires institutionalization of links among local units--in particular, the institutionalization of links that require separate deliberative units to consider their own proposals against benchmarks provided by other units. Because practical reasoning requires a search for best solutions, decision-makers need to explore alternatives to current practice. A natural place to look for promising alternatives--including alternatives previously unimagined in the local setting--is in the experience of units facing analogous problems. Thus alongside directly-deliberative decision-making we need deliberative coordination: deliberation among units of decision-making directed both to learning jointly from their several experiences, and improving the institutional possibilities for such learning. These considerations lead us to our conception of directly-deliberative polyarchy--intuitively, a system with both substantial local problem-solving, and continuous discussion among local units about current best practice and better ways of ascertaining it.

Before filling out this intuitive idea by exploring its basic operating principles, we underscore that directly-deliberative polyarchy describes the form of problem-solving institutions: it is an order in which problem-solving proceeds through connected institutions and organizations that meet a set of abstract conditions of directness and deliberativeness. But the institutions and organizations that meet those conditions might vary widely, from networks of private firms, to public institutions working alongside associations. In this respect, the idea of directly-deliberative polyarchy operates at a different level of analysis from idea of associative democracy or workplace democracy. The idea of associative democracy is to solve problems through means other than states or markets: the nature of the "organizational instrument" matters. Similarly, workplace democracy specifies a particular institutional arena--the workplace. With directly-deliberative polyarchy, what matters is that the conditions are met, not the organizations that satisfy them.

Constitutional Principles

To describe the basic structure of directly-deliberative polyarchy, we need to answer three questions. First, what are the requirements of democratic process within and among units? More particularly, what does it mean for their decisions to be made deliberatively? Second, what conditions should trigger the operation of these deliberative mechanisms? And third how should the circle of membership in the deliberative bodies be drawn?

Deliberative Process within and among Units. At the heart of the deliberative conception of democracy is the view that collective decision-making is to proceed deliberatively--by citizens advancing proposals and defending them with considerations that others, who are themselves free and equal, can acknowledge as reasons. The shared commitment of citizens in a deliberative democracy is that the exercise of collective power should be confined to cases in which such justification is presented. Citizens contemplating the exercise of collective power owe one another reasons, and owe attention to one another's reasons.

But not all reasons are on a par. So the kind of attention owed must be calibrated to the kind of consideration offered. Thus, constitutional reasons are considerations that command substantial weight in decision-making. In deciding which considerations are to be assigned such weight, we look for a close connection to the standing of citizens as free and equal members of the political society: considerations affirming that standing have substantial weight, whereas those that deny it are weightless. Thus, citizens must have fundamental political and civil rights because those rights are backed by reasons that affirm the standing of citizens as free and equal, whereas the denial of those rights requires appeal to considerations that throw such standing into question--perhaps by denying that members meet all the qualifications for citizenship. But denials of qualification--assertions that some member is not to be regarded as a free and equal citizen--do not count as reasons at all because they are not considerations that command respect from those whose standing is denied. So effective participation rights cannot, except perhaps in very special circumstances (perhaps cases of extreme emergency), permissibly be denied. Similarly, proposals backed by reasons rooted in interests fundamental to the standing of members as free and equal can be rejected only upon offering alternative, more plausible projects for advancing those interests.[26] Thus, a requirement of ensuring a basic educational threshold--a threshold defined relative to participation as citizen, and more generally, as cooperating member of society--would be a constitutional reason, and a proposal that would ensure such a threshold would be rejected in a well-ordered deliberative body only in favor of an alternative, better designed scheme.[27]

The first and most fundamental requirement of a directly-deliberative polyarchy is, therefore, that it affirm its character as democratically deliberative by giving stringent protection to claims backed by constitutional reasons.

Of course, not all acceptable reasons for public choices are of constitutional magnitude. The class of policy reasons comprises those considerations whose endorsement is neither required by nor incompatible with a conception of citizens as free and equal, and which are relevant to an issue under consideration. A proposal framed by such considerations may reasonably be rejected by a counter-argument that articulates an alternative balancing of the reasons generally understood as relevant to allocating the resource in question. Consider again the case of education. In deciding how to allocate resources, some relevant and potentially competing policy reasons are: helping each student fully to achieve potential; ensuring that students who are performing least well are given special attention; ensuring common educational experience for students of diverse backgrounds. In the case of health care, the reasons include: helping those who are worst off; helping those who would benefit most from medical resources; assisting larger numbers of people; ensuring that people have fair chances at receiving help, regardless of the urgency of their situation and of expected benefits from treatment.

As these examples suggest, the policy reasons relevant to particular domains are complex and varied, and there often will be no clear, principled basis for ranking them: Different, equally reasonable participants in deliberative process (and, a fortiori, different deliberative bodies) will weigh them differently. Reasonable people and reasonable collective decision-makers reasonably disagree, and recognize the results of a deliberative process in which such reasons are aired as legitimate.

This distinction between constitutional and policy reasons brings us to the second broad condition, a requirement of substantive due process on the operation of directly-deliberative polyarchy: The process is to give is due consideration to reasons of both types, suitably weighted (and allowing for reasonable differences of weight).

Moreover, we require, third, that this consideration be explicit. It is not sufficient to require that outcomes be rationalizable--that the deliberative process issue in decisions for which appropriate reasons could be cited--and to leave it to another institution, say, a court, to determine whether that condition is met. Outcomes in directly-deliberative polyarchy are to be arrived at through discussion in which reasons of the appropriate kind are given by participants. Four considerations lead to this conclusion:

1. Though deliberative justifiability itself is important, it must be aimed at to be achieved; that is, it will not in general be true that results achieved through a process of exchange or bargaining, or outcomes that reflect a balance of power, will be defensible by reasons of an appropriate kind. So requiring actual deliberation helps to establish a presumption that results can be defended through reasons, and thus a presumption that the outcomes of collective decision-making are legitimate.

2. Offering reasons to others expresses respect for them as equal members of a deliberative body. So actual deliberation plausibly helps to foster mutual respect, which in turn encourages citizens to confine the exercise of power as the deliberative idea requires. No similar result can be expected if we assign the job of assessing the justifiability of outcomes to a separate institution.

3. Actual deliberation provides a better rationale for relying on majority rule, should there be disagreement. With reasons openly stated, everyone can observe that the supporting considerations were relevant reasons, despite disagreements about their proper weight. It is manifest to participants, then, that people are not being asked simply to accede to the larger number, but to accept what they can see to be a reasonable alternative, supported by others who are prepared to be reasonable.

4. In actual reason-giving, citizens are required to defend proposals by reference to considerations that others acknowledge as reasons, and not simply by reference to their own interests. To the extent that such public reasoning shapes preferences, conflicts over policy will be reduced, as will inclinations to strategically misrepresent circumstances. Moreover, actual deliberation is, by its nature, a form of information pooling: when people take seriously the task of providing one another with reasons, information about circumstances and outlooks that is relevant to improved policy is brought to bear by those in possession of it. No similar effects on preferences or on information are likely to issue from non-deliberative processes subject to subsequent review. Indeed, understanding the process of review as the natural forum of principle may well encourage strategic, as distinct from deliberative, conduct.

Requiring explicit reason-giving rather than rationalizable outcomes may, however, have a downside. Critics of deliberative decision-making fault it for being doubly exclusionary.[28] Deliberation, they say, is a particular discursive style, with all the conventional indicia of the rational: formal, deductive, and unemotional. By insisting on abstraction from the personal and particular, deliberation excludes both people and information. People, because it silences citizens whose discursive style is detailed, narrative, and passionate; information, because it only invites contributions cast in general terms. As a result, deliberation is unfair and ineffective. Urging more of it is a reform strategy, but not an especially inviting one.

This objection makes two assumptions, both unwarranted. First, that requiring an explicit statement of reasons implies that nothing other than reasons can be stated--as though a conception of deliberative justification supported a ban on undeliberative humor. Second, that the canonical form of deliberation is the justification of a regulation from first principles: the argument for progressivity in the tax system on grounds of a conception of political fairness. Deliberation may take this form, but nothing in the concept of reason-giving requires that it do so. Nor, more immediately, is the reason-giving that occupies us here naturally expressed in the form of deductions from general political axioms. To the contrary, deliberative problem-solving is by its nature focused on addressing specific problems in local settings. Giving reasons under these conditions is, generally speaking, a matter of offering considerations recognized by others as pertinent to solving the problem at hand. It is simply impossible to limit in advance the kinds of considerations that might be relevant, or the form in which those considerations are to be stated. Indeed, deliberation will characteristically involve debating the implications of general principles (standard operating procedures, rules of thumb) in light of the particulars of local experience, and inviting discussion of such experience in whatever terms suits participants--including the ironic "yeah, yeah" that condemns the latest implausible suggestion.

Still, it might be said that requirements of deliberation unfairly bias decision-making in favor of the verbal, that we may end up with a pluralistic logocracy, in which the many forms of verbosity are all on display, but the shy, quiet, and reserved are left out. We agree that there is a difficulty here, but why isn't it remediable? In settings of deliberative problem-solving, everyone has something to contribute so the first task in improving the operation of deliberative arrangements is to ensure that all participants understand that and are encouraged to contribute. The potential for deliberative failure is no argument against efforts at such improvement.

These reasons for preferring decisions by actual deliberation--particularly the last consideration--suggests a fourth requirement of democratic process: that there be like deliberation among units as well. The advantage of actual, deliberate consideration of alternatives by citizens of equal standing but diverse experience and disposition is that the diversity of viewpoints brings out the strengths and weaknesses of diverse proposals. Moreover, the diversity of proposals reveals strengths and weaknesses in viewpoints that make for more careful assessment in later rounds. Extending deliberation across units allows each group to see its viewpoints and its proposals in light of alternatives articulated by the others: in effect, it ensures that the exercise of practical reason is both disciplined and imaginative.

To be effective in provoking this kind of informative comparison, information provided for this purpose must be supplied in a way that both anticipates and reflects this use: In accounting for their own decisions, decision-making processes, and outcomes, units must take into account information about the relevant practice elsewhere, or make a case that apparently better practice is either not genuinely better or irrelevant to their circumstances because of differences in population or resources. A standard way of doing this is through benchmarking: evaluation of one's own activities by comparison with others, judged to be similar, by means of metrics inherent in the choice of the comparison. Benchmarking thus requires a survey of possible comparisons, evaluation of possible metrics, and revision, when necessary, of initial choices of both; and the effectiveness of such surveys, evaluations and revisions depends on the willingness of all participants to disclose information in view of the investigations of the others. This amounts to requiring that, as when acting alone, units actually deliberate among themselves in the sense of taking account of respective reasons, and not content themselves with deliberative justifiability. This requirement implies that units that show poorly in public comparisons will be under substantial pressure to improve their practice to meet the standard of performance set in other comparable units.

As we will see below, responsibility for ensuring that deliberation among and within units meets these four conditions falls ultimately to authorizing and monitoring agencies--legislatures and courts. But the responsibility is to be discharged by ensuring that the relevant decision-making bodies act deliberatively, not--so far as possible--by substituting for their decisions.[29]

Triggering. With these core conditions in place, we come to the areas of policy for which directly-deliberative polyarchy is suited. Generally speaking, the institutions of directly-deliberative polyarchy are designed to do well where current institutions do badly. Hence they should be deployed where the problem-solving capacities of current institutions is subject to serious doubt. These doubts are greatest when the following conditions hold:

(a) The sites at which a problem arises and requires address are too numerous and dispersed for easy or low cost centralized monitoring of compliance with regulations. Even if uniform regulations were appropriate, these conditions would suggest a need for decentralizing the capacity to monitor compliance. Discussions of workplace health and safety regulation commonly emphasize this problem: too many workplaces for a central inspectorate to review.

(b) The diversity of sites at which similar problems arise suggests that problem solvers at different sites will want to employ different means to achieve similar aims and specify their aims differently.

(c) The volatility of sites suggests that a need for continuous reflection on means and ends, and the importance of adjusting both in light of new information about the environment.

(d) The complexity of problems and solutions--where problems are substantially the product of multiple causes and connected with other problems, crossing conventional policy domains and processes--implies that the appropriate strategy requires coordination across those domains. Urban poverty, local economic development, and effective social service delivery are among the familiar problems that occupy this class. Solving them plausibly requires cooperation across quite different institutions and groups--for example, lending institutions, health care providers, technology diffusers, education and training establishments, housing authorities, community development corporations, neighborhood associations.

When all these conditions are in force, we have a strong case for directly-deliberative polyarchy, with its linked, local problem-solvers: Because of the numerosity and diversity of sites, we want a structure of decision-making that does not require uniform solutions; because of volatility, we want a structure with built-in sensitivities to changing local conditions; because of the complexity of problems, we want a structure that fosters interlocal comparisons of solutions.

To be sure, departures from these four conditions imply a less strong case for directly-deliberative polyarchy, and a correspondingly stronger case for markets or regulatory solutions. But even in the face of departures from these ideal conditions, two considerations support the case for directly-deliberative polyarchy. First, as we have urged, it fits with democratic values, and that fit will tip the balance in unclear cases. Second, our basic premise is that is that existing strategies of problem-solving are not working well. So we may be aided in diagnosing the shortcomings of those strategies if we try this alternative. Among other things, it will test the thesis that the troubles emerge from a mismatch of problems and institutions of collective choice.

In describing the virtues of directly-deliberative polyarchy, we have been emphasizing its advantages over conventional political solutions. To highlight the force of these observations about triggering, consider the circumstances under which problem-solving through directly-deliberative polyarchy is preferable to solution through market exchange--here understood as a form of social coordination in which agents need not arrive at a common decision nor defend their separate decisions by giving reasons to others.[30]

Thus suppose we are concerned about the production and allocation of a good that is widely regarded as urgent--that citizens can claim as a matter of basic right or need--and about whose proper production and/or allocation there is disagreement.[31] Because the claims for the good are urgent, arrangements of provision should be open and accountable; moreover, urgency and disagreement together establish a presumption that decisions about the good's provision should be backed by an acceptable rationale. That presumption can be defeated in the case of goods (for example, bread or cars) for which there are a large number of providers and about which it is relatively easy (either for consumers or a centralized monitor) to acquire accurate information. Assume, then, that the good is best supplied by a restricted range of providers, and that there are high costs to switching among those providers: there can, then, be no presumption of voluntarism in the choice of provider. Add, now, that information about the good is difficult to acquire or summarize because a large number of dimensions are important to its evaluation, people disagree about the relevant dimensions and their relative weights, and the conditions of its production and allocation are volatile. Under these conditions, we want goods to be provided through mechanisms in which decisions are backed by reasons and based on pooled information. In short, we have reason to favor directly-deliberative polyarchy over market.

Membership. Finally, as to membership. The basic standard is that directly-deliberative arenas are to be open to providers and parties affected by the extent and manner of provision. (In the case of schools, for example, parents, teachers, and residents of community served by school.) While very little can be said in general terms about the requisite representational form: how many members of different groups, affected parties, etc.--once more, there is every reason to expect at least as much variation as we currently see in polyarchies--a few considerations are to frame debate about whether deliberative bodies all who are entitled, or are instead objectionably exclusive. In general terms, the considerations pull in two directions, reflecting the ideas of political equality and deliberativeness that define the directly-deliberative conception. The value of equality suggests a one-person/one-vote composition of deliberative bodies, whereas the requirement of deliberativeness suggests a constitution that assigns membership in ways that foster the provision of relevant local information and the crisp articulation of alternative views.

More particularly, then, three considerations need to be balanced in decisions about membership. First, citizens can object that the composition and scope of directly-deliberative bodies is objectionably discriminatory--for example, that their geographic range has been gerrymandered on racial or ethnic lines. Second, there is a presumption in favor of equal membership for affected parties--open meetings, with equal rights to participate in discussion and decision-making for all affected parties. Third, rights to participate might also be awarded to organizations with special knowledge that is essential to the problem area in question (for example, neighborhood organizations in the area of public safety), or who are able to articulate a point of view in ways that foster deliberation among alternative solutions.


Why expect that such problem-solving will have concrete benefits? How might it be able to overcome the problems of limited information and diversity of sites that vex state action? Five considerations are important.

First, the parties to the discussion are presumed to have relevant local knowledge; moreover, they can put that information to good use because they understand the terrain better and have a more immediate stake in the solution.

Second, assuming a shared concern to address a problem, and an expectation that the results of deliberation will regulate subsequent action, the participants would tend to be more other-regarding in their political practice than they would otherwise be inclined to be. The structure of discussion--the requirement of finding a solution that others can agree to, rather than pressuring the state for a solution--would push the debate in directions that respect and advance more general interests. Other-regardingness would encourage a more complete revelation of private information. And this information would permit sharper definition of problems and solutions.

Third, pursuing discussion in the context of enduring differences among participants would incline parties to be more reflective in their definition of problems and proposed strategies for solution; it would tend to free discussion from the preconceptions that commonly limit the consideration of options within more narrowly defined groups, thus enabling a more complete definition and imaginative exploration of problems and solutions. The same is true for the federalism of problem-solvers that emerges from requirements of discussion across units-- here, too, comparisons of solutions at different sites, and bench marking of local solutions by reference to practice elsewhere, suggests a basis for improving local practice.

Here, notice that directly-deliberative polyarchy--understood as a form of problem-solving--is not thwarted by, but instead benefits from, heterogeneity of participants. Of course, the participants must--as our discussion of deliberation indicates--share a view about relevant reasons. But this is, we think, a rather weak constraint that does not demand substantial homogeneity--certainly not homogeneity of comprehensive moral outlook.

Furthermore, monitoring in the implementation of agreements would be a natural byproduct of ongoing discussion, generating a further pool of shared information.

And, finally, if things work, the result would be a mutual confidence that fosters future cooperation.

In all these ways, then, deliberation about common problems with diverse participants might thus reasonably be thought to enhance social learning and problem-solving capacity.


We conclude with a brief discussion of the implications of directly-deliberative polyarchy for the design of and expectations on basic political institutions.

First, directly-deliberative polyarchy is, as we have indicated, a form of polyarchy. So we assume the continued presence of the legislatures, courts, executives, and administrative agencies, controlled by officials chosen through free and fair elections, in which virtually all adults have rights to suffrage, office-holding, association, expression, and face alternative, legally protected sources of information.[32] Though the operation of these institutions and arrangements changes, they remain and continue to serve some of the political values with which they are conventionally associated: peaceful transitions of power, restraints on unbridled power, fair chances for effective influence over authoritative collective decisions, opportunities to develop informed preferences, etc.

But with the shift in the locus of problem-solving, the operations and expectations of basic institutions changes markedly.

Consider first the role of legislatures. Directly-deliberative polyarchy is animated by a recognition of the limits on the capacity of legislatures to solve problems--either on their own or by delegating tasks to administrative agencies--despite the importance of solutions. Rejecting the Neo-Liberal Constitutionalist idea that the problems are essential recalcitrant to collective address, and the modern Civic Republican idea that their address requires only a more vigilant exclusion of private interests from national policy making (and a correspondingly more acute intervention by technically adept guardians of the common good), the legislature in a directly-deliberative polyarchy takes on a new role: to empower and facilitate problem-solving through directly-deliberative arenas operating in closer proximity than the legislature to the problem. More particularly, the idea is for legislatures, guided by the conditions of triggering, to declare areas of policy (education, community safety, environmental health) as open to directly-deliberative polyarchic action; state general goals for policy in the area; assist potential deliberative arenas in organizing to achieve those goals; make resources available to deliberative problem-solving bodies that meet basic requirements on membership and benchmarking; and to review at regular intervals the assignments of resources and responsibility. To be sure, legislatures can only play this role if they are able to identify problems needing solution and agents with the capacity to solve those problems, even when they cannot themselves produce the solutions. But once we acknowledge the importance of diversity and volatility in shaping acceptable solutions, this assumption is entirely natural.

Administrative agencies, in turn, provide the infrastructure for information exchange between and among units--the exchange required for benchmarking and continuous improvement. Instead of seeking to solve problems, the agencies see their task as reducing the costs of information faced by different problem-solvers: helping them to determine which deliberative bodies are similarly situated, what projects those bodies are pursuing, and what modifications of those projects might be needed under local conditions.

And the responsibility of constitutional courts is neither simply to inspect procedure for its adequacy as representative, nor to reorganize institutions by reference to substantive constitutional rights, but to require that decision-making proceed in directly-deliberative way: that is, to require that problem-solvers themselves make policy with express reference to both constitutional and relevant policy reasons. You might describe this as a genuine fusion of constitutional and democratic ideals: a fusion, inasmuch as the conception of democratic process includes a requirement that constitutional reasons be taken into account, as such. The aim is a form of political deliberation in which citizens themselves are to give suitable weight to constitutional considerations, and not leave that responsibility to a Court.

These remarks sketch, in sparest terms, how basic political institutions might shift in expectation and responsibility under conditions of directly-deliberative polyarchy. Further details will vary greatly, certainly as much as they do in existing polyarchies. Rather than outlining the dimensions of such variation, we propose to clarify and deepen this account of transformed conventional institutions by addressing an objection to the very coherence of directly-deliberative polyarchy as a form of problem-solving that conforms to basic democratic values. Generally speaking, the objection is that directly-deliberative polyarchy is an unstable combination of institutionalization of democratic values: either central institutions will not supervise local arrangements enough to avoid local tyrannies, or they will over-supervise, thus regenerating the problems of centralized control that directly-deliberative polyarchy is supposed to avoid. More particularly, the objection is that directly-deliberative polyarchy needs to meet two requirements that are at war with one another: Deliberative problem-solvers are supposed to satisfy various conditions (on membership, deliberativeness, and external links to other problem solvers). But directly-deliberative problem-solvers will not meet these conditions as a matter of course, nor is their satisfaction a self-enforcing equilibrium. So the responsibility for ensuring that they are met falls to authorizing and monitoring agencies. If, however, problem-solvers are to achieve the variation in local solutions demanded by conditions of diversity and volatility, then authorizing and monitoring agencies must also ensure them autonomy.

These two conditions are arguably in tension. For directly-deliberative problems solvers can act in ways that conflict with the constitutive values and conditions of democracy itself, either by deciding on the basis of considerations that conflict with those values, or by failing to take them fully into account. If a fundamental, generic responsibility of authorizing and monitoring bodies is to ensure that decision-making is democratic, then those authorizing bodies are obligated to review and pass judgment on the decisions of the authorized bodies. But this creates two related troubles for directly-deliberative problem-solving: first, reduced autonomy in the name of ensuring democracy may substantially limit interest and enthusiasm for participating in problem-solving bodies. Second, if ensuring democracy means constantly second-guessing the solutions chosen by directly-deliberative problem-solvers, then those problem-solvers may decide to avoid troubles by imposing uniform solutions (choosing solutions that have already passed muster), disregarding the suitability of those solutions to their circumstances.

We have four replies to this problem. The first is to introduce a note of realism. The objection is entirely familiar from current discussions of federalism and of relations between courts and legislatures. Focusing on the latter, it is commonly agreed that courts should, whatever else they do, uphold democratic process, ensuring that all citizens have rights to participate as equals in that process.[33] Sometimes majorities violate that requirement, and when they do courts have a responsibility to overturn the results of those violations. The tension noted above is, generally speaking, simply an instance of this problem, which is commonly called the "countermajoritarian dilemma": it is not a problem created by the proposal advanced here, but a reflection, within our proposal, of a problem that any adequate conception of constitutional democracy needs to face.

Second, accepting that the general structure of the problem is familiar, it might nonetheless be argued that a deliberative conception of democracy--or an idea of directly-deliberative polyarchy inspired by that conception--worsens the problem by imposing more stringent standards of democracy. Though a wide range of views will permit review and rejection of decisions of grounds of incompatibility with democracy, the deliberative view embraces an expansive conception of democracy and a correspondingly expansive and therefore invasive account of when the judgments of problem solvers are properly second-guessed. The force of this objection depends on a belief that is widely shared but simply misguided: that deliberation, properly conducted, issues in consensus. We have already explained our reasons for rejecting this claim. Deliberation is a matter of balancing relevant considerations, and arguing in light of such balance: competent deliberators will work out the balance differently; and, correspondingly, competent deliberative bodies will typically arrive at different conclusions, or arrive at the same conclusions differently. Indeed, there is no compelling a priori argument that the range of acceptable results of deliberative processes is smaller than the range of acceptable results of aggregative processes. So we reject the claim that the deliberative view worsens the familiar problem.

Indeed, third, we think that the deliberative conception may reduce the tension between democracy and autonomy. To explain how, we introduce the idea of a "division of deliberative labor." On a conventional view of collective decision-making within a constitutional regime, the division of labor assigns to legislatures the responsibility for devising laws that advance the common good, and to courts the responsibility for ensuring that those laws respect the constitution and the political values implicit in it. As our earlier discussion indicates, the deliberative view rejects this way of dividing deliberative labor. When objections are raised on constitutional grounds to decisions reached by problem-solvers--when it is argued unacceptable reasons animated the decision, or that fundamental constitutional values were neglected by it, the role of courts (and legislatures) is not to substitute their own judgment about the proper outcome, but to require that the deliberative body revisit the issue, taking the full range of relevant considerations explicitly into account--and exploring the experience of similarly situated problem-solving bodies. Suppose, for example, that a decision to impose an English-only requirement on schools is challenged on grounds that students who are not native English speakers will be disadvantaged by it, and consequently disadvantaged as citizens. The response should be to require that the school committee responsible for imposing the requirement revisit the decision, attending both to the importance of education for equal citizenship and to the experience of other multi-lingual districts in solving the problem. In short, the deliberative view rejects the conventional division of deliberative labor, proposing instead that all bodies making collective decisions share responsibility for upholding the democratic constitution by treating its principles and values as regulative in their own decisions.

Our final reply builds on this third point. Suppose that deliberative decision-makers are required to arrive at decisions with explicit attention to constitutional values and comparable experience. Still, they may make decisions that conflict with the democratic constitution, and courts may be required to review their decisions in this light. But when they are, they will have a record of fact and reasoning to draw on in making their decisions. Because they have imposed requirements of due consideration on problem solvers themselves, courts will have the information they need to decide whether means are suitably tailored to ends, and whether ends are specified in ways that satisfy constitutional constraints. Judgments about whether or not to defer to problem-solvers will be backed by fact, and not simply by a priori estimates of institutional competence.

Conclusion: The Idea of the Public in Directly Deliberate Polyarchy

The public is to democracy what work is to Marxism and sex to feminism: the place where the features of our selves that make us most human, and the obstacles to the realization of that human potential, become manifest in the ways we form and deform each other. The public in directly deliberative democracy is the place where practicality in the form of problem solving meets constitutional principle in the form of deliberation through reason giving among citizens who recognize themselves to be free and equal. This public is convoked not only in the national legislature, but also in the work of the local school governance committee, the community policing beat organization, and their analogs in areas such as the provision of services to firms or to distressed families. It connects and so effaces the distinction between high politics of state affairs, and the political maneuverings of everyday affairs. This connection crosses and erases just the lines that distinguish the public from the private not only in familiar theories of representative democracy, but also, and perhaps surprisingly, in current understandings of radical democracy. Hence by way of conclusion, to underscore the defining features of directly deliberative polyarchy and to point towards the unfinished work of our project, we look briefly at the fate of the modern radical-democratic understanding of the public as it appears in the works of such representative figures as Habermas and Arendt and their innovative followers, and contrast that fate with the idea developed above.

From this vantage point the most striking feature of contemporary views of radical democracy is the measure to which they have become rear-guard, defensive strategies, directed as much, if not more, to limiting--at times by novel means--the erosion of the institutions of nineteenth century parliamentary democracy as transforming and extending them. In part these limited ambitions are a prudent response to the temper of the times, hostile, as we saw at the outset, to the very idea of public action, and particularly hostile, since the fall of the plan economies, to any hint of collective control over life choices of individuals. But on a deeper level this self limitation reflects a sharp and nearly unbridgeable distinction, long established in the social and political theory from which much radical democratic theory stems, between a higher, political world of self determination through great public deeds or through the pursuit of truth itself, and a lower realm of survival governed by the rules of technical necessity. It is this distinction, and the postulates upon which it rests, that the idea of the public in directly deliberative democracy questions most immediately.

Take as a first illustration of this divide the notion of the public as it appears in the work of Habermas. In his view human interactions differ fundamentally according to whether their purpose is to achieve worldly success or understanding. In the first case information is manipulated strategically to advance individual or group interests, as in economic exchange. In the second, discussion is aimed at truth seeking, and those who engage in it know themselves to be bound by the requirements of sincerity (Wahrhaftigkeit), veracity (Wahrheit), and moral probity (Richtigkeit) that not only govern dispassionate discourse, but also create the ethical substratum upon which human sociability ultimately rests. Scientific investigation is the most explicit model of this discourse ethics.

Given this distinction, democracy becomes a device by which the discursively formed understanding of common purpose can be used to steer (or, in more anguished formulations, besiege), the administrative apparatus of the modern state, understood itself as a series of technically constrained instrumentalities for guiding the (still more constrained) activities necessary for society to reproduce and advance. Democratic steering is itself divided into stages or phases, ordered by their distance from the apparatus of actual decision-making, and hence their freedom from technical constraint: Parliamentary debate is limited by its connection to administration, the disputations of within political parties by their connection to parliamentary debate. The modern public in this conception is the place where democracy is most authentic because least constrained. It is the dispersed forum where citizens, as individuals and in groups, connected by the means of mass communication, form currents of opinion in seeking the truths that bear on the resolution of the great questions of the day. Because the purpose of debate is near enough to the clarification of questions of right and wrong, and far enough from the resolution of particular disputes, to count as a form of truth seeking, the members of the dispersed assembly that is the public respect the rules of discourse ethics and the assumptions of autonomy and equality on which they rest. Contributions to public debate, therefore, can be critical and criticizable in a way that brings general principles of justice to bear on matters of political moment.

But whether, and to what extent, the public's critical contributions can steer the state in the direction of just must remain, given Habermas' fundamental partition of human action, an open question. The freer the communication within the public -- the greater the immunities from state interference with the formation of opinion, the more accessible the newspapers, the less venal the television -- the greater clarification it can attain. Indeed, the call for the democratization of public debate that follows naturally from Habermas' emphasis on the potential political significance of discourse ethics is exactly the aspect of his general theory that classes it as a type of radical democracy. But the dualism of success-oriented action and understanding at the heart of that theory suggests as well that even the most radical extension of the public sphere would be of limited consequence precisely because the logic of technical necessity, to which administration, parliament, and party must in turn respond, set limits -- but which ones? -- to the direction that justice informed by discourse ethics can steer. At its most paradoxically self defeating, Habermas' view seems to be that the democratic public can not be just and effective because to be just it must be informal in the sense of constituted free of institutions, while to be effective it must be institutionalized in forms that hinder the pursuit of justice (Faktizit鋞 und Geltung, S. 372 ff.) In the end, radical democracy becomes more a call to rally opposition to the encroachments of the technical "system" than a program for the redirection the ensemble of institutions through public deliberation.

As a second illustration of the way the divide between the truthful and the creative on the one hand and the technically necessary on the other have withered the concept of radical democracy and its public, consider the position of Arendt. For Arendt, human activity counts as labor if directed to the rhythmic necessities of biological reproduction, as work if directed to the construction of those durable artifacts, from houses to highways, that provide the scaffolding and outward signs of our social life, and as action if directed to those (re-) foundings that give direction and meaning to each by giving collective, political purpose to all. The public in this view is just the citizens in action; and this citizenry in action embodies democracy in its most radical, constitutional, aspect. The dilemma for this view, of course, is that democracy continues after its constitution -- a constitution that, upon enactment, required immediate re-constitution would not be one; and on Arendt's understanding concerns arising within the constituted polity would fall from the sphere of action and the political to the spheres of work and labor. For practical purposes, the public would be purposeless. Thus if the Habermas view raises doubts as to whether the public in radical democracy can ever influence the constitution of the state, Arendt's view raises doubts as to whether it can play any part in governance after having done so.

Recent efforts to modify the idea of the public in both views to respond to these kinds of criticisms by softening the distinctions on which they rest only underscore the constraints of the original schemes. In both cases the modifications focus on the role of social movements -- of women, of racial or ethnic minorities, of citizens concerned about the environment, and many other -- as forms of the public so dispersed within society to be acting outside of institutions -- and hence untainted by their technical or workaday constraints -- yet directly enough engaged with changing particular social arrangements to influence them. In those writings on social movements that refer, critically, to Habermas and other like him, the emphasis is on spontaneous citizen action precisely as a protest against the risks of (increasing) technical manipulation of the social and natural worlds: The risk of nuclear catastrophe calls forth a social movement against the construction of nuclear power stations, the industrial use of toxics calls forth a movement for their regulation. But this suggests only that democracy in the attenuated and almost inarticulate sense of visceral protest in response to immediate threats can play a part in limiting the reach of a "system" whose innermost mechanisms remain beyond political influence.[34]

Writings on social movements that take Arendt as their critical referent are more ethereal still. From this perspective social movements are seen as the potential precursors to constitutional conventions in the small; or, rather, the movement itself is a kind of fleeting and fugitive assembly whose existence testifies to the possibilities of a democratic refounding when nothing else does. In both variants, the, the newer views make the defensiveness of their antecedents into a matter of principle and reduce radical democracy to an argument about the possibility of protest against the further subjugation of freedom to necessity .

Directly deliberative polyarchy, in contrast, does not seek refuge in social movements; it holds out the promise of transforming the institutions of social steering, not merely containing their erosion. It claims that the apparent limits on the applicability of democratic principles, and the dualism of creative or truth-seeking public and banausic world, can be overcome by understanding how those principles can contribute to problem solving, and how problem solving can contribute to the reinterpretation of those principles. At the core of this mutual re-elaboration is the idea of deliberation as reason giving in context -- that is, relative to purposes that concern the citizens. Purpose here does not, as in the Habermasian view, vitiate deliberation. Rather it guides and enables deliberation by suggesting the comparisons and contrasts that give meaning to diverse and mutually informative points of view. And purpose here means purpose of all kinds. As against Arendt, the idea of deliberation in directly deliberative polyarchy does not distinguish between constitutional and operational tasks. Indeed the notion of continuous adjustment of means and ends at the notion of deliberation in directly deliberative polyarchy -- the very feature that recommends it as a method of problem-solving in diverse and volatile environments -- undercuts the distinction between these types of activities.

But in stating the contrast between the current, defensive ideas of radical democracy and the possibilities of directly deliberative polyarchy we are advancing our case by signing promissory notes. We have offered some empirical hints of new institutional developments to warrant our conjectures, and subjected those conjectures to the preliminary tests of internal consistency, and noted their appeal as alternatives to the despairing prospects of current debate. If we are right in thinking a new, radically participatory form of democracy is beginning to stare us in the face, the obvious and urgent thing to do is stare back.

(click on footnote number to return to appropriate paragraph)

1 Joshua Cohen is professor of philosophy and Arthur and Ruth Sloan Professor of Political Science at MIT; Charles Sabel is professor of law and social science at Columbia University Law School.

2 For background on diagnosis and remedies, see Joshua Cohen and Joel Rogers, Associations and Democracy, especially the concluding chapter, Joshua Cohen, "Procedure and Substance in Deliberative Democracy," in Seyla Benhabib, ed. Democracy and Difference: Contesting the Boundaries of the Political (Princeton: Princeton University Press, 1996), pp. 95-119; and Charles Sabel, "Learning By Monitoring: The Institutions of Economic Development," in Handbook of Economic Sociology, eds. Neil Smelser and Richard Swedberg (Princeton: Russell Sage and Princeton University Press, 1995). For a companion essay to this one that focuses on the institutional innovations central to the new problem-solving entities and their implications for the organization of government and constitutionalism, see Michael Dorf and Charles F. Sabel, "A Constitution of Democratic Experimentalism," working paper, Columbia Law School, March, 1997.

3 See Friedrich Hayek, The Constitution of Liberty (Chicago: University of Chicago Press, 1960); James Buchanan, The Limits of Liberty (Chicago: University of Chicago Press, 1975); Antonin Scalia, "The Rule of Law Is a Law of Rules," University of Chicago Law Review, William Riker, Liberalism Against Populism; William Riker and Barry Weingast, "Constitutional Regulation of Legislative Choice: The Political Consequences of Judicial Deference to Legislatures," Virginia Law Review 74, no. 2 (March 1988): 373- 401.

4 See Robert Kuttner, Everything for Sale: The Virtues and Limits of Markets (New York: Knopf, 1997).

5 See Robert Putnam, Democracy and the Civic Community: Tradition and Change in an Italian Experiment, (Princeton: Princeton University Press, 1992); Robert Putnam, "Bowling Alone: Democracy in America at the End of the Twentieth Century" Journal of Democracy (1995): [pages]; Amitai Etzioni [something]; Sandel's remarks on a new public philosophy mix the sociological anxiety characteristic of much communitarianism with concerns about the political economy of citizenship that are closer to our own focus on new arrangements of democratic governance. See Michael Sandel, Democracy's Discontent: America in Search of a Public Philosophy (Cambridge, MA: Harvard University Press, 1996), pp. 324-38.

6 The Chicago experiment in community policing -- the Chicago Alternative Policing Strategy (CAPS) -- involves assignments of police officers to single beats (30 square block areas) for an entire year; neighborhood-based organizations called "problem solving groups" that work in partnership with police in each beat; and open meetings with community and police each month. The emphasis on community participation distinguishes the Chicago scheme from other strategies that share the label "community policing." See Archon Fung, "Street Level Democracy," May 1996 (unpublished).

7 For discussion, see Charles Sabel, "Local Development in Ireland: Partnership, Innovation, and Social Justice" (OECD, 1996); "Milwaukee Jobs Initiative Consortia Employment Project Description," September 1996 (unpublished draft).

8 [References to Wolin, Habermas, Arendt, Chomsky]

9 The ideal of democracy also has considerable force for organizations whose collective decisions are not binding. But the rationale for democratic decision-making is most compelling in the case of binding collective choices: that is, when members of the collectivity are expected to regulate their own conduct in accordance with its decisions.

10 See Democracy and Its Critics (New Haven: Yale University Press, 1989), pp. 221-22.

11 [See Locke's idea that people are naturally equal in that all have the natural right to freedom.] For related discussion, see John Rawls, Political Liberalism (New York: Columbia University Press, 1993), I: 3, I: 5, VIII: 6, VIII: 7; Charles Beitz, Political Equality (Princeton: Princeton University Press, 1989), ch. 5.

12 In addressing the question for this case, we do not mean to suggest that democracy is important only when these background assumptions are in place, but that the answer for different cases will vary, and that important considerations are likely to get lost if we confine attention to answering the more general question.

13 A common rationale for democracy is that it treats people as equals by giving equal consideration to their interests. See Robert Dahl, Democracy and Its Critics (New Haven: Yale University Press, 1989); Thomas Christiano, The Rule of the Many (Boulder: Westview, 1996). We avoid this rationale because we do not find the idea of equal consideration of interests normatively plausible.

14 Social Contract, 3.14.

15 In the case of the US Senate, for example, votes are of unequal weight because the political system relies on a scheme of territorial representation in which districts (in this case states) correspond to political subdivisions: in this case, the inequality seems less objectionable because it can be provided with a rationale that does not offend against the requirement of treating members as equals.

[16] Such instrumental considerations played an important role in the Supreme Court's classic apportionment decisions, which urged that the same instrumental reasons supporting universal political rights also support equally weighted votes. Thus, in Gray v. Sanders: "No right is more precious in a free country than that of having a voice in the election of those who make the laws under which, as good citizens, we must live. Other rights, even the most basic, are illusory if the right to vote is undermined." Cited in Reynolds 377 US 533, at 558. Or in Reynolds v. Sims: "Especially since the right to exercise the franchise in a free and unimpaired manner is preservative of other basic civil and political rights, any alleged infringement of the right of citizens to vote must be carefully and meticulously scrutinized." Reynolds, at 562.

17 Theory of Justice, p. 359.

18 According to an epistemic conception of majority rule, the rationale is that decisions supported by a majority are more likely to be right, not simply that the process visibly assigns equal weight to the interests of each.

19 See Yannis Papadopolous, "A Framework for Analysis of Functions and Dysfunctions of Direct Democracy: Top-Down and Bottom-Up Perspectives," Politics and Society [ref].

20 Complexities arise when we think of systems with strong parties with well-defined policy positions and ways of disciplining members who depart from those positions. Such systems have a direct aspect. But we abstract from these subtleties here.

21 It is not reasonable to demand to win: assuming disagreement, any decision will be opposed by some people.

22 See, for example, Carl Schmitt, The Concept of the Political; Adam Przeworski, "Minimalist Conception of Democracy: A Defense," unpublished.

23 On the idea of consolidation, see Stepan, 1996 #98. For doubts about the importance of consolidation, see Adam Przeworski, 1996 #95. But note that Przeworski, et al., do not consider the importance of consolidation, as characterized in the text.

24 For discussion of these three views, see Cohen and Rogers, Associations and Democracy, pp. 000-000.

25 Moreover, if legislators can secure their own reelection by servicing constituents, then the limits on problem-solving are not a large source of electoral instability.

26 [footnote here on suspect classifications and fundamental interests]

27 Amy Gutmann argues for a "democratic threshold principle" in her account of the distribution of primary schooling. See Democratic Education, pp. 136ff. But she confines the threshold to "effective participation in the democratic process" -- as though there were not an equally good claim to effective participation in labor markets. [make reference to San Antonio]

28 See Iris Marion Young, "Communication and the Other: Beyond Deliberative Democracy," in Seyla Benhabib, ed., Democracy and Difference: Contesting the Boundaries of the Political (Princeton: Princeton University Press, 1996), pp. 120-35; Lynn Sanders, "Against Deliberation," Political Theory [ref]; Sidney Verba, Kay Lehman Schlozman, and Henry E. Brady, Voice and Equality: Civic Voluntarism and American Politics (Cambridge, MA: Harvard University Press, 1995), pp. 500-08.

29 Susan Sturm, "A Normative Theory of Public Law Remedies," Georgetown Law Journal 79, 5 (1991): 1357-1446. Also, Abers discusses the requirement that citizen-budgeters incorporate considerations of fair distribution in the deliberations in the Porto Alegre system. See Rebecca Abers, "Learning Democratic Practice."

30 Bargaining and command are ways of making collective decisions without mutual reason-giving; confession is the practice of giving reasons to others for individual decisions.

31 The paragraph that follows presents an account of public goods suited to the special setting of a conception of deliberative democracy.

32 See Robert Dahl, Democracy and Its Critics, p. 221.

33 The classic statement of this view is John Hart Ely, Democracy and Distrust (Cambridge, MA: Harvard University Press, 1980).

34 For thoughtful discussion of debates on social movements from a viewpoint close to the one adopted here, see Hans Joas, Die Kreativit鋞 des Handelns, Suhrkamp, Frankfurt a.M., 1992, pp. 348 ff.