PRAGMATIC EQUILIBRIUM

A conception of democracy articulates the central values that regulate collective decisions and actions in a democratic society, provides a benchmark against which to judge existing institutions, and guides efforts to improve those institutions (Pettit 1997, 102). Much of the debate in democratic theory concerns whether elite, pluralist, deliberative, participatory, or other conceptions are attractive and feasible. Every conception of democracy must possess at least three components to be complete. First, it must offer an articulation of the values that relate collective decisions and actions to the interests and views of the individuals who compose a collectivity. All democrats adhere to the value of nontyranny—that no individual or group should decide collective issues regardless of others' interests and preferences. Beyond that, different kinds of democrats variously endorse other values, such as public accountability, self-government, reasoned rule, common good, political equality, private liberty, self-actualization, and competent government. Second, every democratic conception must recommend institutions—for example, political liberties, competitive elections, universal suffrage, civic associations, referenda, town meetings, and peak bargaining arrangements—that advance its underlying values. Third, values and institutional prescriptions are typically connected deductively by presuming empirical facts—often quite stylized—about the political psychology and capabilities of individuals and about sociopolitical dynamics. The conceptions of democracy that normative theory produces can thus be conceived as packages of values, facts, and institutions in which institutional prescriptions are deduced from core values and presumed facts.

A conception of democracy is in pragmatic equilibrium just in case:

• PE1. The consequences of
• PE2. the institutions that it prescribes
• PE3. realize its values
• PE4. well and better than any other feasible institutional arrangements
• PE5. over a wide range of problems and contexts.

Pragmatic equilibrium thus specifies a relationship among underlying values and goals, institutional prescriptions, and the empirical consequences of those prescriptions. This formulation of pragmatic equilibrium articulates a consequentialist standard that is at one level unobjectionable. Conceptions of democracy offer institutional prescriptions to realize certain values.

Applying the standard of pragmatic equilibrium is straightforward in principle but complicated in practice. The first step begins with a well-specified conception of democracy—one with clearly articulated values and institutional prescriptions such as those in the main streams of democratic theory. The second step is to delineate a set of problems that require collective decision and action (PE5). Because there are infinitely many social and political problems, it is sensible to begin with the large categories of public problems that are important to citizens and to democratic governance. These might include not only national decisions relating to social, economic, regulatory, and foreign policy but also local concerns such as education and community safety. The third step is to test that conception by noting the consequences—through direct observation when possible and informed reflection otherwise—of making collective decisions through the institutions prescribed by the democratic conception under review (PE1, PE3). Fourth, the theorist then specifies the alternative ways in which collective decisions might be made for the problem in question and assesses the consequences of those alternatives (PE4). Again, it is impossible in practice to assess all feasible alternatives, so one reasonable way to proceed is by analyzing the most common ways of making the kinds of decisions in question together with uncommon methods that seem likely to have normatively desirable properties. If the originally prescribed institutions yield superior consequences, then that conception of democracy is in pragmatic equilibrium.

This standard creates a close link between normative theory and empirical research that is related to classical pragmatism, and especially to the political thought of John Dewey, in five distinct ways.

First, classical pragmatism takes problems as the source of reflective intelligence (Dewey 1896) and it takes interpersonal problems, or externalities, that arise from social and political interactions as the source of political institutions (Dewey 1954, 34–35). Pragmatic equilibrium takes problems and the contexts in which they occur as the domain of democratic theories. Just as theories of justice formulate judgments and principles to resolve disagreements between competing interests (Rawls 1951, 177), democratic theories describe the values and principles that should guide collective decisions about social problems. The same underlying values and principles may well require different decision-making arrangements for different problems and contexts. A democrat for whom deliberation and public reason are cardinal values need not prescribe citizen deliberation, or even deliberation, for every problem. This focus on variations in problems and contexts is analogous to the rejection of universalism and the problem-driven approach to empirical investigation urged by Donald Green and Ian Shapiro (Green and Shapiro 1996; Shapiro 2005, 51–99). Ceteris paribus, a conception of democracy that encompasses a wider range of problems and contexts is better—because it is more complete—than one that addresses fewer problems and contexts. For example, a conception of democracy that successfully accounts for collective decision-making at the local, national, and international levels is better than one that covers only the national level.

Second, as classical pragmatism holds that the meaning of an idea or belief is given by its practical consequences (James 1975, 28–29; Pierce 1878), pragmatic equilibrium focuses attention on the consequences of institutional designs (PE3). Third, pragmatists are fallibilists; they “recognize that even those beliefs about which [they] are most convinced may turn out to be false” (Knight and Johnson 1996, 567). The pragmatic equilibrium standard focuses theoretical attention on the mistakes of democratic conceptions. In particular, a democratic conception is false when the facts of the world depart from its stylized facts; its institutions are likely then to generate consequences that fail to realize its values. The following sections show how such unanticipated consequences render the major conceptions of democracy unstable in the sense that they are consequentially inconsistent; this is what it means to be out of pragmatic equilibrium.

Fourth, pragmatic equilibrium focuses on institutional variation and choice. There are many ways to organize collective decision-making in democracies and Dewey writes that the task of political analysis is to understand them (Dewey 1954, 100; Knight and Johnson 1999, 583): “All intelligent political criticism is comparative. It deals not with all-or-none situations, but with practical alternatives.” The menu of institutional alternatives is far richer than the dichotomy between representative and participatory democracy supposes, and most of the items on that menu remain empirically and normatively unexplored. PE2 and PE4 of the above formulation stress this diversity and the importance of exploring institutional designs.

Fifth, the approach offers one way to make sense of the pragmatic rejection of the dichotomy between facts and values (Putnam 1995). In particular, pragmatic equilibrium and the method for reaching it reciprocally determine the democratic values and institutional consequences that constitute a democratic conception.

Much of the recent scholarship on pragmatism in political theory has addressed its substantive political and institutional implications. Richard Posner (2003), for example, contends that contemporary pragmatists ought to endorse minimal, Schumpeterian institutions. Knight and Johnson (1996) respond that pragmatism entails radical democratic commitments. Though they disagree fundamentally with each other, both Posner and Knight and Johnson maintain a pragmatic orientation toward evidence that is shared by the approach laid out here. They seek to support their democratic conceptions with empirical materials and positive political theories (Knight and Johnson 1999). In contrast to these views, the idea of pragmatic equilibrium and its associated method of practical reasoning is initially neutral with respect to substantive conceptions of democracy. It is constructed as a tool for democratic theorizing rather than as a way to justify a particular model, position, or set of values. Prior to applying the test, there are no special reasons to think that minimal conceptions will fare any better (or worse) than radical democratic ones. Still less does pragmatic equilibrium include classical pragmatic commitments such as Dewey's endorsement of participatory institutions or his ethical ideal of growth, human flourishing, and self-realization (MacGilvray 2004, 132–33).

PRACTICAL REASONING ABOUT INSTITUTIONS AND VALUES

When a conception of democracy fails the test of pragmatic equilibrium, it may be modified in ways that bring it closer to fulfilling that standard. One straightforward procedure is to treat institutions instrumentally, as means to realize ends that are given by the values of a democratic conception. A procedure of revision would then search the space of feasible institutional alternatives for the combination that maximizes the accomplishment of the values of a democratic conception over the relevant range of problems. But the difficulty with this instrumental approach is that institutions are deeply connected to the democratic values they advance. Imagine a Schumpeterian who begins with a commitment to a certain kind of official accountability and to citizens' private liberty as basic values, but then realizes the facts of the world favor quite participatory arrangements to achieve these ends. Such outcomes can create pressures on a conception of democracy to revise not only its institutional prescriptions but also its underlying values.

The approach to pragmatic equilibrium favored here subjects both the values and the institutions of a democratic conception to revision through reflection. This procedure extends Henry Richardson's (1994, 57–61; 2002) account of practical reasoning for individual agents. According to Richardson, individual deliberation is not just instrumental; individuals' interests and goals can change when they act in concrete situations as a result of reflection, interaction, and discussion. Pace homo economus, it is a completely ordinary experience for individuals to use their practical reason not only to select the means that best satisfy their preferences and accomplish their goals but also to deliberate about ends themselves. When faced with new information, experiences, or situations, individuals may come to see the vagueness in their goals, recognize contradictions among them, and even adopt new ends. Imagine a rational politician whose overriding purpose is reelection. Surveying the field of possible wedge issues, he seizes on the scourge of homelessness (about which he knows little) as a means to political victory. But after meeting homeless men and women and their children during the campaign, he takes a genuine interest in their welfare. He becomes unable to stomach the cynicism that treating homeless people only as a means to electoral victory requires. Indeed, after his successful reelection, he occasionally supports policies that benefit the homeless in his city but that cost him votes. Homelessness begins for him as a means to the end of political success but it becomes an end in itself (Richardson 1994, 82–86).

Analogously, a democratic theorist may adjust the institutional and ethical commitments of his conception of democracy by reasoning practically about which institutions will work best and then about what is democratically valuable. To illustrate, consider the stages through which practical reason about a problem in context can work to internally transform a conception of democracy. Suppose that some city actually adopts a participatory democratic system of neighborhood assemblies to govern important local concerns such as the provision of city services, urban infrastructure, and land use. That city, furthermore, makes substantial investments that create favorable conditions for the success of this participatory endeavor—it funds community organizers to mobilize participants and training workshops to equip them to take on the work of sublocal governance. If these arrangements generate effective and just public decisions through a deliberative process that transforms individuals into more public regarding citizens, then a participatory conception of democracy is vindicated for at least this context.

Unintended consequences, however, may trigger the second step of the practical reasoning process. In many contexts, real facts do not conform to the stylized facts of a conception of democracy. Reality thus intrudes on theory by invalidating its institutional deduction. Suppose, for example, that in some neighborhoods where participation is high, so many people show up to meetings that decision processes become unwieldy and many can only listen and observe rather than express and engage. In others, too few participate to make decisions competently. The system ends up working best where a handful of residents take a deep and ongoing interest in these affairs. The individuals in that small group gain the expertise necessary to comprehend the complexity of urban governance choices. They serve as a focal point for other neighbors who lack participatory inclinations to communicate their needs and interests. And they are generous and representative, so they produce public goods and perform genuine service.

The third step of this reasoning process involves revision of the conception of democracy in light of such new facts and experiences. The participatory democrat might respond by searching for other institutional forms that better vindicate his prior value commitments (to political equality, self-government, reasoned rule, and participation). In light of this experience, he may come to think that the best institutions for realizing participatory democracy involve small groups of relative amateurs focusing on public problems in a sustained way, rather than on larger groups (in principle everyone) that deliberate less intensively. So long as there are enough small groups and issues so that everyone participates in some decisions (but not everyone in any particular decisions or anyone in every decision), he may regard his values as well served. More deeply, the participatory democrat might also revise the values that ground his conception. This experience might show, for example, that making wise infrastructure or land use decisions is difficult work and he might come to appreciate more deeply the consequences of bad decisions. This realization might promote a previously subordinate value—that of governmental competence. He may also soften his objection to representation and thus respecify his notion of participation when he recognizes that what the small groups of volunteers in the neighborhood association are doing is in part representing the interests and views of those who do not participate directly.

In this third step, then, incommodious facts trigger the revision of a conception of democracy in two possible ways. When facts may make it clear that a received institutional prescription (everyone should participate in direct deliberation) fails to advance the values (self-government, participation, political equality) that justify it, the first response is to cast about for alternative institutional arrangements that better realize those values. A second response is to revise the democratic values themselves in ways that respond to the possibilities and constraints revealed by empirical analysis or reduce vagueness or inconsistency in a conception's values.

This process of practical reasoning occurs within a single conception of democracy. But if facts of contexts break the tight connection between configurations of democratic values and their institutional implications, then there is at least a possibility of agreement—in particular contexts—between previously antithetical conceptions of democracy. It may be, for example, that a Schumpeterian comes to acknowledge all of the failures of governance through a corrupt representative machinery in some city and to support the same institutions as a participatory democrat there. They may agree upon the same set of institutions based upon their very different unreconstructed value commitments. Or, more deeply, they may rearticulate the values that ground their respective conceptions of democracy in ways that are more compatible with one another. In the urban governance example, the Schumpeterian may come to think that the ordinary citizens need not “fall to a lower level of mental performance” at least when considering local issues and thus that some stronger notion of self-government is possible and desirable (Schumpeter 1942, 262). He may also come to think that in particular hostile contexts, extensive citizen participation is necessary for effective public accountability.

This method shows only that it is possible for proponents of particular conceptions of democracy to revise them rationally by reflecting upon the consequences of political practices, not that they will necessarily do so. For many problems, compelling evidence regarding the comparative effects of alternative institutions may be unavailable. In such cases, conceptions of democracy should make their empirical and institutional speculations explicit to facilitate further investigation. Even when faced with substantial disconfirming evidence, however, commonplace cognitive and social biases may lead theorists to ignore or reject it. The recent history and sociology of science has demonstrated that physicists and chemists often resist new facts that would require them to revise their settled theories. Political theorists are not immune from these same tendencies. But it is possible to mitigate their effects. Theorists who engage in practical reasoning should seek the stance of inquiry of Rawls's (1951) “competent judges.” A competent judge “exhibits a desire to consider questions with an open mind, and consequently…is…always willing to reconsider it in light of further evidence and reasons” and “knows, or tries to know, his own emotional, intellectual, and moral predilections and makes a conscious effort to take them into account in weighing the merits of any question.” This formulation is highly congenial to the emphasis that American pragmatists such as Pierce, James, and Dewey placed on actively seeking experiences and evidence that falsify beliefs and their hostility to dogma (Putnam 1995, 171). The standard of pragmatic equilibrium vastly enlarges the range of potentially disconfirming data, and so the sources of reflection and refinement, by demanding that democratic theories account for the results of empirical research.

A related concern stems from the inevitable sensitivity of assessments to the specific problems and alternatives subjected to analysis through the pragmatic equilibrium standard. Suppose that we have reached what appears to be a pragmatic equilibrium for some conception of democracy by working through many contemporary problems and institutional prescriptions. It may be that considering additional problems or other institutions would reveal that this point is not a pragmatic equilibrium at all. Since the argument for any particular pragmatic equilibrium depends on the cases that have been selected to reach it, the method is vulnerable to selection bias. In this regard, however, it does not differ from most empirical methods and the general remedies apply. Those defending a conception—and their critics—should seek to test it with ever more problems and alternatives. Though the illustrations of practical reasoning below utilize brief case study materials, this method should in principle incorporate quantitative empirical studies as well. Even so, such a process can never be exhaustive. But in the course of progressively developing a conception of democracy, its values and institutions become more adequate, relevant, feasible, and actionable as it incorporates additional problems and contexts.

FOUR CONCEPTIONS OF DEMOCRACY

To understand how the pragmatic equilibrium standard and the practical reasoning procedure operate in practice, consider four major conceptions of democracy—minimal, aggregative, deliberative, and participatory—as value-deduced institutional packages.¹

RULES OF THE GAME

From time to time, every democratic system must revise the arrangements through which political leaders are selected. These changes include adjusting the boundaries of electoral districts as well as altering voting rules, for example, by shifting from majority rule to proportional representation. Such procedural revisions are less common but more fundamental than the formulation of routine laws and policies. They also reveal substantial difficulties in the four conceptions of democracy.

Consider three alternatives for deciding such questions:

• I1. Elected representatives, authorized to make other laws and policies, also make decisions about political rules.
• I2. Politically insulated and/or neutral experts—such as jurists, mathematicians, or civil servants—resolve such matters. In 20 American states and Canada, redistricting decisions are made at least in part through independent or bipartisan commissions (McDonald 2004).
• I3. A body of ordinary citizens, as in a criminal jury, decides the issue. In 2004, the Canadian province of British Columbia created a Citizens' Assembly composed of 160 randomly selected individuals from across the province to deliberate about whether B.C. should replace its majoritarian electoral system with some form of proportional representation.

Empirical evidence regarding the consequences of the first alternative (I1) poses problems for the first three conceptions of democracy (McDonald 2004). Even if legislators are properly authorized to make many kinds of laws and policies, it may not be appropriate for them to fashion the rules according to which they gain and lose those powers. In such decisions, many elected officials may be principally interested in crafting rules that protect their incumbent individual or partisan advantages, whereas their electors desire nearly the opposite: extensive political competition or wide choice among representatives. In U.S. states, redistricting decisions occur primarily through the ordinary legislative process. Thirty-eight states make their Congressional redistricting plans and 26 compose the electoral boundaries of their state legislatures in this way. Two outcomes tend to result. If one or other party controls the legislature, they produce partisan gerrymanders that favor the controlling majority. If control of government is divided between the houses of a state's legislature, or between the legislature and the executive, lawmakers tend to ratify a compromise incumbent protection plan in which each party makes its own seats safe. A rule-making process whose central result is to protect the prerogatives of those who make the rules has little democratic value.

Commissions of experts (I2 earlier) are the principal alternative to ordinary legislative processes in fashioning the rules of politics (Thompson 2002, 173–79). Compared to legislative redistricting, such commissions seem attractive for two reasons. First, they may be politically insulated from the corrosive calculus of incumbent advantage. Second, these commissions may be composed of individuals who possess legal, mathematical, ethical, or political expertise that enables them to make wise decisions. But autonomy is difficult to achieve when the stakes are so high. Most redistricting commissions in the United States do not even pretend to possess such independence. Members are frequently appointed because of their political affiliation and reliability. As a result, U.S. redistricting commissions often make decisions that resemble the incumbent-protecting outcomes of ordinary legislative processes.

But a degree of political insulation is possible. In Arizona, for example, a popular referendum in 2000 created a distinctive redistricting commission that operates with evident independence. State legislators appoint four members, two from each major party, from a pool created by the judicial appointments panel. Those four members then appoint a fifth, politically unaffiliated, member to serve as chair. The commission's charge is to develop a redistricting map that is attentive to compactness, existing boundaries, and competitiveness. The law prohibits the commission from considering party registration and voting history data in the initial phase of the mapping process, and from identifying the locations of incumbents' or candidates' residences altogether. The commission approves maps by majority vote and their plans are not subject to legislative review or gubernatorial veto. Though it is widely acknowledged as an exemplar of independent commission redistricting (Macedo et al. 2005, 58), the Arizona process is not without difficulties. Hispanic democrats have sued Arizona on the grounds that the map fails to produce sufficiently compact districts and the city of Flagstaff has sued on the ground that the commission failed to respect traditional communities of interest. Since 1992, more than 90% of incumbents seeking reelection in Arizona have won (Thompson 2002).

The important criterion of independence, however, does not substantially differentiate well-constituted commissions (I2) from citizens' assemblies (I3). Indeed, a group of randomly selected citizens may be more politically autonomous than notables who have been vetted by legislators. But an assembly of experts may be more competent than one composed of ordinary citizens. Conversely, the enduring popularity of juries in criminal and civil trials (Abramson 2000; Gastil 2000; Leib 2004) suggests that the lay character of citizen assemblies confers a certain democratic legitimacy and virtue. They may introduce popular values and perspectives without the freight of partisan self-interest.

If there is a sharp and inevitable trade-off between the competence of expert commissions and the popular sentiments and legitimacy of citizens' assemblies, then different conceptions of democracy will balance these reasons differently. If, on the other hand, ordinary citizens generally possess or can with reasonable effort acquire the knowledge and skills necessary to make wise decisions in this area, then (other factors equal) considered judgment favors assemblies of citizens over experts deploying their technical prowess.

A recent experience from Canada suggests that ordinary citizens can acquire the competence necessary to design political institutions. The Liberal Party government of British Columbia recently created a participatory mechanism to recommend whether the province should keep its system of single-member, plurality-winner elections or replace it with some other voting system (British Columbia Citizens' Assembly on Electoral Reform 2004). The Citizens' Assembly was composed of 160 citizens who had been randomly selected from provincial voting lists. To assure a degree of descriptive representativeness, selection was stratified by region and gender.¹

Members decided that B.C.'s electoral system ought to serve three fundamental values: fairness, understood as proportionality in the allocation of legislative seats; local representation, understood as the connection between an elected representative and her geographic constituency; and voter choice, understood as the number of candidates and parties. To analyze the merits of alternatives, members simulated the operation of various voting systems. Eventually, Assembly members settled on two alternatives—a mixed member proportional (MMP) system and a version of the single transferable vote (STV). The STV option defeated MMP by a 123 to 31 in a vote of Assembly members. Bypassing the legislature, the citizens of British Columbia considered this recommendation in a provincial referendum in May 2005. The threshold for passage consisted of a “double-majority” of (i) more than 60% of the total ballots cast and (ii) more than 50% of the ballots cast in 48 of the 79 constituencies (i.e., a simple majority in more than 60% of the ridings). The measure won a majority in all but two of the constituencies, but it garnered only 57.4% of the total vote. Though the result fell just short of the required super-majority threshold, it appears that many voters did consider the Assembly process legitimate. As of this writing, the Liberal Party government has announced that it will hold a second referendum on the Citizens' Assembly proposal in 2008 to allow time for greater public debate. The situation also creates a certain awkwardness; the Liberal Party holds power by virtue of an electoral system that a majority has voted against.

Experience and analysis does not yet provide definitive conclusions regarding the consequences of these different institutional arrangements. Suppose, however, that the following propositions—backed by the compressed descriptions cited earlier—are true. First, elected officials will act to advance their own electoral prospects rather than other important values such as competitiveness or the coherence of districts when given the opportunity to make rules of the game. Second, citizens' assemblies are more easily insulated from undue political influence than expert commissions. Third, it is possible and feasible to structure a citizens' assembly in such a way that the participants gain sufficient mastery of the subject. That is, suppose that the alleged success of the British Columbia Citizens' Assembly can be repeated for other voting rules decisions and for electoral redistricting. These consequences compel revision to the four conceptions of democracy.

Prior to confronting these realities, the minimal democrat favors the prevalent method of redistricting in the United States: let the legislature work it out. After that, he supports the expert commission due to its presumed expertise, and his last choice (again for reasons of competence) is the citizens' assembly. His initial institutional preference ordering is I1 > I2 > I3. On considering the facts just stipulated, he must alter this ordering along with deeper elements of his democratic conception because the institution (I1) he favors fails to advance his values. Letting a team of politicians make the rules of the game at one point in time undermines the competitiveness of the institution later, and competition is the lynchpin of public accountability for him. For similar reasons, he cannot support commissions that have difficulty maintaining their independence, I2. He is quite surprised that participants in the Citizens' Assembly managed to gain the topical mastery that they evidently did and that they were able to endorse a common solution. So, at least in this case, he reluctantly supports I3 instrumentally, as a set of institutions that better advances his democratic values of public accountability and private liberty.

But these revisions in his assessment of individual political capacities and appropriate institutional forms also lead to shifts in his underlying values. Reflecting on the Citizens' Assembly, he comes to see how direct citizen participation can be valuable (perhaps even indispensable) to periodically check the power of elites. The Citizens' Assembly experience also opens for him a limited sense of possibility regarding the values of common good and self-government. All citizens do have in common the good of an electoral system that preserves competition among elites. A kind of democratic self-government does seem possible in determining the content of that system, though he remains skeptical that these values are relevant for other issues.

The Citizens' Assembly is a peculiar institution for the aggregative democrat because it successfully combines the interests of British Columbians but does so by privileging a handful of citizens. The aggregative democrat has little difficulty accepting the facts about institutional alternatives stipulated earlier. Given the incentives that institutions create, there is ample reason to think that the conclusions of a Citizens' Assembly would match more closely the interests of citizens at large than a legislative committee or commission of experts (Dahl 1989, 338–41). The institution thus offers the aggregative democrat a better way to establish rules that make good on his commitments to political equality and self-government. But the Assembly's experience is also troubling and may cause him to rethink the values of his conception in at least two ways. First, Assembly participants did not calculate their positions based on their own advantage. Instead, they developed a consensus around values that an electoral system should serve and then deliberated about which system best advances those values. This evidence presses the aggregative democrat to incorporate the possibility of deeper, deliberative agreement on fundamental questions such as design of the political structure. Second, he may be pressed to weaken his commitment to self-rule through universal participation, at least on complex and obscure issues such as electoral design. Most citizens of British Columbia—and most members of the Assembly when they began—had no command of alternatives such as MMP and STV systems. Over the course of their deliberations, many of the Assembly's members became expert on these issues and so quite unlike the electorate at large. If their decision has democratic value, it is in part because Assembly members came to have views that were more informed and wiser than the general electorate and because they exercised more authority over that decision than other citizens.

The Citizens' Assembly experience triggers relatively minor instrumental revisions for the deliberative democrat. It helps settle what was for him an open question: whether a legislative committee, independent commission, or popular assembly is more likely to produce electoral policies backed by reasons that are acceptable to all. In revising his view in this minor way, he moves closer to deliberative democrats who favor direct citizen participation (Fishkin 1995; Gastil 2000; Leib 2004) than those who view representative bodies as the main sites of political deliberation (Bessette 1997; Richardson 2002). But his central commitment to self-government understood as reasoned rule survives unscathed.

Although the directly deliberative character of I3 makes it preferable to I1 and I2 for the participatory democrat, he objects to the extremely limited scope of participation in the Citizens' Assembly; 160 people is a miniscule fraction of the population. Especially for decisions of quasi-constitutional importance, he favors shifting more authority to citizens at large. For example, a Citizens' Assembly might frame several options and then put them to the entire electorate through a referendum (Barber 1984, 281–89). Suppose, however, that quality of deliberation in the wide-open public sphere—even under quite favorable conditions—is necessarily substantially lower than that of a Citizens' Assembly.

This unanticipated trade-off between two important values—the quality of a decision and the quantity of popular participation—would compel revision in the participatory conception. The participatory democrat might, for example, revise his institutional prescription to fit a weaker principle of participation in which (i) every citizen participates in some important public decisions, (ii) no citizen participates in every decision, (iii) there is no decision in which every citizen participates intensely, and (iv) some citizens participate intensely in every important public decision. That is, some participate in the Citizens' Assembly about voting rules, others in decisions about electoral districts, and still others in the governance of public services. That institutional vision—in which everyone benefits from some experiences of face-to-face self-government—may be the best feasible reconciliation of the values of participation and competence. But that revision would compel a reconsideration of the values of the participatory conception. In particular, it would force the participatory democrat to acknowledge the important role for representation; even in a citizens' assembly, participants represent the broader population in a certain sense.

TYRANNY OF MINORITIES

Decisions about political structure are by definition extraordinary and perhaps a democratic conception should not be much faulted for failing to account for them. A far more common problem in the governance of all contemporary democracies is the tyranny of powerful minority factions. Though democratic theorists have focused on majority tyranny and its remedies, minority tyranny remains a large and daunting challenge in every society. There are many kinds of minority tyranny and many have been rejected by pluralist critics—such as the notion that governments of capitalist societies operate as executive committees of the bourgeoisie—as either too vague to verify or plainly incorrect (Dahl 1989, 265–79). Nevertheless, political science has established at least four major sources of unequal influence: material resources that can be translated into political influence (tyranny of the rich, see Cohen and Rogers 1983), knowledge (tyranny of experts, see Dahl 1989), strategic position (tyranny of concentrated interests, see Lowi 1989 and Wilson 1980), and institutional advantage (e.g., residents of small states vs. those of large ones).

In the urban areas of many developing countries and some developed ones, these sources of minority tyranny conspire to create a politics of patronage whose consequences are rejected by the four conceptions of democracy. When governance is infected with patronage, politicians make decisions about public investments in clinics, schools, houses, roads, and other basic infrastructure by selecting contractors and employers who advance their political fortunes rather than selecting projects to improve public welfare. Those who benefit from using public infrastructure are typically much more numerous, and so more difficult to organize, than the producers who gain by reaping public funds to build that infrastructure. Producers can therefore exert undue influence on decisions about what gets built by whom. In classic clientelist arrangements, they exert greater influence still by using their resources to organize popular support for politicians who direct favors to them. These arrangements are frequently protected from scrutiny and reform by the shield of expertise. Politicians, line departments, and contractors all benefit from claiming that such decisions are the province of analysts and technicians.

“Participatory Budgeting” (Orçamento Participativo, or OP) is a method of allocating public investments that has addressed the challenge of patronage. It was developed first in the southern Brazilian city of Porto Alegre and has spread in recent years to dozens of other cities in South America. Prior to 1989, control over municipal funds in Porto Alegre lay principally in the hands of city councilors. They made these investments according to a logic of patronage in which benefits were doled out in exchange for the support of well-organized groups of contractors and well-off residents. In 1989, the left-wing Workers' Party (Partido dos Trabalhadores, or PT) was elected to the city executive based in part on its promises to empower the city's civic organizations. Over the next 2 years, they developed the OP (Abers 2000; Avritzer 2002; Baiocchi 2003, 2005). The mechanism shifts decisions over the capital portion of the budget from the city council to a system of neighborhood and citywide popular assemblies. Through a complex annual cycle of open meetings, citizens and civic associations meet to determine investment priorities. The infrastructure portion of the city's budget results from aggregating these priorities.

Though it is a procedural reform, the Participatory Budget was born of a substantive motive to “invert” public spending priorities by shifting them away from the wealthy areas of the city to poor neighborhoods. It has achieved that goal remarkably well, and Porto Alegre's poor residents enjoy better public goods and services as a result. The percentage of neighborhoods with running water has increased from 75% to 98%, sewer coverage has grown from 45% to 98%, and the number of families offered housing assistance grew 16-fold since the initiation of the OP. Furthermore, the reforms seem to have suppressed patron–client relationships. In surveys, 18% of long-time neighborhood association leaders admit to engaging in client–patron exchanges of benefits for political support, whereas only 2% of leaders who became active after the initiation of the OP did so (Baiocchi 2005). Another study found that 41% of associations secured benefits by directly contacting politicians prior to the OP, whereas none relied on such unmediated channels after its establishment (Avritzer 2003).

Suppose that there is a set of sociopolitical urban conditions under which allocating power over public infrastructure decisions to elected officials (I1) or public agencies (I2) will result in patronage. But adopting a system of participatory budgeting (I3) reduces patronage. These empirical assertions may be a fair characterization of the recent history of Porto Alegre and other cities in developing countries.

These consequences press the minimal, deliberative, and participatory conceptions closer to the aggregative one. Nothing about the consequences of I1 or I2 is inconsistent with the minimal democrat's emphasis on competitive elections or with its underlying commitments to public accountability and nontyranny. If patronage problems rise above some threshold—say under investing in public health to the extent that an epidemic strikes—the party in power would presumably suffer electoral consequences. Instead, pressure on the minimal conception comes not from the ills of patronage but from the possibility of a coherent and more demanding conception of self-governance revealed by the Participatory Budgeting experience. Minimal democracy favors comparatively low standards of public accountability in part because it supposes that citizens' preferences are unclear or unwise. The Participatory Budget shows that there are conditions under which citizens can have clear and reasonable preferences over issues concerning basic urban infrastructure and that feasible political institutions can make government responsive to those preferences. In these circumstances at least, the minimal democrat's skepticism regarding the existence and rationality of citizens' desires is unwarranted.

In light of these possibilities, the minimal democrat is compelled to revise the values of his conception. In particular, the very same reasons that justify the threshold of public accountability provided by competitive elections also favor increased governmental responsiveness when that notion is sensible, as it is in the case of Porto Alegre. Institutionally, competitive elections are for the minimal democrat an instrument of public accountability. If the invention of the Participatory Budget is more efficacious in that regard, it calls for revision of his institutional prescription.

The Participatory Budget is more congenial to the aggregative conception. It constitutes the discovery of a form of direct participation for a relatively large city that seems to allow greater opportunities for participation and more responsive government than the aggregative democrat had previously supposed possible. Adoption of participatory budgeting is a relatively minor instrumental revision to the institutions of the aggregative conception that requires no adjustment of its underlying democratic values.

Residents may deliberate to an extent when they discuss the merits of various projects for the quality of neighborhood life. Even then, decisions in the Participatory Budget may revolve centrally around (aggregative) bargaining to satisfy individuals' needs and wants. Insofar as participants assert their interests and preferences without offering reasons to one another and pay little attention to any good that they have in common, the Participatory Budget falls short of deliberative expectations. But the gains from a more deliberative OP are unclear. Unlike deep moral disputes or conflicts over the rules of a common political system, the best reasons to support one public infrastructure choice—a clinic versus a school or paved street—over another may well hinge on the extent to which those choices satisfy preferences fairly. If so, then infrastructure investment decisions constitute an area in which deliberation reduces to aggregation; cases in which preference aggregation is the most reasonable procedure are degenerate instances of deliberation. It isn't that the participatory budget isn't deliberative, but rather that deliberative accounts add little, in this case, to the aggregative one.

The empirical assertions that drive these revisions depend on circumstances of institutional and political development. The extravagant participation that Participatory Budgeting requires can be regarded as a cost, but one worth paying to reduce corruption. If some fortunate cities possess professional bureaucracies, competitive elections, and vigorous media that are sufficient to generate honest and competent infrastructure investment decisions, then many of the reasons to favor participation fall away. The various conceptions of democracy would then diverge on institutional prescriptions. Minimal and perhaps aggregative democrats would favor representative arrangements whereas participatory democrats would still value the opportunities for direct citizen engagement created by the Participatory Budget.

These empirically informed thought experiments show how challenges such as minority tyranny can create occasions for rationally revising conceptions of democracy. Clear pressure for revision comes from the unintended and undesirable consequences of institutions recommended by various conceptions of democracy. But these institutional mismatches arise in turn because various problems and contexts make different democratic values salient and urgent. With making fair political rules, deliberation and constrained forms of citizen participation can be instrumentally valuable even to a minimal democrat due to the difficulties of political insulation faced by the other alternatives. With minority tyranny in urban clientelism, by contrast, fairness is secured primarily by institutions that accurately aggregate preferences over public goods and services, and public deliberation seems to be less relevant to the problem.

At minimum, these reflections suggest that each of the pure conceptions of democracy is institutionally incomplete. Each will be compelled to adopt ethical commitments and practices from the other views in the course of justifying particular democratic institutions in context. That is, the deliberative democrat will likely be compelled to make room for elements interest aggregation, and the minimal democrat may incorporate occasional deliberation and even direct participation. But this analysis does not produce a substantive judgment about the relative promise of the four conceptions of democracy. That would require considering many more problems and institutions. Other issues and contexts—perhaps political stability in transition societies (Przeworksi 1999) or issues in which technical expertise is indispensable but difficult to acquire (Dahl 1989, 332–35)—might well highlight the comparative advantage of minimal democracy as a starting point. Nevertheless, these applications illustrate how practical reasoning presses existing conceptions of democracy to develop institutional prescriptions and underlying values that are more differentiated and conditional in order to respect the complex texture of contemporary problems and conditions.

PRAGMATIC AND REFLECTIVE EQUILIBRIUM

Pragmatic equilibrium as a standard for conceptions of democracy is a species of John Rawls' (1951, 1971, 46–53) general idea of reflective equilibrium. Several differences between democratic theories and theories of justice make the version of reflective equilibrium developed here—a variant that focuses upon empirical consequences—more useful for the assessment and development of democratic conceptions than Rawls' formulation.

Rawls offers reflective equilibrium as a test for, and method of developing, principles of justice (Harman 2004; Pettit 1997). Reflective equilibrium seeks fit between considered moral judgments in a limited set of settled cases on one hand and, on the other, a conception of justice composed of general principles that yield judgments in those and other cases (Daniels 1996; Scanlon 2002). Rawls writes that “A conception of justice characterizes our moral sensibility when the everyday judgments we do make are in accordance with its principles” (1971, 46). Reaching reflective equilibrium is not simply a matter of inducing principles of justice that best fit the data of considered judgments; rather it is a dialogic process that works back and forth between principles and judgments. That is, an appealing set of principles that generates a verdict contrary to strong moral intuition may cause one to revise one's judgment in that case in order to achieve greater coherence across many cases.

Pragmatic equilibrium shares this dialectical structure. The major difference is that its “data” consist of empirical evidence about the consequences of institutions rather than considered moral judgments. Three considerations necessitate this shift. First, democratic theory lacks the sort of relevant reliable judgments that Rawls contends are available for theories of justice. Even an intuition as basic as “one person, one vote” is contradicted by common and well-justified practices such as reserved legislative seats for women or minorities and the representation of political units without regard to population as in the U.S. Senate and the United Nations.¹

After his discussion in A Theory of Justice, Rawls (1974) and subsequent theorists (Daniels 1996; Scanlon 2002) distinguished between narrow and wide reflective equilibrium. Pragmatic equilibrium is in one sense narrow but in another sense wide. In narrow reflective equilibrium, one works between a particular conception of justice (say, utilitarianism, libertarianism, or a Kantian view) and considered judgments about cases. This procedure may allow for “the smoothing out of certain irregularities” (Rawls 1971, 49) without subjecting the initial conception to fundamentally different and challenging alternatives. To attain wide reflective equilibrium, by contrast, one compares all of the possible conceptions of justice against one another to find the conception that best coheres with one's considered judgments. Rawls argues for his device of the Original Position and its veil of ignorance as justified by reflective equilibrium considerations. His pairwise comparison of the two principles with competing conceptions of justice is an argument that his two principles are those that survive the test of wide reflective equilibrium (Daniels 1996, 48–50).

The concept of pragmatic equilibrium is clearly a kind of narrow reflective equilibrium. Each conception of democracy evolves to more coherent and consistent points through the practical reasoning process beginning from its initial values and institutional prescriptions. Pragmatic equilibrium as stated earlier offers no analog to the Original Position, no Archimedean point from which to compare minimal, aggregative, deliberative, participatory, and other conceptions of democracy. It may be that democratic theorists will develop such a device in the future. Or, it may be that a widely compelling justification for one pragmatic equilibrium democratic conception over the others will never be available because differences of fundamental values are too deep or, as Dewey (1927, 31–32) thought, because the interplay among changing problems, appropriate institutions, and related values renders all such equilibria temporary. These important questions lie beyond the current enterprise.

But the current level of development of conceptions of democracy, deliberative theory notwithstanding, is such that attaining the coherence among values, institutions, and consequences that pragmatic equilibrium requires—even temporarily—would be a breakthrough. Despite decades of work on both participatory and deliberative democracy, there is little in the way of suggestions at all, much less agreement, regarding the sorts of institutions that those conceptions require. Furthermore, working toward narrowly reflective, pragmatic equilibria for various conceptions of democracy may well eliminate some conceptions as unfeasible or undesirable. The process of conceptual development through practical reasoning may also reduce the extent of the differences among the conceptions. I think it likely, for example, that the minimal conception will, on reflection, shift toward the aggregative one and that the participatory conception will be pressed toward the deliberative one. The contribution of narrow reflective equilibrium to the development of democratic conceptions is large and more important at this juncture in political theory than for theories of justice that are by comparison mature.

Another way to distinguish narrow from wide reflective equilibrium, orthogonal to the first, refers to the sources of information used in making considered judgments that discipline more systematic and principled conceptions. On this dimension, narrow reflective equilibrium draws only on firm judgments that stand independently of most contingent and empirical considerations. Reflection that is wide on this dimension draws on the full range of knowledge about contexts, institutional alternatives, consequences, and other “all things considered” concerns. One can interpret Rawls moving from narrow to wide reflective equilibrium in this sense when he describes the “four-stage sequence” in which the veil of ignorance is gradually lifted to allow increasing amounts of information in the selection of principles of justice, constitutional design, legislation, and finally administration, adjudication, and citizen action (Rawls 1971, 195–201). These two dimensions of wide and narrow reflective equilibrium are depicted in Table 1.

In A Theory of Justice, the course of Rawls' reflective equilibrium moves schematically from I to III and then IV in the cells above. That is, he takes the conceptions of justice that have been worked out in political philosophy (I) and then develops the device of the original position to select the preferred conception of justice (III). For those two principles, he works out the institutional implications in light of the facts of advanced industrial societies (IV).

I have suggested a different course for the development and justification of conceptions of democracy that moves schematically from I to II without speculating about whether or how to move to IV. Democratic theorists should widen their sources of inspiration and constraint to include the disciplined consideration of the consequences of the fullest range of institutional alternatives for collective decision making and action. That course of practical reasoning toward pragmatic equilibrium would enhance the plausibility, feasibility, and coherence of the major conceptions of democracy and others as well. Practical reasoning may also reduce the extent of disagreement between contending conceptions at the level of institutions and specific problems. Those two advances would mark substantial progress in political theory. Achieving them, however, requires synthesizing the normative and empirical study of democracy.