Power Sharing in Deeply Divided Places

This volume considers an array of power-sharing systems in divided cities and states, with critical evaluations of their merits and defects as well as explanations of their emergence, maintenance, and failings.

Power Sharing in Deeply Divided Places

Joanne McEvoy and Brendan O'Leary, Editors

2013 | 448 pages | Cloth $85.00
Political Science
View main book page

Table of Contents

Chapter 1. Power Sharing in Deeply Divided Places: An Advocate's Introduction
Brendan O'Leary

Part I. Power Sharing and Electoral Systems

Chapter 2. Electoral Rules and Ethnic Representation and Accommodation: Combining Social Choice and Electoral System Perspectives
Bernard Grofman

Chapter 3. The Track Record of Centripetalism in Deeply Divided Places
Allison McCulloch

Chapter 4. Electoral Engineering for a Stalled Federation: A Countrywide Electoral District for Belgium's Federal Parliament
Kris Deschouwer and Philippe van Parijs

Part II. Historical and Conceptual Forays on Power Sharing

Chapter 5. A Theory of Accommodation Versus Conflict: With Special Reference to the Israel-Palestine Conflict
Ronald Wintrobe

Chapter 6. The Success of Religion as a Source for Compromise in Divided Empires: Ottoman and Safavid, Past and Present
Benjamin Braude

Chapter 7. Geopolitics and the Long-Term Construction of Democracy
Randall Collins

Chapter 8. Courts, Constitutions, and the Limits of Majoritarianism
Samuel Issacharoff

Part III. Contemporary Power-Sharing Questions

Chapter 9. A Revised Theory of Federacy and a Case Study of Civil War Termination in Aceh, Indonesia
Alfred Stepan

Chapter 10. We Forbid! The Mutual Veto and Power-Sharing Democracy
Joanne McEvoy

Chapter 11. Northern Ireland: Power Sharing, Contact, Identity, and Leadership
Ed Cairns

Chapter 12. Public Opinion and Power Sharing in Deeply Divided Places
Colin Irwin

Chapter 13. The Balkans: The Promotion of Power Sharing by Outsiders
Florian Bieber

Chapter 14. Governing Polarized Cities
Scott A. Bollens

Chapter 15. Power Sharing in Kirkuk: The Need for Compromise
Liam Anderson

Chapter 16. Power Sharing: An Advocate's Conclusion
Brendan O'Leary

List of Contributors
Index
Acknowledgments


Excerpt [uncorrected, not for citation]

Chapter 1
Power Sharing in Deeply Divided Places: An Advocate's Introduction
Brendan O'Leary

The Mafia makes offers that cannot be refused. In one peace process a politician was once accused of making offers that no one could understand (O'Leary 1990). Do these statements explain the difference between power and power sharing? Is power coercive capacity, whereas power sharing is incomprehensible?

Power sharing is not incomprehensible, but it is frequently misunderstood. To aid comprehension a comparison is useful. In standard English, power is the ability to act, to be able to produce an intended effect (Russell 1992 [1938]). The powerless lack the capacity to do things they might want to do. The powerful are in the opposite situation. Power sharing, therefore, suggests spreading access to the capacity to get things done. Power is also a synonym for authority, jurisdiction, control, command, sway, or dominion, as well as the capacity to persuade, induce, constrain, oblige, or force. It follows that power sharing minimally means widening the access of persons or groups to the same domains or attributes. In standard usage power is also "a possession," "held" by those with authority or influence over others, especially public officials, governments, officers, managements, or establishments who constitute what Paul's Letter to the Romans described as "the powers that be." Power sharing, therefore, broadens membership of "the powers that be." It also requires that the included parties have access to key and observable "decision making." There must be no important "non-decision making" taking place off stage, that is, no hidden possessors of power who control the agenda or exclude some issues from being addressed. There must instead be an open and negotiable public agenda among the power-sharers, or at least among their leaders. Any suppression of (controversial) issues must be mutually agreed upon among those who share power.

Theorists contrast "power to" and "power over" (see Morris 2002; Parsons 1969). "Power to" is ability, "power over" is domination. The contrast resembles that between "positive-sum" and "zero-sum" relationships. "Positive-sum" power is joint, collaborative, or cooperative. All gain from its exercise, even if the benefits are not the same for all. "Zero-sum power," by contrast, describes a distinct antagonism: if power could be measured, then A's gain and B's loss would sum to zero. Positive-sum and zero-sum conceptions do not exhaust the logical possibilities of power relations. The exercise of power may generate net losses (a "negative sum") or the mutual ruin of the contending parties. It may create winners and losers; there may be disparities in benefits among the winners as well as in losses among the losers; and only one party may gain, while the others experience no net losses. Power sharing, for its proponents, is defended as "power to." It enhances collective capacity; it is "positive sum." Those who share will gain from a constructive way of making public decisions, from which all stand to gain, notably through the preservation of order and peace. Critics, by contrast, suggest that power sharing shapes public life at the expense of other and better kinds of politics—more competitive, individualist, or harmonious.

The opposite of power sharing is power's monopolization by a person, faction, group, organization, or party. On inspection, it is usually true that the chief power-holder has to delegate some power to organize and maintain the monopoly. But to delegate power is not to share it. The principal who delegates requires the delegated agent to perform specified tasks and may withdraw the mandate.

Monopolies of power exist, at least formally, in tyrannies, despotisms, military autocracies, monarchies, lordships, papacies, theocracies, and one-party dictatorships. They also exist, however, in democracies, a more unsettling idea. To say that democracy may coexist with monopolistic domination requires no commitment to theories suggesting that behind the façade of electoral competition lies the power of a ruling class or a power elite (see, e.g., Miliband 1980 [1969]; Domhoff 1990; Mills 1956). For example, no matter how competitive or free elections may be, critical political power can be monopolized between elections by the incumbent president, prime minister, cabinet, and nominated judges associated with the dominant party, ethos, or ideology. Even a temporary domination (between elections) is nevertheless domination, and the opportunities for elected leaders to dominate their societies against widespread or deeply held public preferences are significant (see, e.g., Nordlinger 1981, 92-94, 111-12, 130-32).

That democracy might lead to domination was the theme of the "tyranny of the majority," which deeply concerned eighteenth-century republicans, such as James Madison, and nineteenth-century liberals, such as Alexis de Tocqueville and John Stuart Mill. They were mostly preoccupied, however, with the impact of that possible tyranny on the individual's property and liberty (including the individual's religious beliefs) rather than on national, ethnic, or linguistic minorities as such (Madison, Hamilton, and Jay 1987 [1788]; Mill 1997, 5-6, 81-82, 192-94; de Tocqueville 1988 [1835, 1840], vol. 1., chap. 7, esp. 250ff).

Democracy is, however, also straightforwardly compatible with the (temporary) tyranny of a minority, especially democracies with institutions that encourage the "winner" to take all. For example, an ideological faction, not supported by a majority of voters in a country, may nevertheless control a cabinet, which in turn controls a party, and which in turn controls a legislature. In consequence, law or public policy may be dictated in the interests of the faction as long as its control is maintained.

Defining Power Sharing, Deeply Divided Places, and Well-Ordered States

These considerations suggest the following broad definition of power sharing: Any set of arrangements that prevent one agent, or organized collective agency, from being the "winner who holds all critical power," whether temporarily or permanently. This suggestion explains why the synonyms of power sharing usually include the following generally positive connotations: "coalition" or "cooperative" government and "consensual" and "inclusive" decision making. Critics of power sharing just as powerfully insist upon negative connotations. They refer to power-sharing arrangements as "rudderless" or "leaderless," and they complain of "stalemated," "deadlocked," or "blocked" decision making.

The general definition of power sharing just suggested is broad if not vague. It does not, for example, specify how power is shared among the parties. It is capacious enough to include arrangements such as the Roman Republic's executive, based on the annual election of two consuls, and its tribunes, who were able to veto legislation; ancient Sparta's two kings and ephorate; the mercantile republican aristocracy of Venice; and the Institutes promoted by Calvin in Geneva. These are examples of power-sharing arrangements, and the definition thereby displays a key advantage: it does not presume that all power-sharing arrangements are virtuous by our current standards. A definition is of considerable merit if it makes it possible to approve or disapprove of the use to which power-sharing systems are put.

In this book our authors' attention is mostly on contemporary power-sharing systems. The major exception is Benjamin Braude's discussion of limited power-sharing provisions under the Ottoman and Safavid empires (Chapter 6). In contemporary political science, to summarize a very large literature, power sharing is defined both by a regulatory goal and by specific instruments. The goal is the arrangement of political institutions to prevent the monopoly, permanent or temporary, of executive, legislative, judicial, bureaucratic, military, or cultural power. Four principal sets of instruments accomplish this goal.


  1. The first are overtly political bodies (executive, legislative, judicial, and administrative) organized to ensure both "shared rule" and "self-rule" among the relevant agents. These political bodies are organized through partly self-governing communities or territories, or both. Differently put, and as we shall elaborate below, these political bodies may be consociational (based on communities) or federal (based on territories). These bodies usually respect some combination of the principles of parity, proportionality, and autonomy.

  2. The second are security bodies: militaries, which normatively face outward for defense; police, which face inward to preserve order; and intelligence agencies, which engage in lawful surveillance and threat assessments. Security bodies must be organized so that power sharing within the political bodies is meaningful.

  3. The third are economic policies, principally wealth-sharing formulae, that reinforce the power sharing within the political bodies through some combination of the principles of parity, proportionality, and autonomy that also apply within the political bodies.

  4. The fourth are policies and practices that preserve cultural pluralism. Modern power sharing deliberately avoids the full-scale integration or coercive assimilation of "cultures" within the polity; that is part of their monopoly-rejecting ethos.


The last set of instruments may be called "cultural protectionism," both by their critics and their proponents. They distinguish modern power-sharing systems from the ancient, medieval, and early modern examples cited above: the Romans, the Spartans, the Venetians, and the Calvinists never intended to promote cross-cultural power sharing within their republics. In modern power-sharing systems, "cultures," and their evolution, are not left to the free market or subject to the governing diktats of the largest group. Instead (at least some key components of) the cultures of the parties to the power-sharing system are protected. The overall power-sharing settlement may promote an inclusive overarching shared public identity, but that identity must complement rather than wholly replace the previously existing "cultures."

"Cultures" is put in quotation marks because agreeing what is entailed in "culture" is highly contested. As employed here culture encompasses the languages, national and ethnic traditions, religions and philosophies of life, customs, mores, and the ethos of peoples. As used here, no supposition is made that particular cultures are homogeneous, intrinsically holistic, static, wholly authentic and unaltered transmissions from antiquity, or mutually exclusive. It is, however, usually true that in deeply divided places at least some key agents believe that their "cultures," in whole or in part, are threatened by others.

Contemporary power-sharing systems promote the coexistence of at least some cultures. This idea is partly reflected in the language of "multiculturalism." But to anticipate some false and facile criticisms, modern power sharing does not promote all cultures (e.g., headhunting, tribal scarring, genital mutilation of males or females, or foot-binding). Nor does it presume to freeze the cultures of the contending parties as they were when the power-sharing settlement was made. Rather, modern power sharing, through "encoded pluralism," enables the partners to the political settlement to have the power to govern changes in their own cultures—through autonomy (self-rule) and through joint agreement with others (shared rule). "Cultural protectionism" works not through the wholesale freezing of certain practices but through empowering specified groups to control their own cultural evolution—both autonomously and jointly.

"Deeply divided places" has more obvious connotations than power sharing, but warnings are in order. "Places" is a better expression than "societies" because it is a mistake to presume that a divided place contains just one society; that may be an issue in deep dispute, and a deeply divided place may be characterized by rival, parallel, or segregated societies. In a deeply divided place there may be more than one "civil society," and their relations may be far from civil. All moderately complex societies are divided (stratified) in ways that may matter politically, for example, by age cohorts, by sex or sexual preference, or by income, wealth, class, and status. But within deeply divided places these standard stratifications are superseded, or profoundly reinforced, by further divisions of nationality, ethnicity, race, tribe, language, or religion. Deeply divided places are, as the designation suggests, sites of actual or potential "civil" or intergovernmental wars. They are where genocide, ethnic expulsion, or coercive assimilation are threatened, or have taken place; they are the places for which power sharing is often recommended.

There needs, however, to be some prospect of "stateness" or "governability" for power sharing to work as a recipe for deeply divided places. States matter more than societies in building inclusive power sharing because they define societies and their possibilities. Impersonalized institutions that have some degree of centralized and procedurally governed political decision making characterize functioning states. They have coercive capacities to ensure security: they can regulate all instruments of potential public violence and prevent or inhibit their own agents from being predators. They express authentic legal authority over persons, property, and their movements, and are recognized as such entities by their citizens, civil society organizations, and other states. Through self-help or alliances they can defend themselves. Lastly, functioning states are defined by their recognized sovereignty over their territory and its accompanying prerogatives: control over entry and exit of persons and entities. If states lack these capabilities they cannot protect human rights, promote human development, be inclusive, or share power effectively. Credible power sharing requires credible commitments to governability.

Conversely, failing and failed states are personalized: previously dominated by rulers, a family, clan, or clique, which did not distinguish public from private realms and have become "kleptocracies," governments of thieves, before or during the collapses of their regimes. They lack coherent, institutionalized, rule-governed patterns that inhibit predation. The "rulers" are indeed predators. They have usually lost their monopoly on the regulation of coercion and are challenged by guerillas, paramilitaries, terrorists, Mafiosi; they may be invaded, looted, and occupied by other states. They neither make nor enforce law. Those over whom they have failed to rule despise them as much as they fear them. These properties of failed states remind us that inclusion and power sharing work best within well-ordered states. Power sharing, inclusivity, and human development require more than the diffusion of the right values; they need the soil of functioning states because they are unlikely to grow in "anarchia." Order, or its likely realization, is therefore a key condition of power sharing in a deeply divided place, a negotiated order that brings widespread human security.

Thomas Hobbes was right to emphasize the necessity of order for a worthwhile human life but deeply wrong to mandate everywhere an authoritarian solution or a sole sovereign. Unlike Hobbes, we know that our states are far more lethal than the alleged "war of all against all," and for that reason alone we need to prevent them from becoming Leviathans. Power sharing is one route to controlling the violence of states, not just the violence of civil wars. States are the most powerful agencies of exclusion, and governments the major murderers in human history; and many state-builders and nation-builders have been people-killers and nation-killers (Connor 1972). Rudolf Rummel has calculated that in the twentieth century governments killed nearly 170 million people within their borders, a figure that exceeds those killed in wars between states (Rummel 1997). Genocide—killing peoples because of their presumed ascriptive characteristics—has been more common than most countries' official histories acknowledge, and governments have been the major perpetrators: the list is not confined to Ottoman Turkey, Nazi Germany, and Interahamwe Rwanda. "Politicide"—killing those deemed political opponents—has also been recurrent in modernity. The Soviet Union, especially under Stalin, was the major killing regime of the last century: nearly sixty-two million may have perished under its yoke. Maoist China was often as brutal, killing over thirty-five million people. "Democide"—the killing of peoples—is the ultimate form of exclusion, and governments its major agency. Governments have also organized, encouraged, or not stopped the expulsion of whole categories of persons from their land borders or their shores—people whom they have helped define as undesirable, nonindigenous, nonnational, or disloyal. The twentieth century is justly described not just as one of death by government but as one of expulsions, of the "cleansing" of populations. Governments have also been the prime architects of policies of discriminatory control: organizing dominant national, ethnic, linguistic, or religious groups, and disorganizing and subordinating others through systematic discrimination, what we are now encouraged to call "exclusion."

These should be commonplace thoughts. Regrettably, they are not sufficiently appreciated. States define people's life chances, and it is to their practices that we must look to see how greater inclusion, promised by power sharing, may best be facilitated. Exclusion, in the regulation of national, ethnic, linguistic, and religious differences, is the product of coercive homogenization and special treatment of privileged communities: counting some categories of persons and communities "in," as members, and others "out." Genocide, expulsion, and the unilateral partition of territories are ways of homogenizing peoples that are now internationally outlawed, even if the laws are dishonored. If international norms against genocide and expulsion are rigorously enforced, through diplomatic engagements, economic sanctions, military embargoes and interventions, and the international criminalization of genocidal officials, then the worst forms of exclusion may eventually be halted or inhibited. But success in these endeavors requires moral universalism and a reorientation of the foreign policies of the great powers so that both their policymakers and domestic constituencies see the prevention and punishment of genocide and expulsion as in their interests. We are far from there.

The record of recent history is, however, mixed: it need not occasion despair. Reforms and sanctions against exclusionary practices—against discriminatory racist, religiously intolerant, and xenophobic regimes—had major successes in the last century. The undermining of apartheid in South Africa had both international and domestic sources, both moral and political causes. Decolonization, desegregation, autonomy, the advancement of language rights, and civil rights movements were significant elements in the more inclusionary advances of the twentieth century. Power sharing is one of the most important instances of enhanced "inclusion," and it was revitalized in the last century and in this one. It is a standard prescription for protracted national, ethnic, and communal conflicts in deeply divided places, especially ones focused on antagonistic self-determination claims, but it is not the sole one; and no sensible advocate of power sharing assumes it is a panacea. Power sharing in deeply divided places is the subject of this collection of essays. Our contributors address variations in power-sharing systems and the instruments used to calm deeply divided places to make them more peaceful, civil, and stable, and to provide credible commitments to well-ordered democratic states.

Constitutionalism: The Separation of Powers, Competing for Power, and Power Sharing

Democracy can operate as a tyranny of the majority, and as the tyranny of a monopolistic faction, and therefore may worsen sociability and civility among the contesting peoples in a deeply divided place (McGarry 2010b). One response to such fears is to suggest that well-designed "constitutionalism" will block such tyrannies through protecting some core rights of citizens from potential violations by the governing majority (or faction) or from the stifling of individuality by crushingly conformist public opinion. Constitutionalism also works through making officeholders publicly accountable, requiring them to follow rule-governed and transparent procedures in their performance of their functions and through dividing public authority among multiple offices and institutions. Is constitutionalism therefore the relevant way to control monopoly? Is constitutionalism the key to power sharing?

Power sharing is indeed normally constitutionalized, that is, encoded in formal constitutions, written agreements, treaties, or charters, but that does not mean that all constitutions are power-sharing systems, especially regarding the management of cultures. Constitutions in principle may assign term-limited authority to use power in a monopolistic fashion to a person, faction, party, or role, for example, "The President shall exercise the full plenitude of the executive power for a seven year term of office." The constitution may grant emergency powers to a single-person executive, which may include provision for the suspension of basic rights, and fundamental freedoms, or dramatic "war powers." A constitution may mandate that there is just one nation, one religion, one language, one ethnicity as the basis of eligibility for citizenship. A constitution may mandate that the territory of the state is indivisible, thereby blocking the formation of federal regions. A constitution may make nonamendable or powerfully entrench some identities at the expense of others. The 1982 Constitution of Turkey has many of these features.

But even when constitutions are more pluralistic they need not be power-sharing systems. If constitutional rules permit the domination of one person (faction, party, or national, religious, or ethnic group) over all the key institutions that shape major political decisions (and nondecisions), then the abuse of political power is possible. A powerful president or prime minister, backed by a regularly electorally endorsed party, may lawfully execute the powers of the office, promote legislation that reflects the preferences of the dominant nationality, race, religion, or linguistic group, and fill the judiciary and administration with their appointees. The "rule of law" is therefore often the rule of the dominant majority (or faction). Periodic elections may not prevent such domination. Accountability to some of the electorate, or to the courts, may not protect visible minorities, especially if the laws (and the constitution) have been drafted according to majority preferences. Courts staffed by appointees of an executive and legislature controlled by the same party may not act as guardians of individuals' rights, let alone collective minorities. Constitutionalism per se does not prevent cultural homogeneity from becoming the hallmarks of policy and the state. Constitutionalism, formal or informal, is probably a necessary condition for successful power sharing across cultures, but it is definitely not sufficient. Indeed modern power sharing is conceived of by some of its supporters as a necessary supplement to constitutionalism, required to prevent the monopoly of cultural power, and especially required by large minorities.

Another way to make much the same argument distinguishes three remedies for despotic power found primarily in liberal thinking, namely the division of power, competition for power, and power sharing. All these three ways of preventing despotic power may be constitutionalized, although it is the separation or division of powers that is historically associated with "constitutionalism."

The Separation of Powers

In the liberal political tradition, influenced by the arguments of Montesquieu and James Madison, and strongly present in American federalism, the separation of powers holds a famous place. Here dividing political power is seen as critical to preventing despotism. The tradition commends separating executive, legislative, and judicial institutions. Inhibiting a monopoly of power, especially in the executive, avoids kingly dictatorship. The separation of civilian from military power, of nomination from appointment, of police powers to arrest and interrogate from the judicial power to prosecute, of federal governments from state governments, or local governments from central governments are less recognized but just as important parts of the same logic. To divide power is to prevent its abuse; to check power with power controls public officials. Ambition tempers ambition.

Some think that properly organizing the division of power to prevent domination is what really matters in deeply divided places (see Roeder and Rothchild 2005; Roeder 2007). A well-structured division of power, they say, inhibits national, ethnic, or communal majorities—or minorities—from dominating others. Instead of one majority, multiple majorities, or no majority, may be created. By placing power in different institutions, with different electoral procedures, and with staggered timings of elections, for filling key officials, different persons and different majorities (or pluralities) in different institutions may claim democratic mandates and thereby check one another. Proponents of integration and assimilation often make this claim. They celebrate the creation of dual executives (a president, and a prime minister and cabinet), a dual legislature (a house of representatives and a house of states or regions, or an elected house and a house of experienced experts), and autonomous legal institutions (perhaps with both a supreme court and a constitutional court). For such advocates, separating powers is power sharing as well as power-division because they think that the separation of powers in different institutions obliges power-holders to work cooperatively in anticipation of the checking and balancing capacities of the others.

One tempting response to this apparently persuasive rhetoric might be to observe that the well-attested division of powers in the U.S. Constitution proved compatible with institutionalized and racialized slavery between 1787 and 1860; racism and a white tyranny over blacks in the deep South from the 1880s until the 1950s; and the termination, expulsion, and demographic erosion of Native Americans in an unfinished process of destruction. The U.S. Constitution has also proven wholly compatible with regular eruptions of persecutions of immigrant and religious minorities. Far from always supporting pluribus, the Philadelphia Constitution has enabled the unum of the dominant people palpably to take charge and avoided neither the tyranny of the majority nor the tyranny of faction. While that rebuttal is in order, it would be too quick a response because it does not address the relevant difficulty: it is a feature of our concepts that the expressions "power sharing" and the "division of powers" can be construed as synonyms by the informed, as well as the less well informed. In constitutional advisory work in Iraq, for the Kurdistan Region between 2003 and 2009, and in Sudan for the United Nations in 2009-10, I experienced the difficulty in the flesh and in texts. "Power sharing" in English would be translated into Arabic and then later translated back into English by another person to confirm that all negotiators and their advisors agreed the meaning of the relevant text. More often than not when translated back, "power sharing" would emerge as "the separation of powers."

There are philological and etymological similarities between sharing and division that explain this conflation: through dividing, after all, people or groups receive shares. But power sharing and the separation of powers are distinct, albeit potentially intersecting, notions. Power sharing mandates both coordinated jointness in shared decision making and autonomy in group or territorial decision making. The separation of powers, by contrast, does not prescribe a coordinated policymaking system. Rather under the separation of powers, policy and order are expected to emerge (if at all) from the clash of ambitious power-holders scattered across multiple institutions, and the separation is not organized to facilitate group organization. Power-sharers seek to share power across national, ethnic, religious, and linguistic groups through their representatives making joint decisions in executives, legislatures, and judiciaries; power-dividers seek to break up the formation of such groups and to individualize what they disparage as communal politics. As we shall see, power-dividers are mostly integrationists—only centripetalists, whom we shall shortly describe, express the desire to compete as advocates of power sharing. But though they say that they support sharing power, centripetalists rarely wish to institutionalize autonomy among encompassing national, ethnic, linguistic, or religious communities, or to have their representatives jointly share power within an executive; what they want is power sharing among politicians incentivized to be moderate toward others who are different from them. That said, there can be a power-sharing coalition within a federal or central government that respects a formal separation of powers among executive, legislative, and judicial institutions, and between the federal/central government and the regional/local governments, and related variations.

The Competition for Power

In the liberal tradition, found also in its "constitutionalist" wing, there is a distinct focus on how public officials "win" powerful positions, be they executive, legislative, judicial, or bureaucratic. Competition, as in the marketplace, is seen as a way of preventing nefarious monopoly. In this tradition, competition for executive and legislative posts should occur through elections. Liberals, especially outside the United States, are more doubtful about elections to judicial and administrative positions, for which they generally favor competitive meritocratic appointment through professional associations and transparent and reviewable procedures. The minimal definition of representative government is a political system in which officials compete for authoritative positions in free and fair elections for citizens' votes; in which elected officials hold office for limited terms, make laws, and give orders to unelected officials within constitutional norms that ensure accountability—both through the ballot box and recourse to the courts. Given these premises, the competition for power is a sine qua non of democratic government.

The division of powers and competition for power are established and intelligent principles. But on their own, advocates of power sharing submit, they are unlikely to calm deeply divided places and may cause conflict. Where the competition for power resembles an ethnic, religious, or linguistic census, elections may not check governments; rather they may encourage the tyranny of the majority. The combination of the separation of powers and the competition for power may also be conducive toward the oppression of national, ethnic, and religious communities. After all, the competition for power expresses or creates majorities—and such majorities may be constructed from national, ethnic, or communal cleavages. A sustained majority from the same community may win control over all major offices and governments—even if the powers of those offices and governments are divided and checked—and then propose and implement discriminatory public policy or biased conceptions of merit.

Integrationists and assimilationists often prescribe the division of powers and the merits of the competition for power but often presume that a nation of individuals is in existence, or that one should be built. They may forget (or ignore) that many states are multiethnic and multiconfessional—and that many are plurinational—and thereby pass over the fact that the competition for power (with or without the division of powers) may be a recipe for conflict in deeply divided places. Sri Lanka has not wanted for a separation of powers or for elections. Neither has Kenya nor Northern Ireland. To commend procedures that advance the position of nationalizing majorities when there are rival national self-determination claims is partisan or, alternatively, utopian. It is partisan when one community seeks to nationalize the state or region in its image on no better claim than might (numbers) makes right; it is utopian when (potentially or actually) antagonistic communities are expected or instructed to fuse. Partisans and utopians have had opportunities many times in the last two centuries, too often producing bloodbaths. That is why many contemporary liberals commend power sharing as a supplementary approach to avoiding the worst outcomes in plurinational, multiethnic, and multireligious states.

Power sharing, however, commends not only the sharing of power but also the division of power and the competition for power. Power sharing should add to rather than subtract from the liberal, constitutional, and democratic experience. Power sharing commends "coalition" as a considered way of doing things, but not as a wholesale substitute for the division of power or competition for power.

If Rudolf Rummel's calculations in Death by Government are even approximately correct, the last century was the most lethal in human history. Power sharing claims to offer some prospect of reducing domestic lethality and war in human affairs. The argument for power sharing is more sophisticated than acknowledging that what cannot be won on the battlefield is best allocated through a shared forum and a shared executive. Power-sharers follow Rousseau's declared method in The Social Contract that commends taking "men as they are, and laws as they might be," but because they do not seek just one community they reject Rousseau's particular proposals as disastrous, namely inalienable, indivisible, and absolute sovereignty, the rejection of partial associations, and one vigorous homogenizing civic religion. Power-sharers do not seek a social contract among a unified people; they seek social contracts for sociability among divided communities or between territorial governments. The first of these possibilities leads to what are called "consociational" directions; the second leads toward territorial or federative power sharing. These two possibilities can be combined in complex forms, as we shall see.