Chapter Four—
Models and Pitfalls

When I said earlier that, for a democratic, accommodative regime to emerge, it is necessary to will the means as well as the ends, I meant to imply that what looks like a surprising degree of interracial consensus on goals in South Africa can be defeated by dissensus on specific measures that might actually achieve those goals. Small mistakes have large consequences in this field. It is useful to remember the remark of the Confucian Hsün-tzu: "Yang Chu, weeping at the crossroad, said, 'Isn't it here that you take a half-step wrong and wake up 1,000 miles away?'"

Right at this point, there is an important difference between engineering, the design of physical structures, and constitutional engineering, the design of political structures. In engineering, as in constitutional engineering, there are multiple potential sources of failure. Consequently, engineers create redundancy (backup systems) and factors of safety (overbuilding) to counter failures and mid-course correction features to reverse them when they occur.^[1] In constitutional engineering, mid-course correction is extremely difficult, because interests quickly crystallize around whatever arrangements are adopted; and so, even if the institutions fail in their public objectives, there are actors whose private success depends on the maintenance of the arrangements. South Africa cannot count on mid-course correction. This puts the burden of

averting failure on precision at the outset, on safety factors, and on redundancy, but in a divided society it is fortunate if there is initial consensus on reasonably apt institutions. Consensus on precisely apt institutions, let alone overbuilt or redundantly apt institutions, will be very hard to reach.

All of this counsels against the advice frequently heard in divided societies—that if it is wrong at the outset, it can be fixed later^[2] —and argues strongly for efforts to enhance initial agreement on institutional objectives, rather than on constitutional provisions that the parties can merely tolerate. A deliberate process can invest the participants in the product of their own hands, instead of being merely a necessary prelude to the contest for power.

These cautions make clear why Yang Chu's apprehension frequently applies in constitution making across the board—in careless borrowing of institutions, in citing the wrong examples and drawing the wrong lessons, in choosing one mode of decision making rather than another, and in choosing one set of institutions over another. All of these are problems in South Africa.

A particularly helpful way to see this is to review some of the major plans that have been considered for South Africa. Some are much more elaborate than others. They run the gamut, from doing very little except expand the franchise, to structuring extraordinary safeguards, to dividing up the state. As I intend to show, virtually all of the main models that have been advanced are inapt for the conflict, and most are also inapt for the metaconflict. More than this, they tend to overlook some very important distinctions relating both to the formulation of plans for accommodation and to the content of those plans. Failure to attend to these distinctions renders constitutional plans vulnerable to some predictable pitfalls. These I shall present later in the chapter.

Binationalism

The prevalence of racial discrimination makes it easy to conceive of South African society as characterized by what I called in Chapter 1 ascriptive ranking and subordination. The tendency to view the society that way is powerfully reinforced by the whole history of European colonialism in Africa and by some of its important by-products, such as

slavery and segregation in the Western Hemisphere. There is, of course, no doubt that the Western, and particularly North American, focus on South African issues is at least as heavily determined by the experience of race relations in North America as it is by South African conditions and events. There is a strong element of projection of home-country guilt onto South Africa, which accounts for much of the importance of the South African issue in American (or British or French) politics.

This element of projection makes it difficult to conceive of the South African conflict in terms other than those of racial subordination. There have been numerous attempts, for example, to explain the history of Afrikaner nationalism to Western publics,^[3] but they have largely fallen on deaf ears. More often than not, Afrikaners are portrayed as unreconstructed racists, throwbacks to the bad old days of Jim Crow and White supremacy.^[4] The internalization of this same image—and the guilt consequent upon it—is largely responsible for the steady growth of Afrikaner liberalism in the post-World War II period.

There is, however, an alternative way to conceptualize the conflict, one that has long come naturally to Afrikaners themselves. An amalgam of Dutch, French, and German strands, the Afrikaners, even in the eighteenth century, saw themselves abused by and cut off from the Netherlands. "By 1795 the European roots of the white South African community were almost completely severed. Most of its members were South Africans of at least the third generation. Very few of them had even visited Europe, or were in communication with anyone in Europe. They were white Africans."^[5] Later, of course, they considered themselves oppressed by British colonialists. All along, they fought battles with neighboring Black ethnic groups. Afrikaner victories and defeats burned themselves into a sharply defined ethnic consciousness.^[6] Slowly, in the twentieth century, there has also been some further merger of the Afrikaner and the English-speaking populations, including increasing rates

of exogamy. In politics, for example, the National Party draws support almost equally from both White groups.

Conceived in this way, the current struggle can easily be regarded as involving Afrikaner nationalism or, more broadly, the nationalism of both White groups, on the one hand, and Black nationalism, on the other. Two important consequences follow from such a conception of what is happening in South Africa. The first is that the aspirations of the Whites become far more legitimate. Now they are not—or at least not merely—racist oppressors seeking to preserve privilege but nationalists struggling not to be swallowed up in a larger, competing nationalism. The second is that, if this characterization is accurate, political formulae might well take account of it. Needless to say, racial discrimination would not be countenanced, but nationalists, White and Black, may be entitled to a degree of self-determination.

If there are, for the sake of argument, just two nationalisms—one White, one Black—in South Africa, then perhaps South Africa does not resemble Mississippi in 1950 so much as it resembles Canada in 1980. In short, if there is a White nationalism and a Black nationalism, then the appropriate accommodation might be one that builds political institutions, as the Canadians have, on recognition of two "founding nations" that must live together in a single state but can only do so decently if the design of the state prominently reflects what is called "dualism."^[7]

To be sure, the outcome in Canada is by no means certain. The "patriation" of the Canadian Constitution in 1982 established something like a loose federation, but this Quebec saw as insufficient. In 1987, the Meech Lake Accord attempted to secure the accession of Quebec by recognizing that "Quebec constitutes within Canada a distinct society."^[8] The provinces remain more or less equal, but a province would have been permitted to declare a provision of the Charter of Rights inoperative, and Meech Lake would have allowed a province to opt out of certain federal programs. It also would have expanded the category of constitutional changes requiring unanimous approval of the provinces and required the courts to interpret the Constitution to take the binational character of the country into account. In 1990, amidst considerable Anglophone bitterness, the Meech Lake Accord lapsed with-

out securing the required assent of all the provinces. Nevertheless, slowly and haltingly, Quebec has been pushing Canada toward a confederation that recognizes the state as having, not just an Anglophone majority and a Francophone minority, but "two majorities," each of which will have some sovereign powers. The tortuous path toward any such recognition is a sign of how great the obstacles to binationalism might be in South Africa.

Nevertheless, a somewhat analogous plan for South Africa has been advanced for a ten-year transitional period by Hermann Giliomee and Lawrence Schlemmer. Judging the struggle to be "primarily between Afrikaner and African nationalists,"^[9] they argue that the most sensible change for the time being is to a binational state, probably with three electoral rolls—White, Black, and open^[10] —and a dual set of symbols for the state, to allay the fears of Whites that political change will necessarily entail a "symbolic reversal."^[11] At the center, they envision a joint Black-White government of national unity based on the requirement of concurrent majorities for action.

All of this they see as transitional. As confidence is built, what they call "nation-building" can take place. The long-term future, in their view, will be based on majority rule, following a dissolution of the binational system, a dissolution that will take place by mutual consent.

Of course, South Africa would not be alone if it adopted such arrangements. Not only Canada but also Belgium has elements of a binational polity; other Western European countries, of the consociational variety, have arrangements for decision making by consent of all groups. Czechoslovakia has dual symbols, including two national anthems, one Czech and one Slovak. In short, there are precedents for these proposals, although generally the binational solution is based on nations that have separate territorial homelands. The overarching Black and White groups occupy the same single homeland.

In this respect, there is an interesting distinction between traditional and contemporary conceptions of political community. Whereas kinship alone might frequently have been sufficient to constitute traditional polities, modern political systems are based heavily on inclusive conceptions of territorial proximity. No doubt there have been departures from

this notion, but the idea of territorial sovereignty, which is now worldwide, is difficult to square with a full-blown division of sovereign authority based on origins rather than on territory. Skepticism about such approaches is epitomized in a scathing characterization of the government's emphasis on group autonomy and "own affairs": "it permits you to carry your own border around with you."

Apart from what I shall soon say about group rights and consociationalism—some of which might apply to the Giliomee-Schlemmer proposals if they were spelled out in more detail—there are two main additional objections that could be raised. The first relates to the temporary character of the arrangements. If there really are contending Black and White nationalisms, do nationalisms disappear in a decade? The second relates to the identification of the nationalisms in question. Leave aside the incomplete merger of Afrikaners and English into a White group. What of the merger of Zulu, Sotho, Tswana, and Xhosa into a Black group? If the polity is to be divided along sub-South African national lines, then Zulu nationalism, as represented by Inkatha and KwaZulu, might have a claim to recognition, but such claims are denied at the threshold by the binational character of the scheme. It is not clear that, once binationalism is to be recognized, some Black ethnic groups will not make strong demands for multinationalism.

Again, Canada comes to mind. The effort, in the 1960s, to foster bilingualism and biculturalism ended by promoting bilingualism and multiculturalism, as various non-British, non-French groups questioned the legitimacy of biculturalism.^[12] The same might well happen in an initially binational South Africa.

The temporary character of the plan appears explicable in terms of its proponents' commitment to a nonracial society and to untrammeled majority rule, even as they recognize alternative ways to view the polity. Or, to put it more sharply, South African elites find divisions of the future polity so thoroughly illegitimate—reacting as they are against the past divisions of the polity—that proposals like these are difficult to justify, to oneself and to others, except as transitional. There are, in short, ideological crosscurrents in South Africa, some based on previous experience, some based on assessments of the present and future, and they sit most uneasily with each other.

The same point can be made about the boundaries of Black nation-

alism. To recognize Zulu or Ndebele or Sotho nationalism might be to confirm the South African government's prior recognition of these nationalisms and its embodiment of that recognition in the form of the so-called homelands. To recognize nationalism below the level of an inclusive Black nationalism is to run afoul of an important South African taboo.

Once constitutional planners think in terms of contending nationalisms in South Africa, they will inevitably be faced with major obstacles of precisely the sort that confront Giliomee and Schlemmer. It remains true, however, that, over the short and medium term, no constitutional progress can be made without White assent. Perhaps Whites will be willing to assent to a scheme like this, and to no other. Of course, we do not know that this will be the case, since no proposed schemes have been put to the test. Nevertheless, if that does prove to be the inclination of Whites, Blacks will have to think long and hard before rejecting a scheme of this sort. South Africa, it needs to be emphasized, is not assured of either an easy transition or a democratic result. Any arrangement that can make both more likely has to be considered seriously.

There is, however, a good chance that Black leaders will reject such a scheme out of hand—so strong is the power of certain, relatively fixed ideas in South Africa. These ideas form part of the metaconflict described in Chapter 1. They were expressed by Walter F. Sisulu, the long-imprisoned ANC leader, in February 1990, when he noted that President F. W. de Klerk "still speaks of Black representation on an ethnic basis: Black people must elect Black people. The ANC will not even think of taking part in an election for Black representatives—it is a step back to ethnicity."^[13] But suppose for the sake of argument that Blacks do not reject the plan for a binational state at the outset. Will the proposed arrangements make the transition and a democratic result more likely, if they are adopted?

There are grounds for a negative answer. If the plan is accepted by Blacks because, given White apprehensions, it is the only way Whites can be induced to agree to any substantial changes, that is scarcely a basis for optimism about the durability of the plan. There is a good chance then that the plan will be overthrown by resentful Black leaders at the earliest opportunity, and there is no assurance that what is put in

its place will be at all apt for South Africa's divided society. Even if the plan is not overthrown, it merely postpones the debate over the choice of institutions that are ultimately going to be appropriate for South Africa, including, for example, the kind of electoral arrangements discussed at length earlier. In the meantime, South Africans will become habituated to completely different arrangements, building up a new set of inhibitions and taboos. And if the Giliomee-Schlemmer plan is adopted in order to make constitutional progress but is then overthrown, the postponed institutional debate may never take place.

Here, then, may be a concrete case of some caveats I shall enter in more detail later. Any agreement is not necessarily better than no agreement, and the existence of an agreement is no evidence whatever of its probable durability.

Partition

Needless to say, if binationalism is ideologically unacceptable to a good many influential South Africans, partition of the country into more than one state is anathema. The sources of this reaction are several: the common antipathy of politicians to the prospect of losing territory in which power can be exercised; the conception of partition as another form of divide and rule, comparable to cutting the Black homelands out of South Africa and declaring their inhabitants to be noncitizens; and suspicion that a partition plan will give the predominantly White successor state the major share of good land and natural endowment, particularly mineral resources and rainfall—a suspicion solidly grounded in experience with the boundaries of the crowded, resource-poor homelands.

The comparative experience with partition in the post—World War II period has generally been unhappy, not in the sense that the alternatives to partition were necessarily superior but in the sense that partition was costly in lives and in protracted, post-partition conflict. The examples of India-Pakistan, Israel and the rest of Palestine, Cyprus after 1974, and Ireland after 1968 (although it was partitioned almost a half-century earlier) are all pertinent here.^[14] Partition may, in the end, be unavoidable, but for a great many reasons it is generally not the policy of choice.^[15]

Still, South Africa is a society so divided that fallback positions can hardly be excluded a priori. Possible partition plans have been assessed in South Africa for a very long time.^[16] Among the various plans, the most carefully considered, from the standpoint of history, economic integration, demography, and—especially—fairness in the allocation of resources, is the one described (but not advocated) by Gavin Maasdorp.^[17] Maasdorp's plan is particularly generous, in terms of resources, to the part of the bisected state that would have an African majority. But to examine briefly even this careful design is to see the ethnic problems that would be left afterward.

The successor state with the larger White percentage—"Capeland"—would be centered on the Western Cape, extending eastward to, roughly, the Fish River (east of Port Elizabeth), and would also include two districts in the southwestern area of the Orange Free State. On the basis of the 1970 census, Africans would constitute only 23.6 percent of the population. The remainder of South Africa—"Capricornia"—would have a strong African majority. Nearly all Indians would be in Capricornia, and nearly all Coloureds would be in Capeland. More Whites would be in Capricornia than in Capeland, but since Capeland would have less than 20 percent of the total South African population, Whites would form a larger share of the Capeland population.^[18]

Despite the effort to disentangle populations, Whites would still form a minority of only 27 percent in Capeland (and of 15 percent in Capricornia). Coloureds would make up nearly half the Capeland population, and Africans would be (even as of 1970) almost as numerous as Whites there. Neither state would be free of the same sort of conflict that now troubles South Africa.

No doubt it is true that there was once a possibility of solidifying the links between Whites and Coloureds in the Cape, but those links have been weakened and, in a good many cases, broken since the apartheid policy of the National Party disfranchised and forcibly displaced many Coloureds from their residences in the 1950s. Segments of the Coloured community remain quite conservative. As I noted in Chapter 2, in a 1985 survey in Natal, 31 percent of Coloureds chose President P. W.

Botha as their preferred South African leader, nearly three times the number that chose Nelson Mandela.^[19] But a great many Coloureds have identified with the struggle of Africans against the White regime. In Capeland, there is little doubt that, in place of intergroup harmony, there would be considerable political conflict between Coloureds and Whites. Moreover, the end of influx control and pass laws in the mid-1980s has produced enormous growth in the African population around Cape Town, so that it is no longer a foregone conclusion that Whites would even constitute the second largest population group in Capeland. Capeland would be an especially severely divided, triethnic society.

About 80 percent of the population in Capricornia would be African. The White minority there would presumably have relatively little political power. Capricornia would, after all, be designed to be the African-dominated successor state of South Africa. If Whites had grievances, the presumption would be that they might move to the "non-African" successor state, Capeland, where, as we have just seen, their position might also be precarious. And to the extent that White power was diminished in Capricornia, conflicts among Black groups could be expected to emerge there, uninhibited by the comfort they would otherwise have given to powerful Whites. In fact, in Capricornia, as in Capeland, a few large groups would compete for power. Ironically, this would alter one of undivided South Africa's only favorable conflict conditions—the existence of a multiplicity of ethnic groups, no one of which could, under democratic conditions, easily capture power by itself.^[20]

It seems obvious that partition along these lines would not solve the problems of intergroup conflict—though it would rearrange the conflict somewhat—or the problems of Whites, or the problems of anyone else. Any partition that stood a chance of solving such problems would require a massive population transfer, principally of Whites to the Cape, which might not be able to support them.^[21] Even if a transfer could be arranged, its impact would be to turn the Coloureds, almost a majority under Maasdorp's plan, into a minority under the domination of Whites, who have not exactly been solicitous of their interests in the past. Partition cannot be ruled out, but it is not a promising idea.

The Zimbabwe Model:
Majority Rule, Minority Overrepresentation

The so-called lessons of Zimbabwe are much discussed in South Africa. Whites in Zimbabwe feared for their future under majority rule. As is well known, no major calamity has befallen them. On the contrary, they prosper as farmers and business people under a regime that understands their contribution to the gross national product. In fact, within a few years after independence, some former Rhodesian diehards were returning to the country from South Africa. In a moment, I shall have something to say about what Zimbabwe teaches with respect to the controversial question of group rights. At this point, three other lessons can be derived from Zimbabwe.

The first is that, whatever the economic virtues of political quiescence, Zimbabwe is essentially a one-party state and not a democratic one. Political parties in Zimbabwe broke fairly neatly along ethnic lines, and under first-past-the-post elections the Shona majority could easily outvote the Ndebele minority. Repression was the dominant mode of nation building in Matabeleland until the capitulation of Joshua Nkomo and the institution of a de facto single-party system in 1987.^[22] So the first lesson of Zimbabwe is to confirm the tendency, to be described more fully in Chapter 5, toward census-type elections and majority domination in divided societies unless appropriate precautions are taken. This is indeed a lesson of considerable relevance to South Africa.

The second lesson relates to the indigenousness question I have touched on previously. Dutch settlement in South Africa began at roughly the same time as Dutch settlement in New York, in the seventeenth century. The survivors of a wrecked Dutch ship began a sojourn near what is now Cape Town in 1647, the same year that Peter Stuyvesant arrived at Fort Amsterdam in what is now Manhattan. By contrast, the majority of Rhodesian Whites actually arrived in Zimbabwe after World War II.^[23] Quite apart from the tiny fraction of the Zimbabwe population consisting of Whites, it seems appropriate to use the term settler for

such recent immigrants. One can readily understand, then, why Whites in Zimbabwe abandoned the aspiration to a real share of political power in order to preserve untouched their economic power. One could hardly envision a similar transaction in South Africa, where Whites aspire to retain a share of political power and Blacks aspire to attain a share of economic power. The second lesson of Zimbabwe is, then, that the Whites got it backward—perhaps not for Zimbabwe but for South Africa.

The third lesson of Zimbabwe takes us back to a theme discussed earlier: timing and planning. Following its Unilateral Declaration of Independence (UDI) in 1965, the Ian Smith regime took every opportunity to delay reaching a settlement. Mordechai Tamarkin has shown that, from 1974 onward, Smith had the possibility of negotiating power-sharing arrangements that would not only have been favorable to Whites but, more important, were supported by President Kenneth Kaunda of Zambia, by the West, and by the Soviet Union.^[24] The settlements that were offered were certainly durable, at least over the short and medium term. Whether the institutional arrangements would have been conducive to a democratic, multiracial, multiethnic Zimbabwe is difficult to say in retrospect. What is clear is that the bad faith of the UDI regime precluded any such possibilities, accelerated guerrilla activity, and ultimately forced an international settlement that disadvantaged not so much the Whites as the Ndebele, who suffered violence in the countryside for years after independence in 1980. As is customary in such matters, when the time is propitious, the urgency seems absent. When the urgency is present, the time may no longer be propitious.

What was ultimately provided in the 1980 Constitution of Zimbabwe was temporary White overrepresentation in the legislature.^[25] For a period of seven years, 20 House of Assembly members were to be elected by Whites on a "white roll," while the remaining 80 House of Assembly members were to be elected on a common roll consisting of other voters.^[26] Less than 2 percent of the population thus held 20 percent of the parliamentary seats. At the earliest lawful moment, in 1987, the Zimbabwe government abolished these reserved seats and joined White voters to the common roll.^[27] The 20 seats were converted to seats to be held by nominated members of any race.

To be sure, the government then nominated Whites to most of the

converted seats.^[28] Nevertheless, the experience with reserved White seats shows that guaranteed disproportionate representation was seen to be reversible, that it created a desire to reverse it (despite the need for White capital), and that it did not create any real power that the White community did not in any event enjoy.

No doubt, as has been pointed out, South African Whites would not be content with the Zimbabwe arrangements.^[29] But the lines along which much thinking has gone come down to minority guarantees, safeguards, and special rights.^[30]

From what I have already said, it is not difficult to see the deficiencies of most versions of group rights—or, really, special group privileges. They provide illusory security, easily pierced. Even if they continue to function, they consign minorities to minority status. Unless they offer a minority veto—in which case the urge to abolish them will grow—they ratify the exclusion of the minority from power. So, in the first respect, group rights provide too much—benefits that are disproportionate and are, on that account, unlikely to survive. And, in the second respect, group rights provide too little, for they do not aim at minority participation at the seat of power. The overrepresentation of Whites in the 1980 Constitution of Zimbabwe had both of these vices.

In Zimbabwe, as in a number of other countries, minority overrepresentation was anchored to nothing. No doubt, overrepresentation served as a symbolic assurance by the incoming regime that it bore no special ill will toward Whites, despite the past. It was a transitional, confidence-building measure, not a provision that can help produce democracy in divided societies.

If one is looking for African democracy in a divided society, the place to look is not Zimbabwe but Nigeria. In 1978–79, despite severe ethnic conflict, military coups, civil war, and a long period of military rule, the Nigerians proceeded to design a sound regime for managing their het-

erogeneity in a democratic fashion. When that experiment was interrupted by a military coup on the last day of 1983, the Nigerians still did not give up. In the late 1980s and early 1990s, they began the process of returning once again to democratic rule, and once again they have designed institutions to foster interethnic accommodation. That is where many of the African lessons are, but they seem far away, little known, and less understood in South Africa.

Consociation

What divided societies like South Africa are likely to get is the worst of both worlds. Majorities may ultimately get majority rule, but without genuine inducements to accommodate minorities. Minorities may demand and receive guarantees of protection that, in the end, will protect nothing worth protecting.

Consider, for example, the admittedly rough sketch of a future South Africa drawn up by Heribert Adam and Kogila A. Moodley.^[31] Adam and Moodley argue that "factors unique to South Africa suggest that the dream of a relatively democratic society has a better chance of being realized in an integrated South Africa than elsewhere."^[32] These unique factors are said to be (1) minority rather than majority claims to exclusion;^[33] (2) the noncorrespondence of racial divisions with cultural or linguistic divisions; and (3) the economic interdependence of all groups.

It should be said in passing that none of these conditions is unique to South Africa. Excluded groups often make claims to inclusive principles, and dominant groups reject them—whatever the numerical status of the groups. In many severely divided countries, religion and language span group boundaries. And the economic interdependence of South Africans is hardly different from the interdependence that characterized many societies later shattered by ethnic violence, among them Lebanon, Cyprus, Uganda, and Sri Lanka. If South Africa is to be saved, it will hardly be saved by suggesting that it may not even be a thoroughly plural society. It will be saved by overcoming the obstacles in the way of a democratic future.

At some level, Adam and Moodley must realize this, for they advance the outlines of a scheme predicated on overcoming difficulties. What, then, is their plan? Their preferred outcome would be consociational democracy—a form of collective decision making that is said to mitigate unbridled majoritarianism in divided societies by requiring the concurrence of all groups in all major decisions and the delegation of decisions involving only the interests of particular groups to those groups themselves. It is, its most persuasive proponents say, the best, even the only, democratic system for severely divided societies.^[34]

But, Adam and Moodley suggest,^[35] it is too late for consociationalism in South Africa. Consociational systems in Switzerland and elsewhere in Western Europe antedate industrialization. Industrialization in South Africa has made "a more democratic means of representation of interests" necessary.^[36] Consequently, their scheme envisions universal suffrage, but under a system that can guarantee "the influence of minority political parties . . . through proportional representation—or even overrepresentation—as well as veto rights."^[37] As a matter of fact, they even contemplate "corporate federalism"^[38] —that is, nonterritorial, group-based federalism that might even be represented in the upper house of the central legislature—a good measure of cultural autonomy, even group self-policing through local police forces, group proportional representation in the armed forces, perhaps even overrepresentation for minorities in the military.^[39] A society with allegedly excellent raw prospects turns out in the end to require extraordinary precautions.

There is no need to dwell on the disjunction between Adam and Moodley's benign diagnosis and the strong medicine they prescribe. South Africa is a severely divided society; it needs a prescription.

If one person, one vote is required—and it is—then special rights for particular groups, rights designed to detract from the effects of universal suffrage, will not work. The contradiction between the two principles will be visible, and the former will win out. If the universal suffrage

that is contemplated seems conducive to majority oppression, ask not what can be done to counter universal suffrage but whether the right electoral system has been adopted.

Although Adam and Moodley deny that their prescription is consociational, the medicine has a strong aroma of consociation about it. The most elaborate consociational scheme for South Africa consists of several elements that add up to what its author, Arend Lijphart, calls "power-sharing."^[40] (Just to keep the terminology clear, I do not mean consociation in the former South African government's sense of officially defined groups, enforced segregation, and cooptation of unrepresentative elites, but in Lijphart's sense.) These elements include a broadly inclusive coalition; an executive that represents all the significant groups; "internal autonomy for groups that wish to have it," embodied in federalism if the groups are geographically separate and in the nonterritorial, group equivalent of federalism if they are intermixed; "minority veto on the most vital issues"; a system of election based on proportional representation; and proportional allocation of civil service positions and public funds to the significant groups.^[41] Since Adam and Moodley also espouse proportional representation, group vetoes, nonterritorial federalism, cultural autonomy for groups, and group proportionality in the armed forces, if not in the entire civil service, how their scheme differs from consociationalism is difficult to say.

I have enumerated elsewhere several reasons to doubt the applicability of consociational models to severely divided societies in Asia and Africa.^[42] Here I only want to underscore a few points of immediate relevance to South Africa.

Perhaps the most important is that the consociational model contains no mechanism. The consociational idea has been aptly described as government based on a "cartel of elites."^[43] But there is no reason to think that politicians—especially electorally minded politicians—will behave in accordance with its rules and form such a cartel across ascriptive lines. There are good reasons to think that they will not.

In divided societies, there are some studies indicating that elites are less ethnocentric than their followers, but there are more showing that ethnocentrism increases with education.^[44] There is little or nothing in the available South African studies to suggest a different conclusion. There is direct evidence from several studies that increased education is generally associated with negative attitudes on the part of Black South Africans toward out-groups. Research conducted in KwaMashu, near Durban, found that the most highly educated male respondents displayed the greatest interethnic and interracial social distance.^[45] A study of African interracial attitudes concluded that "a higher educational qualification was significantly associated with a negative attitude toward Afrikaans-speaking Whites. . . ."^[46] A survey found African students at universities in the Western Cape to be racially exclusive "hard nationalists."^[47] In a fourth study, based on an earlier Soweto sample, increasing parental education was associated with slightly reduced social distance between Black respondents and most other racial groups.^[48] The

results were the same for social distance between the Zulu subsample of 75 respondents and most other racial groups; but the modestly increased levels of parental education reflected in the subsample produced an average increase in social distance between Zulu respondents and other African groups, a substantial one in the case of distance from the Xhosa.^[49] Among Whites, although increased education might be thought to produce greater interracial tolerance, the South African results are at best equivocal.^[50] Even if tolerant leaders could "bring the tendentiously hostile masses to accept coexistence," notes Theodor Hanf, " . . . this begs the question of how hostile and intolerant masses are going to produce such tolerant and open-minded leaders."^[51]

Since the mute equation of educated elites with accommodative attitudes is unsupported, there is no reason to think automatically that elites will use their leadership position to reduce rather than pursue conflict. Analysts need to avoid projecting their own accommodationist impulses onto political leaders, who may entertain other ideas and whose behavior, in any event, will be governed not just by what they believe but by the situation they face.

If anything is to induce politicians to play by conflict-limiting rules, it lies in the structures in which they find themselves. By far the most important of these is the electoral system. I shall have much more to say later about the version of proportional representation (PR) that has been advanced as part of a consociational model for South Africa. Suffice it to indicate at this point that proportional representation makes a multiparty system, and therefore government by coalition, more likely than the present South African electoral system would, but it provides no

assurance whatever that all the significant groups will be embraced in the coalition. Like other electoral systems, PR is conducive to government and opposition. Those parties that, individually or in coalition, have a majority of seats proceed to form the government, while those that have a minority of seats form the opposition. That is, after all, the way PR systems function in the Irish Republic, in Germany, in Scandinavia, in Australia, and even in Israel, whenever one of the major parties and its allies have a clear majority of seats. To obtain a broadly inclusive consociational coalition, one that routinely embraces all the major groups and tendencies, requires agreement to much more than a PR electoral system.

The same goes for group autonomy, consensus decision making, minority veto, and group proportionality in the bureaucracy. These features amount to a full-blown scheme of government that does indeed require agreement among a cartel of elites. But notably lacking in this prescription is any reason, anchored in electoral politics, for leaders to support such a system, to make such an agreement, or more generally to foster intergroup compromise. Even if leaders committed themselves to such a plan at the outset, under democratic, competitive conditions, centrifugal forces from among their own followers and from among their more extreme electoral competitors would easily undermine the durability of the agreement.

There have been other versions of consociational proposals for South Africa, perhaps most notably those of Frederik van Zyl Slabbert and David Welsh, who advocate a system based on the "principle of coalescent politics," which seeks to be "maximally representative [of parties and of groups] while operating on a broadly consensual basis."^[52] Like Lijphart, Slabbert and Welsh propose a power-sharing executive, inclusive of all the major parties and groups, proportional representation elections, a minority veto, and group-based proportionality in the civil service, armed forces, and police. Unlike Lijphart, they do not advocate group autonomy on either a territorial or nonterritorial basis, and they hope that parties will have reason to appeal for support across group lines.^[53] They, too, however, are counting on enlightened leadership to take steps to avoid "mutual destruction."^[54] Of course, if such

leadership could be counted upon consistently, consociational arrangements would be much less necessary in the first place.

In many ways, proposals like these entrench group rights, although Lijphart is careful to insure that the only groups whose rights will be recognized are those that emerge voluntarily in the electoral process. In this respect, his proposals are quite different from the practices of the apartheid government, which specified the identity of groups from above and assigned individuals to official group categories. Nevertheless, there is no gainsaying the ultimate group character of the protections. Cultural autonomy, for example, cannot be accorded to political parties but only to ethnic or racial groups.

The same problem attaches even more prominently to consociational plans or devices that are explicitly based on group identity. The proposals of the KwaZulu Natal Indaba (discussed in Chapter 5) and of the Soviet academic Gleb Starushenko^[55] envision an upper chamber of the legislature based on racially demarcated seats, in equal numbers for each racial group. They would not meet the criterion I articulated in Chapter 1 of nonracial institutions suitable for a multiracial society.

This becomes a serious sticking point in South African opinion. As we have seen,^[56] there is considerable support for arrangements that preclude majority or minority domination. However, this support turns soft when the arrangements contain safeguards for named groups. A survey question that elicited 53 percent White approval for special White voting privileges received only 1 percent Black approval.^[57] A study conducted for the KwaZulu Natal Indaba asked whether "power-sharing between all races in South Africa" was the "best solution." Ninety percent of African respondents agreed. When, however, the question was framed in terms of "minority protections," only 20 percent agreed on the need for them, 38 percent disagreed, and 42 percent responded "Don't know."^[58] Given the dissensus on means, arrangements based ultimately on group rights have a doubtful future. They run afoul of the metaconflict. They also run afoul of strongly held conceptions of territorial sov-

ereignity, which, around the world, have eclipsed earlier ideas of sovereignty or autonomy on a group or personal basis.^[59] Perhaps unfortunately, proximity has displaced kinship as the fount of government, thus throwing the idea of self-determination into great confusion and undermining group rights.

Entrenched minority group vetoes would seem particularly precarious. It was pointed out by an unofficial South African study commission in 1973 that, for consociational plans to survive, the segments represented must be fairly equal in population.^[60] It is one thing to have each of three or four substantial groups enjoy a mutual veto; it is another to confide such power to two or three racial minorities which together constitute less than a quarter of the population.

There is experience with these matters. Minority guarantees were an integral part of the consociational Cyprus Constitution of 1960. It required concurrent majorities for legislation; a division of the two top offices (one Greek, one Turk), with mutual veto powers; and quotas that generally overrepresented the Turkish minority in everything from the cabinet to the civil service, the armed forces, and the police.^[61] Formal guarantees are virtually always resented by the group on which they are imposed in the same measure as they are insisted upon by the group seeking to benefit from them. Because they are resented, they are not likely to prove durable, unless, of course, they accidentally create an accommodative feature from which politicians can gain and which outweighs the resentment. That did not happen in Cyprus, and the arrangements were soon demolished.

An important source of durability for constitutional arrangements in severely divided societies is that they must work well and be seen to work well from the beginning. This is most likely if they create interests across group lines. Then there will be inhibitions on altering the arrangements. Group rights for the minority, however, are likely to appear temporary, as a thorn to be excised as soon as possible, as they were in Zimbabwe. But even if they were permanent, special rights would do nothing about the problem to which they are ostensibly addressed—the ability of ascriptive majorities to ignore interests other than their

own. If that is the problem, institutional designers will have to do better than reinvent wheels on which Zimbabweans were riding in 1980 and Cypriots were riding as early as 1960.

Bottom-Up

All of the plans reviewed thus far presuppose agreement at the center. By contrast, there has been a recurrent tendency in South Africa to suggest that, as long as progress was stymied at the center, local-level negotiations might create "nodes of interracial co-operation and accommodation"^[62] from which political structures might grow, in spite of intransigence at the center. While top-down plans were merely being discussed, bottom-up action might actually be taken.

This idea was first mooted in 1973 by the Political Commission of the Study Project on Christianity in Apartheid Society, which urged the vigorous use of the governmental structures of apartheid against the aims of those structures, so as to produce "power centres capable of forcing change throughout South Africa at a greater pace."^[63] The aim was not to succumb to immobilism but to use separate development to expand opportunities, improve Black social welfare, increase security of land tenure, reduce restrictions on free speech, phase out petty apartheid, encourage trade unionism and collective bargaining, and place segregated local authorities in ongoing relations with each other—in short, to adjust the existing system in order to reduce drastically the power of a stubborn central government, even before moving into a second phase, consisting of consultation and negotiation.^[64] A variety of plans for multiracial, bottom-up cooperation were advanced in the 1970s and 1980s.^[65]

Such an incremental strategy was of limited utility as time passed, for the same reason as gulps, rather than sips, became essential to the process of political change. Moreover, boring from within was not really possible so long as many major Black organizations declined to partic-

ipate in what they saw as utterly illegitimate governmental structures. A suggestion that the multiracial Regional Services Councils might, for example, provide such "nodes" of cooperation^[66] was met with the obstacle that the councils were based in turn on racially based local authorities.

The possibilities, however, seemed greater with the mass activity of the United Democratic Front in the mid-1980s. At the local level, there were frequent consumer boycotts of White businesses, sometimes leading to negotiations with White merchants and to resolution of local grievances. In a number of cases, the negotiations were more wide-ranging. In the course of the negotiations, some considerable local understanding appears to have been built up on both sides. It is exemplified by a statement issued during a 1985 consumer boycott by a Black civic organization in Port Alfred, in the Eastern Cape:

We emphasize our desire to make Port Alfred a pleasant place to live, for its own sake, and as an example to the rest of the country—after all, Jesus Christ wasn't born in the biggest city of Israel. This committee will always act responsibly in its dealings with its "counterpart" in the white section of our community. It is our greatest wish that this community will, in the very near future, be able to look back at these troubled times and see them as a good omen rather than the opposite. . . . The call of this committee is a simple one: We want a single non-racial local authority for Port Alfred. We are not blind nor wilfully deaf to the Government's new deal; we do not doubt the Government's good faith in seeking to resolve the political problems of our country. What we are doing is trying to contribute, in a positive way, to the search for solutions. We are well aware of the fact that our white "counterparts" have no direct powers to control what the Government does. However, we appeal to the white community as a whole to support this initiative, if only for the future of this particular area and, who knows, for South Africa as a whole.^[67]

This statement had followed a number of productive meetings, and it illustrates the good will that was still present at the local level, even as conflict was reaching new heights at the national level.

Such local-level efforts at political change had various origins. In East London, as in Port Alfred, there was a request by a township residents'

association for a takeover of the Black township administration by the White local authority, which proved receptive. In Pietermaritzburg, there was a planning exercise designed to reduce inefficiencies resulting from racially fragmented local governments. In Cape Town, the White city council began a dialogue with Black and Coloured civic groups and sought ways to restore the equivalent of the former nonracial franchise. In the main, therefore, the negotiations either followed a Black boycott or a White attempt to find a formula to transcend racially based local authorities.

These efforts failed for several reasons.^[68] In some cases, bureaucratic rivalries among governmental units came into play. In others, the central government thwarted the proposals, which ran afoul of its own master plan for racially separate local authorities as part of the devolution of power over "own affairs" to the respective groups. In still others, the UDF declined to participate, or the initiatives were overtaken by local violence. And in some localities, Black civic activists were arrested as the negotiations were in progress.

By and large, it proved impossible to do what was intended—to use existing structures for purposes opposite to those the regime had when it created the structures. The bottom-up relations that were established were not unimportant, because they created the chance for some people to work together across group lines for the first time.^[69] In a new period of change at the center, these understandings can be useful building blocks. For the most part, however, fundamental political change cannot come out of a bottom-up strategy.

The Models:
Too Little or too Much

The models reviewed thus far can be grouped into categories. Some plans presuppose consent on the part of participants; others elide the issue of consent. The bottom-uppers and the partitionists assume that those who participate in their ventures agree on the need to proceed in the ways they propose. There is no point in even discussing a nonconsensual partition in South Africa. In both cases, however, even assuming mutual consent, the likely accomplishments of the plans fall far short

of what would be required to achieve democracy in an environment of reduced conflict. In fact, drastic though partition would be, it might actually make the conflicts in the successor states more stark and serious than in an undivided South Africa. The same cannot quite be said of binationalism and consociationalism. Given the consent of the participants, perhaps they would have some of the desired effects. The problem is that consent is most unlikely to be forthcoming. In the case of group rights or overrepresentation on the zimbabwe model, both defects are present. Neither consent nor effectiveness is likely. Bottom-up and partition plans accomplish too little; binationalism and consociationalism attempt too much; and overrepresentation achieves both too little and too much.

Ironically, the models that lay out plans for consensual rather than adversary politics do not explain how to obtain consent to a consensual plan. Slabbert and Welsh, for example, acknowledge that "African nationalism has not viewed the constitutional provision of guarantees for minority ethnic or racial groups with much favour."^[70] In South Africa, there is a strong ideological aversion to guarantees on the part of Charterists and Africanists. Slabbert and Welsh also refer to "the strongly competitive and zero-sum conceptions of politics in South Africa, and its corollary that South Africa's political system must embody either white minority rule (under the guise of separate development) or black majority rule," which means that "the adoption of a more coalescent style of politics would require a substantial modification of attitudes."^[71] In some countries, attitudes have been modified, but only after disaster has struck. Short of that, the modification of attitudes in the time required would seem highly unlikely.

Although they propose consociation despite this problem, Slabbert and Welsh also note that "the best possible safeguard for minority groups" is one "woven into the operation of the political system itself by making the votes of minority members important."^[72] In Chapter 5, I shall show that the electoral system they recommend will not accomplish this, but the insight remains useful. Weaving concern for the interests of others into the fabric of political calculations is completely different from prescribing altruism ex cathedra. Much of the remainder of this chapter and of this book in general emphasizes this distinction, which is vital to the survival of severely divided societies.

Contracts and Social Contracts

By and large, those Asian and African states that have done best at managing their divided societies in a democratic fashion have done so at one of two junctures: at independence or after a civil war. The European and North American states that have made similar accommodations (Belgium and Canada) tend to fall in between—they did not make appropriate plans at the outset for democratic conflict management, but they did not wait for civil war either. Since each of these times of innovation has a characteristic set of procedures, policies, and pitfalls attached to it, I want to deal with them separately.

First, in a minority of post-colonial societies, accommodations were reached at independence—not by the departing colonialists but by the local groups that would have to spend the future together. The most important cases are the Malaysian constitutional bargain of 1956 and the Lebanese National Pact of 1943, both of which saved conflict-prone societies from early disintegration.

Notice that they are "bargains," "pacts," "contracts." They are treaties between semisovereign peoples based on reciprocity, and they have all the characteristic problems all contracts have: the preferences of the parties change over time; conditions also change; the returns to the parties from the deal are uneven; and the deadlines laid down inevitably arrive. There is typically a good bit of splitting the difference when the contracts are made. Certain characteristic provisions, assuaging aspirations when adopted, create discontent later. If group quotas are used, changing demography may render them obsolete, but they are frozen into the agreement. If some issues are postponed to a date certain in the future, the firm target becomes a new date of reckoning, a new occasion for conflict. If incommensurables are traded—X in return for Y—and if X proves more valuable over the long term, the party that received Y may nurse a sense of grievance. Unless provision is made for amendment, contract alone is not a lasting basis for accommodation. Intergroup contracts tend to be their own undoing.^[73] The problem with intergroup pacts, in a word, is that to endure they need to be flexible; but to be formed they must be, in principle, permanent, since the ripe moment may never come again.

Contrast with these problems the completely different assumptions,

procedures, and provisions of the Nigerian return to democratic rule in 1979—and again a decade later.^[74] This is a second innovative juncture—after a civil war.

The Nigerians had been through severe conflict and civil war, and they did not want a repetition. Since no one could be sure which group might be on the receiving end in any future round of ethnic conflict and civil strife, the Nigerians made, not a bargain but a real constitution, not a contract among groups that knew what their interests would be but a social contract among groups that were not sure what their interests might be next time around. They made a blind, Rawlsian contract.^[75]

Mirabile dictu , the Nigerian social contract worked according to plan. The president, elected by a formula that put a premium on multiethnic support, became a panethnic figure. The new federal system made everything, including ethnicity, more complex. In ethnic conflict terms, the Second Republic was different from the First.

If everything went so well, why, then, was it necessary to go through the same exercise again a decade later? To answer this question, it is necessary to recall that the Nigerians did not have the sober insights of 1978–79 at the time of independence in 1960. In the interval, they developed corrupt regimes, military coups, a civil war, and bloated, greedy armed forces. All of these problems, the indirect result of the ethnic conflict Nigerians failed to anticipate and avert the first time around, came back to undo what the Nigerians did the second time around.^[76] As I said previously, earlier is better, in the sense that more obstacles are arrayed against democratic arrangements that are instituted later. Nevertheless, there is a tradeoff here: the bitter experience on which the later innovations are based may be costly and painful, but it is more likely to give rise to enduring institutions of conflict reduction.

That is because bitter experience is more likely to produce public policy making through the social contract mode than to produce treaty

making through the simple contract mode. By public policy, I mean arrangements that do not merely reflect transient group interests but a design for living together premised on incentives for accommodative behavior transcending group interests at the moment of enactment.

Group rights tend to be negotiated as part of the simple contract mode. They therefore are prone to many of the problems I have identified as being inherent in contract as a tool of accommodation.

South Africa is neither starting completely fresh, as countries on the threshold of independence were, nor sobered by civil war fought to victory. South Africa is neither in the position of Lebanon in 1943 or Malaysia in 1956 nor in the position of Nigeria in 1978. It is somewhere in-between. This middle position may not be an altogether bad thing. South Africa may be able to avoid the civil war needed to reach the Nigerian sobriety and public policy making that followed, but it may be sufficiently chastened by its current prospects to reject mere intergroup horse trading in favor of public policy making.

That is not to imply that group interests are irrelevant or that some intergroup trades are not part of public policy. In the making of the United States Constitution, for example, there was a large element of reciprocity, especially in the trades that were made between small states and large states. But some of the central innovations of the Constitution, such as a strong federation and a separately elected president, had little or nothing to do with anything other than what was thought to be sound public policy. South Africa is a severely divided society, which can become even more severely divided in the future. Groups will need self-interested reasons even to adopt public policy measures. The difference between treaty making and public policy making does not lie in the absence of altruism in the one and its presence in the other. The motive for both is self-interest. The point is to make groups see the long-term superiority of social contracts over simple contracts, of public policy making over mere treaty making. Unpleasant experience and a desire not to repeat it are at the root of the social contract-public policy perspective.

On the Fallacy "Whatever the Negotiators Agree on Will Be Acceptable"

What I have just been saying can be summarized in two more general statements. The first is that constitutive policy decisions really matter. The second is that mode of decision and content of decision are related.

Some provisions are more conducive to intergroup accommodation than others, and some procedures are more likely to produce these provisions.

Previously, I called attention to a contrast between Malaysia and Sri Lanka that illustrates the first of these statements very well. I said that the Malaysians had acted early on their ethnic problems, whereas the Sri Lankans had acted late. What the Malaysians did and the Sri Lankans did not do are every bit as important. Here I have in mind, not the Malaysian constitutional bargain, which later largely unraveled (for reasons to which I have already adverted), but the interethnic coalition that Malaysians formed even before independence. That coalition ran a single electoral slate, and it was flanked by ethnically based political parties. The coalition was the location for interethnic compromise and, in some modest measure, still is. By contrast, the Sri Lankans opted for ethnically based parties and Sinhalese-majority governments; and they implicitly opted against interethnic coalitions and interethnic compromise. These contrasting approaches are heavily responsible for the Malaysians' capacity to moderate a very difficult ethnic problem and the Sri Lankans' capacity to exacerbate a much less difficult ethnic problem.

To underscore the second proposition—that mode of decision and content of decision are related—I hardly need to say more than that, if we are in a deal-making mode, we shall probably make a deal, if we make anything at all. Deal making is common in divided societies—and not merely in ethnically divided societies but in societies divided along class lines as well.^[77] Lack of consensus, even active dissensus, among the component groups, together with spheres of complementarity or ethnic divisions of labor, can make the quid pro quo seem a natural solution. Trades also are attractive for political leaders in building support, for they can show constituents the palpable gains they have made right at the moment of agreement.

By the same token, serious public policy making for interethnic accommodation is rare, for some of the converse reasons and for some other reasons as well. If deal making is attractive because of immediate, palpable gains—because one immediately knows what one has received and has given up, at least for the time being—the very unforeseeability

of gains and losses is what makes public policy making less attractive, absent the Nigerian sobriety induced by bitter experience. Serious public policy making, embodied in constitutions, is also rare because constitutional documents have so often been sold to countries in need of them as so much parchment and so many lofty (albeit irrelevant) phrases required by every self-respecting state in the world community. They have been sold by provision merchants who, more often than not, have only a superficial acquaintance with the problems of the buyer countries. They are sold to politicians who have, in any event, their own agendas.

I do not mean to denigrate deal making in severely divided societies. On the contrary, an exchange, a treaty, a deal, is usually better than simply letting nature take its course in such societies, for that course is generally the course of serious, destructive conflict—or worse. All I want to emphasize is that the results that follow from the public-policy-making mode are different. For the reasons I have given and for some additional reasons that I shall identify later, they are also likely to be superior.

If I am correct about this, there is an important corollary for interethnic negotiation. There is now a large body of literature on negotiation, including interethnic negotiation, for such societies as Cyprus and Zimbabwe. That literature is almost entirely processual: it documents why and how parties reach, can be induced to reach, or fail to reach, agreement.^[78] The tacit assumption has been that, if the parties agree and can live with the agreement, they must have arrived at something like the market-clearing price. Perhaps they have, but the notion that, if an agreement clears the market, the resulting commodity will produce interethnic peace down the road is entirely specious. There are many designs for living together in a severely divided society. Evidence that the parties to a conflict have agreed on one such design only shows that this one was among those acceptable at the moment. It does not show that it was the best or that it will be durable.

There is a powerful bias toward process operating in world affairs. Parties to conflicts are supposed to talk to each other. If they talk earnestly and long enough, it is thought, they will understand each other's

concerns and solve their problems. The South African literature on negotiation is, unfortunately, suffused with the same notions. It emphasizes the "communication process"^[79] between adversaries, the "chemistry of negotiation"^[80] and its ability to produce unforeseen outcomes. "If the process is a reasonable one," claims a proponent of this view, "people will naturally be reasonable with each other."^[81] Such an approach begs the question of what is reasonable for a divided society, as it begs a host of other questions related to agreements, including the one alluded to in Chapter 1—namely, the problem of feigned assent to an agreement as a prelude to achieving hegemonic, rather than democratic, aspirations.

Incentives Versus Constraints

The more enduring and effective arrangements to reduce intergroup conflict are fortified by internal incentives, not external constraints. For present purposes, a constraint is a rule that binds only because it was agreed to at the outset. It has no continuing binding force based on present interest (except, of course, the interest in keeping one's word or in abiding by the law). It can readily be seen that most of the approaches reviewed in this chapter—group rights, consociationalism, binationalism—are based on constraints. An incentive, on the other hand, binds because it is in the continuing interest of the actors. The approaches just mentioned do not attend to the problem of incentives to accommodative behavior. If they create any such incentives, that is entirely accidental. The incentives of greatest concern are, of course, those that operate on politicians and their followers, that harness their self-interest to the cause of intergroup conflict reduction, regardless of their personal feelings, that, in a word, make moderation pay.

The exigencies of certain electoral systems do this, and certain political coalitions that result from particular competitive and electoral settings do the same. The genius of the Malaysian coalition arrangements that I referred to previously was that they forced Malay and Chinese politicians, in heterogeneous constituencies, to rely in part on votes delivered by politicians belonging to the other ethnic group. Those votes

would not be forthcoming unless leaders could portray the candidates as moderate on issues of concern to the group that was delivering its votes across ethnic lines. Consequently, compromises at the top of the coalition were supported by electoral incentives at the bottom. At various times, the Lebanese, Nigerian, and Sri Lankan electoral systems—or parts of them—have offered comparable electoral inducements.

The division of territory has sometimes had moderating effects on intergroup conflict. Before the "linguistic states" that were put into effect by Jawaharlal Nehru's government in India in the 1950s, intergroup conflict seemed to be growing. The advent of linguistically homogeneous states, however, impelled politicians seeking power in those states to compete with other politicians belonging to the same group, rather than different groups, and the effect was to reduce intergroup political rivalries and conflict.

All of this is what I mean by internal incentives built by constitutional engineers into the structure of the selfish calculations of politicians. Sometimes, however, the engineering is inadvertent. Like some other Indian nationalists, Nehru was initially opposed to linguistically homogeneous states, because he thought they constituted a capitulation to the forces of linguistic parochialism and a challenge to national unity.^[82] He did not foresee that linguistic states would tend to reduce separatism and intergroup tension, rather than foster them.

To frame the utility of incentives in terms of their ability to affect the selfish calculations of politicians, harnessing self-interest to the cause of peace, is, of course, to reject the perhaps-loftier aspiration of dealing with the so-called root causes of conflict. It is to resort to an approach that does not seek or need to change hearts and minds in order to succeed. That is not to say that significant periods of accommodative behavior will not soften up what seem now to be intractably difficult conflicts. Perhaps they will, but the efficacy of incentives does not depend on that; it depends instead of changing behavior. In this case, what needs changing is the behavior of the key actors who turn polities toward either conflict or accommodation: the politicians.

By contrast, consider the more usual and generally much less effective approaches, which rely on external constraints. Unlike incentives, the lure of which consists in their being rewarding, constraints are to be obeyed even if they are not rewarding. They are rules from outside.

It should not be thought that only minorities insist on constraints. One set of superficially attractive constraints consists of prohibitions relating to party composition. No party may be registered unless it provides lists of officers and members that demonstrate its multiethnic, multiracial character. Disturbed by the prospect of ethnic claims, a large number of African states outlawed ethnically based parties.^[83] In so doing, they managed to cast a pall over the political process but did not succeed in ending ethnic claims. Such prohibitions are futile where they are not draconian. Many parties are able to provide the appropriate spread of names on their lists of officers, members, and candidates without changing the character of the party at all. Formal legal regulations of this sort can produce a great deal of window dressing without altering the ethnic base of any party. As I shall indicate later, however, there are ways to induce ethnically based parties to alter, if not their membership base, their posture and their behavior on ethnic issues.

Despite their successful experience with electoral incentives, which I shall describe in Chapter 5, and their unsuccessful experience with an attempt to force parties to transcend ethnic lines, Nigeria's military rulers, planning a second return to civilian rule, have created a mandatory two-party system. Each of the parties is to be government created and ethnically balanced.^[84] Although they are not necessarily apt for democracy in plural societies, the two-party ideal and the nonethnic-party ideal are strongly held in much of Africa.^[85]

The ANC's Constitutional Guidelines similarly imply that advocacy of ethnic exclusivism might be unlawful in a nonracial, nonethnic South Africa.^[86] This would mean that ethnically based political parties would be unlawful. "During talks with an ANC delegation in Dakar in July 1988 it was made clear that racially or ethnically based mobilisation in

a hypothetical future dispensation under the ANC would be met with 'liberatory intolerance,' as it was put."^[87] Such prohibitions would affect some of the strongest likely participants, representing potential minorities, in a future democratic system. These might include the largely-Zulu Inkatha and the National Party.^[88] With such a step, so heavily reminiscent of the suppression of ethnically based opposition by Kwame Nkrumah, Sékou Touré, Milton Obote, and other post-independence African leaders, there is a strong sense of déjà vu. In no African case did such measures reduce ethnic conflict. In every one, however, such measures stifled democracy. Ethnically based parties are a serious problem, but outlawing them and depriving their constituencies of representation is not the answer. A much better answer, as I shall point out in Chapter 5, is to provide them with incentives to dilute the exclusivity of their appeals.

Constraints, then, consist of prohibitions as much as they do of guarantees. Neither necessarily has any payoff in inducements to accommodative behavior.

Occasionally, of course, constraints are part of a larger bargain. If the parties are unusually lucky at the outset, the resentment at what was given up is mitigated in the first instance by the knowledge that something else was gained. In the case of Cyprus, the Turkish guarantees, seen as Greek losses right from the beginning, were not offset by Greek gains. But even if they had been, the bargain would have been vulnerable to being undermined in the long term by all the forces that conspire against such contracts. These I have reviewed earlier in this chapter. In South Africa, there might well be a temptation to enact group political guarantees for Whites in exchange for an affirmative action program in education and employment for Blacks. Affirmative action may be desirable,^[89] but the asymmetrical character of the likely returns in this exchange counsels against making it part of a constitutional bargain. Group political rights may be conferred by a stroke of the pen, but the fruits of affirmative action may take years to appear. Differential rates of return, such as these, tend to undermine constitutional bargains. In Malaysia, precisely such a trade—of Chinese citizenship at once in return for a promise to help the Malays progress economi-

cally—tended to undermine the bargain a dozen years later, when Malays expressed discontent at what they had received.^[90] The lesson seems to me generalizable.

Another constraint-based approach consists, not of explicit group guarantees, but of ostensibly neutrally framed bills of rights and other legal requirements, often to be enforced by independent judiciaries. These rights and requirements may then provide shelter for groups seeking areas of autonomy or influence as groups.

There is nothing wrong with bills of rights or with judicial review of legislation. There is certainly an increasing tendency to regard both as useful, if not essential, to democracy. The Canadians have adopted a Charter of Rights and its concomitant, judicial review. In Australia and even in Britain, the seat of legislative supremacy, there have been serious discussions about the adoption of a bill of rights. Proposing a bill of rights, the report of the South African Law Commission on group and human rights is, in most ways, a remarkable and forward-looking document.^[91] The report endorses principles of universal suffrage and nondiscrimination and recognizes the need for wholesale law reform on these matters.^[92]

Although denominated a report on "group and human rights," the report of the commission puts the accent on human rights, rather than group rights. The sole exception is the provision of article 2 of its proposed Bill of Rights, which would permit positive discrimination in order to improve, "on a temporary basis," the position of groups that have been disadvantaged "for historical reasons."^[93] It does not require such affirmative action; it merely allows political bodies to resort to it. For the most part, the commission's work is markedly individualistic.

The problems arise, not from bills of rights or from judicial review per se, but from an effort to solve the problems of severely divided societies through the entrenchment of guarantees for particular groups and the commitment of enforcement to the independent judiciary. This, it seems to me, puts too heavy a weight on constitutional law, evolved

through the courts, rather than on constitutional practice, evolved through elected politicians; and it confers an unsustainable task upon a precariously placed judiciary.

The recent history of independent judiciaries in severely divided societies is, with some exceptions, not an especially happy one. The discord of the larger society may creep into their deliberations and undermine their independence. Even if it does not, self-interested politicians may find ways to achieve the same result, by reducing the jurisdiction of the courts or refusing to obey judicial decrees.^[94] When the overthrown prime minister Zulfikar Ali Bhutto, a Sindhi, was sentenced to death in Pakistan, his appeal was heard by the Supreme Court, which divided along ethnic lines. Sindhi judges voted in Bhutto's favor; Punjabi judges voted against him.^[95] When Turks walked out of the Cyprus parliament, Greek legislators passed a statute creating a Greek-dominated Supreme Court, contrary to the Cyprus Constitution. The Greek majority of the new court upheld the law on grounds of "necessity."^[96] When some Chinese in Malaysia went to court to force governmental permission to open a Chinese-language university, Malay judges were not sympathetic to their constitutional claim, whereas a Chinese judge was.^[97] And, finally, there is a well-known example in South Africa itself. In the mid-1950s, Cape Coloured voters were stricken from the electoral roll, in spite of an entrenched provision of the Constitution, by increasing the size of the Appellate Division that would hear a challenge to the law and by increasing the size of the Senate to insure the requisite two-thirds majority.^[98]

It is doubtful that a rights and requirements approach to divided

societies can survive the fragile state of judiciaries in those societies. All it is likely to do, when politicians are opposed to the claims, is to put undue pressure on a nascent institution that might otherwise evolve traditions of genuine and useful independence.

Given a choice between betting on the solid political insulation and ethnic neutrality of judiciaries in most divided societies or betting on the self-interest of politicians, prudent gamblers would put their resources on the latter. Give politicians a reason to be moderate, and they will find a way. But give them the Cyprus Constitution of 1960, with its heavy burdens and absolutely no political incentives for the majority to accept its minority protections, and the politicians will do just what Archbishop Makarios did—overthrow it.

Plans, Perfect and Imperfect

The Cyprus experience to which I have referred supports an insight of Pierre du Toit about divided societies. Based on experimental evidence with respect to bargaining, du Toit suggests that the more an outcome of bargaining between parties departs from the principle of equality between those parties, the harder it is for the more powerful party to justify such an outcome and for the less powerful party to accept such an outcome in the long term.^[99] This point is related to the earlier one that, for consociational plans to survive, roughly equal-sized segments are preferable to lopsided proportions. If du Toit is right, that does not mean that unequal outcomes will not emerge from bargaining processes, but it does imply that such provisions will not be durable. The Cyprus experience is consistent with both the adoption and the durability point.

It is not necessary to be a purist in these matters. For two reasons, severely divided societies can survive some measure of explicit Cyprus-type protection.

The first is that, without such protection, some agreements, some moves forward, will simply be precluded. If we take an absolute stand on this, we may not maximize interethnic accommodation as a value, because we may raise the threshold too high for some agreements and consign some groups in some states to a conflict-prone status quo. South Africa may be among those states.

The second, related reason is that, although the best-designed structures, filled with incentives to accommodation, may produce the best results, mixed structures may still produce mixed results. That is to say, if a given structure consists of components, some of which reduce conflict and some of which exacerbate it, there may still be a net reduction of conflict—not as great a reduction as might have been obtained had the parties to the conflict chosen a better design, but a reduction nonetheless. As I shall note later, the Nigerian Second Republic had some new, incentive-based features that reduced conflict, and it had some features that perpetuated conflict; the latter were copies of the conflict-fostering institutions of the First Republic. Still, the Second Republic did rather well overall. In terms of ethnic relations, it had mixed results, which were far better, in the aggregate, than were obtained in the First Republic.

The point, then, is not to be pure or entirely right about institutional design in the end. The point is to attempt to think clearly about the matter at the beginning, to urge apt institutions on the designers, and indeed to urge maximum incentives for accommodation in every body and at every level, but not to persuade the designers to reject anything other than the perfect plan. The perfect plan will not come along.

Far from it. Negotiating their way through a maze of conflict into an uncertain future, leaders of the respective groups have every reason to devise plans that do not point in a mixed direction but point in the wrong direction. So the real problem becomes, not one of accepting a less-than-perfect plan, but of being vulnerable to a plan that succumbs to all the pitfalls without possessing accommodative components. Major benefits accrue to polities that start out with accommodative institutions, rather than trying to build them later.

Four Reasons for Pessimism

Consider how many forces conspire against decisions at critical junctures that will operate to reduce intergroup tensions. The forces are at work in most divided societies, and they are at work in South Africa.

First, and very likely, it is already late. Bad things have happened, and good opportunities have been missed. That is assuredly the case in South Africa.

Second, decisions are likely to be taken in a crisis atmosphere, and too little attention will be paid to the distinctions I have been making—between incentives and constraints, as well as between contracts and

social contracts and between reaching any agreement and reaching a genuinely accommodative agreement. Certainly, in South Africa, these distinctions have not been a prominent part of political debate.

Third, since an agreement will be necessary, a deal may be sought, and it will be easy and natural for the parties to slide into the contractual form, with its many disadvantages. Again, in South Africa, the talk has fairly consistently referred to an undifferentiated "negotiated settlement."

Fourth, because the situation at the moment of settlement is likely to be grave, or at least serious, groups will fear to rely on internal incentives, although they are more effective. They will probably demand hard-and-fast guarantees—external constraints—that are far more frail and easy to undo. For years, the atmosphere has been rife with discussions of group rights in a future South Africa.

Because much is known about how to do better, crises in divided societies do not have to come out wrong. Thus far, most of them have. The question is whether South Africa will be in the one category or the other.