Jack Snyder and Karen Ballentine
Nationalism and the marketplace of ideas

The conventional wisdom among human rights activists holds that a great deal of the ethnic conflict in the world today is caused by propagandistic manipulations of public opinion. Human Rights Watch, for example, points the finger at unscrupulous governments who try to save their own skins by "playing the communal card." As antidotes, such groups prescribe democratization, wide-open debate in civil society, and greater freedom of the press.(1) Scholars likewise argue that a major stimulus to belligerent nationalism is the state's manipulation of mass media and mass education to infuse the nation with a sense of in-group patriotism and out-group rivalry.(2) They, too, prescribe greater freedom of speech.(3)

We agree that media manipulation often plays a central role in promoting nationalist and ethnic conflict, but we argue that promoting unconditional freedom of public debate in newly democratizing societies is, in many circumstances, likely to make the problem worse. Historically and today, from the French Revolution to Rwanda, sudden liberalizations of press freedom have been associated with bloody outbursts of popular nationalism. The most dangerous situation is precisely when the government's press monopoly begins to break down.(4) During incipient democratization, when civil society is burgeoning but democratic institutions are not fully entrenched, the state and other elites are forced to engage in public debate in order to compete for mass allies in the struggle for power.(5) Under those circumstances, governments and their opponents often have the motive and the opportunity to play the nationalist card.

When this occurs, unconditional freedom of speech is a dubious remedy. Just as economic competition produces socially beneficial results only in a well-institutionalized marketplace, where monopolies and false advertising are counteracted, so too increased debate in the political marketplace leads to better outcomes only when there are mechanisms to correct market imperfections.(6) Many newly democratizing states lack institutions to break up governmental and non-governmental information monopolies, to professionalize journalism, and to create common public forums where diverse ideas engage each other under conditions in which erroneous arguments will be challenged. In the absence of these institutions, an increase in the freedom of speech can create an opening for nationalist mythmakers to hijack public discourse.

In developing these arguments, we first define nationalist mythmaking and explain the scope of our claims. Second, we present our concept of the marketplace of ideas and offer hypotheses about conditions that facilitate nationalist mythmaking, illustrating these propositions with examples from that classic hotbed of nationalist mythmaking, the Weimar Republic, and from other recent and historical cases. Third, we test our argument against two hard cases, ethnic conflict in the former Yugoslavia and in Rwanda, both dramatic clashes that seem superficially to fit the conventional view that a government media monopoly is the problem, and unfettered speech the antidote. Fourth, we explore cases with comparatively moderate outcomes to determine the conditions under which democratization does not produce intense nationalist mythmaking. We conclude with suggestions for better institutionalizing public debate in new democracies.

Nationalist Mythmaking

Conventional wisdom is right in focusing on inflammatory propaganda as a cause of nationalism and ethnic conflicts. The archetype for this notion is the stem-winding oratory of Adolf Hitler, with its renowned pied-piper effect on malleable audiences.(7) Likewise, the world's first instance of aggressive nationalism, the Wars of the French Revolution, was sparked by an outpouring of warlike commentary in France's newly free press, which swept the demagogic journalist Jacques Pierre Brissot into power.(8) Recent reincarnations of this phenomenon, stressed in analyses by non-governmental organizations (NGOs) and by scholars, include the Hutu "hate radio" stations that encouraged genocide against Rwanda's Tutsi minority, as well as President Slobodan Milosevic's use of the television monopoly to foster an embattled, surly mentality among Serbs.(9)

Conventional wisdom is one-sided, however, when it blames nationalist demagogy primarily on governmental media monopolies and consequently prescribes unfettered free speech as the remedy. A 1995 report by Human Rights Watch, for example, concluded that in ten of the hottest contemporary ethnic conflicts, manipulative governments had "played the communal card" as a way to forestall declining popularity or to pursue strategies of divide-and-rule. "Dictatorship offers the ideal condition for playing the 'communal card'," because "official control of information makes public opinion highly manipulable."(10) Yet almost all of the countries studied in the report - India, Israel, South Africa, Romania, Sri Lanka, the former Yugoslavia, Lebanon, Russia, Armenia, and Azerbaijan - had recently held openly contested elections where powerful opposition groups were more nationalist than the government.(11) In addition, Human Rights Watch argued that, because "conditions for polarization along communal lines are less propitious in a society where public debate is encouraged," where past human rights abuses are vigorously prosecuted, and where there is "free participation in a broad range of voluntary and public associations," the cure is "vigorous civic debate" in a "well developed civil society."(12) But in fact, the Hindu fundamentalist Bharatiya Janata Party, the Armenian Karabakh Committee, and most of the instigators of ethnic conflict in the Human Rights Watch cases are "civil society," that is, voluntary organizations not created by the state. The Weimar Republic had record numbers of newspapers per capita, choral societies, and grassroots nationalist organizations, all indicators of a vigorous civil society; likewise, the nationalistic Jacobin Clubs of the French Revolution were spontaneous emanations of civil society.(13)

Conventional analyses fail to emphasize that a "well developed" civil society is not simply a matter of many clamoring voices, but also the set of institutions and social norms that make pluralism a civil process of persuasion and reconciling of differences.(14) No matter how well-intentioned and knowledgeable, non-governmental organizations promoting human rights tend to understate the tension between their ideal of an open society and the difficulty of establishing its preconditions in newly democratizing societies. As a consequence, their remedies may sometimes fuel nationalist mythmaking rather than dampen it. It is understandable that such groups are reluctant to explore the trade-offs entailed in promoting free speech, since there is undoubtedly a risk that qualifications to the argument for free speech can be exploited by dictators who wish to snuff out freedom of expression entirely. Nonetheless, because the risks of uncritical advocacy of unconditional free speech can be very high, these trade-offs must be analyzed forthrightly.

DEFINING NATIONALISM AND MYTHS

Nationalism, according to the most widely accepted definition, is the doctrine that the state and the nation should be congruent. Nationalism holds that legitimate rule is based on the sovereignty of a culturally or historically distinctive people in a polity that expresses and protects those distinctive characteristics.(15) We examine under this definition both the state-seeking activity of ethnic groups in multi-ethnic states and also the rivalries of established and more ethnically homogeneous nation-states, insofar as their leaders seek to legitimate their power and appeal for popular support for their policies by claiming to fulfill the historical mission or cultural identity of the nation.(16)

"Myths," for our purposes, are assertions that would lose credibility if their claim to a basis in fact or logic were exposed to rigorous, disinterested public evaluation.(17) The assertion that the Holocaust never happened is an example of a myth, in this sense. Nationalist mythmaking, then, is the attempt to use dubious arguments to mobilize support for nationalist doctrines or to discredit opponents. For some scholars, nationalist mythmaking is understood exclusively in this sense of promoting demonstrable falsehoods: in Ernest Renan's words, "getting its history wrong is part of being a nation."(18) Other scholars of nationalism, however, conceive of myth in the sense of mythomoteur, that is, as a story about the origins, special character, and destiny of the "nation."(19) The values embodied in such mythomoteurs are not subject to falsification, no matter how astutely they are scrutinized. Assertions like "it is good to be Ruritanian" or "Ruritanians deserve their own state" cannot be falsified, not because they are irrational, but because they are normative claims which exist independently of objective standards of argumentation. Nationalist mythomoteurs can, on occasion, lead to conflict: for example, when both Ruritanians and Megalomanians identify the same territory as "our historic homeland," and each claims to "deserve" exclusive, sovereign control over it. Because of the non-falsifiability of national mythomoteurs, even a well-constituted marketplace of ideas cannot provide iron-clad restraints against this kind of mythology. A well-constituted marketplace can, however, effectively mitigate the propagation of falsifiable nationalist myths. This capacity is of no small consequence since, typically, nationalist conflict is not the spontaneous emanation of vague mythomoteurs but the product of deliberate elite efforts to mobilize latent solidarities behind a particular political program, which falsifiable myths are used to justify. Thus, some of the justifications offered for why Ruritanians deserve to rule a territory (e.g., "we ruled it six centuries ago") or why they should preventively attack the Megalomanians (e.g., "they were able to kill ten million of us during the last war, because we let our guard down") entail empirical claims or causal inferences that are subject to objective scrutiny and perhaps verification or refutation.

We do not contend that all nationalist ideas are myths, let alone falsifiable ones. Nor do we argue that nationalists are uniquely prone to political mythmaking, nor that nationalism and nationalist conflict stem only from mythmaking. Besides myths, scholars have identified many factors that plausibly contribute to nationalism or nationalist conflict: the rise of the modern state, economic change, political repression, socio-economic inequality, security threats, and so forth. Our focus on mythmaking is not intended to compete with these approaches, but to complement them. We argue that these factors work their effects on nationalism through the process of persuasion and mythmaking in the marketplace of ideas. As many scholars have stressed, whatever background factors may contribute to nationalism, nationalist agitation and propaganda are always necessary conditions to the development of a mass nationalist movement.(20)

Though nationalism and nationalist myths are not the only cause of conflict between nations, we do argue that a tendency to breed conflict is inherent in typical nationalist myths, because they overemphasize the cultural and historical distinctiveness of the national group, exaggerate the threat posed to the nation by other groups, ignore the degree to which the nation's own actions provoked such threats, and play down the costs of seeking national goals through militant means.(21) Nationalist mobilization against alleged threats from other national groups, whether within the state or abroad, heightens the risk of conflict by stereotyping opponents as irremediably hostile, yet inferior and vulnerable to vigilant preventive attack.(22) Whether such myths can be successfully sold depends in large measure on the structure of the marketplace of ideas in which they are advanced.

Imperfect Markets and Nationalist Mythmaking

Liberal conventional wisdom, steeped in John Stuart Mill's argument that truth is most likely to emerge from no-holds-barred debate, optimistically expects the invisible hand of free competition to check the mythmaking of nationalist demagogues. Under conditions of "perfect competition" in the political marketplace, it may indeed be true that, on balance, unfettered debate tends to discredit ill-founded myths by revealing their factual inaccuracies, their logical contradictions, or the hidden costs of acting on their implications.(23) That is probably one reason why mature democracies never fight wars against each other.(24) However, when waning authoritarian power is newly challenged by the forces of mass politics, competition in the marketplace of ideas is likely to be highly imperfect, and opportunities for nationalist mythmaking abound.

In the political marketplace, governmental and non-governmental elites advance arguments about the benefits of policies and commit themselves to these policies in order to gain political support.(25) Consumers in the marketplace decide whom to support based in part on the persuasiveness of the arguments of elite entrepreneurs and on the credibility of the elites' commitments to implement desired policies. Middlemen in the marketplace (journalists and policy experts) and market institutions (the media, analytical institutions, and the laws regulating them) convey political entrepreneurs' commitments and arguments to consumers in ways that provide varying degrees of information about their credibility and accuracy. The better institutionalized the market, the better it scrutinizes arguments and forces ideas to confront each other in common forums, and therefore the better the information the market provides. The media that comprise the marketplace include not only instruments of mass communication like television and newspapers, but also local networks of face-to-face persuasion, as well as elite publications and discourse that generate ideas for mass dissemination.(26)

Thus, the commodities exchanged in the market are consumers' political support and suppliers' policy commitments. The role that ideas play in the "marketplace of ideas" is not that of goods, but that of advertisements for political support.(27) Ideas put forward in the marketplace convey purported information about values and interests, and also about facts and causal inferences from facts. Thus, the supplier uses advertising to convince the consumer to want what the supplier has to offer: for example, to believe that the fulfillment of the nation's destiny is rightful and valuable, and that the consumer's personal interests and values will be served by it. The supplier also tries to convince the consumer that the offered policies will produce the advertised benefits with a high probability at a low cost, and that alternative policies will lead to worse results. To accomplish this, the suppliers' advertising includes claims about, for example, the nature of the nation's opponents, the likelihood of cooperation with other national groups, the history of past interactions with them, and the prospects of success from nationalist mobilization or armed conflict.(28)

Our conception of the marketplace of ideas is based on the description of economic markets provided by standard economic analysis. The structure of the market consists of the degree of concentration of supply, the degree of segmentation of demand, and the strength of institutions regulating market interactions, including those that provide information or regulate advertising. Imperfect competition occurs when there are few sellers because of large scale economies and high barriers to entry, and when products are differentiated for sale to segmented markets. Market segmentation may occur as a result of consumers' distinctive tastes for differentiated products, the artificial inculcation of differentiated preferences through targeted advertising, transportation costs or other advantages in distribution and marketing to a particular set of consumers, or political barriers to exchange between market segments. Under these conditions, sellers tend to engage either in competitive advertising, collusion to divide up market share, or a combination of the two.(29) Rivalry is more likely when barriers to entry are falling, or in a "young industry," where "sellers may not have learned what to expect of rivals" and "may be scrambling to secure an established place in the industry, in the process inadvertently starting a price war."(30) To achieve socially beneficial outcomes under imperfect competition, regulation is needed to break up trusts, prohibit collusion, and insure truth in advertising.

The marketplace of ideas in newly democratizing states often mirrors that of a young, poorly regulated industry, where barriers to entry are falling, competition is imperfect, and oligopolistic elites exploit partial media monopolies in intense competition to win mass support in a segmented market. This kind of imperfectly competitive market may yield the worst of both worlds: elites are driven to compete for the mobilization of mass support, but by targeting niche markets, they can avoid debating in a common forum where ideas are publicly held up to rigorous scrutiny by competitors and expert evaluators. In these circumstances, nationalism may help elites to gain support in an ethnic market niche and also to maintain high barriers to entry by diverting demands for civic participation into mobilization for national goals. Thus, market conditions in newly democratizing states often create both the incentive for nationalist advertising and the conditions for its success, as we explain below.

PARTIAL MONOPOLIES OF SUPPLY

What human rights advocates fear most is a complete governmental monopoly over the press. In this situation, the government can propagate any nationalist myth without having to face countervailing arguments. While we agree that perfect monopoly is hardly desirable, we argue that it is not the only - and perhaps not the most - dangerous condition for nationalist mythmaking. Conditions of perfect monopoly make the audience skeptical. In communist and other authoritarian states, for example, people tend to discount propaganda precisely because they know that it comes from a monopolistic source, and typically turn to informal networks and stratagems for reading between the lines of official discourse.(31) Moreover, perfect monopolists often lack a motive to mobilize their population's nationalism. Facing no active opposition and ruling without popular consent, they face little need to compete for the mantle of popular legitimacy by whipping up mass enthusiasms. Indeed, unleashing mass nationalism would only hinder their goal of depoliticizing domestic politics and would introduce needless complications into their management of foreign relations.(32) For this reason, dictatorships play the nationalist card only under two conditions: when their ability to monopolize power and discourse is slipping, like that of the Argentine junta on the eve of the Falklands War, or when their ascent to power, and hence their legitimacy, have been based on the use of popular nationalism to prevail in an initially pluralistic setting, such as Bonaparte after the French Revolution, Hitler after the Weimar Republic, or the Japanese military regime after the collapse of the Taisho democracy of the 1920s.(33)

Especially prone to nationalist mythmaking are situations of partial monopoly over supply in the marketplace of ideas, which often occur during the earliest stages of democratization. In such conditions of intra-elite competition, governments and other elite interests often enjoy residual market power as the legacy of authoritarian monopoly control: the state or economic elites of the threatened ruling circles may still control key components of the mass media or have the resources to shape its content. Nationalist militaries may invoke their monopoly of specialized expertise to exaggerate foreign threats; the government may tendentiously regulate broadcast media in what it calls "the public interest"; private economic lobbies may buy journalists, supposedly neutral experts, and media access. For example, Alfred Hugenberg, the chairman of the board of directors of Krupp Steel during World War I and the chairman of the German National People's Party during the Weimar Republic, established the Telegraph Union wire service, which gave him control over half of Germany's press.(34) By providing loans, reduced-rate newsprint, and accounting services to inflation-ridden papers, Hugenberg achieved substantial control over many papers while maintaining their facades of independence. Though even small cities often had multiple newspapers, Hugenberg's service fed them all the same nationalist-slanted copy.

As a democratizing political system opens up, old elites and rising counter-elites must compete for the support of new entrants into the marketplace through popular appeals, including appeals to the purported common interests of elites and mass groups in pursuing nationalistic aims against out-groups. In many instances, including the case of Serbian President Slobodan Milosevic, these elites evince little interest in nationalism until rising pressure for mass political participation gives them an incentive to do so.(35) This strategy worked extremely well, for example, for the Kaiser-appointed governments of Bismarckian and Wilhelmine Germany, which faced the dilemma of winning budgetary approval from a Reichstag elected by universal suffrage. Five times between the founding of the Reich and 1914 the government chose to fight elections on what it styled as "national" issues - the Kulturkampf against the Catholics in 1874, the campaign of 1878 tarring socialists as anti-national, the campaigns to support bills to strengthen the army in 1887 and in 1893, and the "Hottentot election" on German colonial policy in 1907. Each time elections were fought on "national" grounds, voter turnout increased and more rightist candidates were elected, in part because conservative candidates got more votes overall and in part because coalitions of right-wing parties were more cohesive. Hidden financing of nationalist movements and publications by the Navy and by industrial interests, combined with prosecutions of opposition voices under a restrictive press law, played an essential part in this strategy.(36)

Rising counter-elites also tend to play the nationalist card; indeed, they often make the initial move in a spiral of nationalist outbidding. From the French revolution to contemporary Armenia, Azerbaijan, Georgia, and Croatia, it has often been elites from outside the ruling circles who pushed nationalist issues to the fore of the public debate, asserting a right to rule on the grounds that old elites were lax in pursuing the national or ethnic interest. In principle, populist counter-elites in conditions of incipient democratization can opt for any of several ideological stances vis-a-vis the old authoritarian regime. They can pursue a liberal strategy, criticizing the old elites' denial of individual civic rights; a socialist line, criticizing class domination; an ethnic line, criticizing the old elite for favoring a particular cultural group; or a more inclusive nationalist line, arguing that the narrow, venal old elite was ignoring the broader national interest. Nationalism is often attractive to rising counter-elites in part for the same reason that old ruling elites adopt it: unlike liberalism or socialism, nationalism allows the aspiring elite to make claims in the name of the masses without necessarily committing itself to a policy of sharing power and wealth with the masses once it has seized control of the state. Other incentives may depend more on context: e.g., which ideologies are discredited by association with the hated ancien regime, whether cultural differences are already salient and thus available to be politicized as ethnonationalism, and whether the rising counter-elite has some particular comparative advantage as the standard bearer for a distinctive national culture, as might be the case with literary or religious elites.(37) Also important, however, is whether the structure of demand in the marketplace of ideas is highly segmented, and thus permissive for nationalist mythmaking.

SEGMENTATION OF DEMAND

A well-constituted marketplace of ideas depends not only on the expression of diverse views by different groups in society, but also on individuals' exposure to diverse ideas. A highly segmented marketplace has the former, but not the latter. In a segmented marketplace of ideas, individuals in one market segment lack exposure to ideas expressed in other segments, or exposure is filtered through sources that distort those ideas.

Demand in the marketplace of ideas is likely to be segmented in newly democratizing states. A common sphere of democratic discourse depends on the development of unifying institutions, such as state-wide non-partisan media, which take time to construct. In many cases, the authoritarian states or colonial powers that were the democratizing states' predecessors leave a legacy of divisive institutions and ideas that were elements in a strategy of divide-and-rule. In other cases, democracy and freedom of expression are newly thrust upon traditional societies whose political horizons have historically been local and communal.(38) Even in democratizing post-communist states with a legacy of hypercentralization, the media's financial vulnerability often leaves it vulnerable to capture by partisan segments, thus spoiling it as a neutral forum for debate.(39)

Narrow market segments magnify the effects of oligopolistic control over supply; because they are more susceptible to domination by a single, myth-purveying supplier. Unlike true monopolists, oligopolists are forced to compete, but they often do so by increasing sales to consumers in segments of the market that they can monopolize, rather than in market segments where they face strong competitors. When this happens, there is no common marketplace of ideas, in which contending discourses and evidence confront each other directly on an even playing field. Instead, the existence of parallel monopoly discourses creates the illusion of market pluralism and free choice of ideas and, by vitiating skepticism, makes oligopolistic propaganda more effective than under pure monopoly.(40) Thus, the more segmented the market, the more the effects of partial monopolies of supply are magnified, and the more feasible is mythmaking.

Nationalist groups in the newly democratic Weimar Republic, including those backed by heavy industrial cartels, competed for mass electoral support against labor parties and liberals not so much by preaching to the constituencies of their opponents as by exploiting partial propaganda monopolies to mobilize their own. Hugenberg had only 50 percent of the overall Weimar media market, but he enjoyed a virtual monopoly over the flow of news to papers in Germany's small cities and towns, the locations that later voted most heavily for Hitler.(41) Exploiting Hugenberg's priming of middle class opinion, Hitler's successes came not from winning over liberal, socialist, or undecided opinion in open debate, but by cornering the nationalist market segment through skillful penetration of grassroots voluntary organizations, such as veterans groups and beer-drinking societies.(42) Since Hitler attained a dominant position in the Reichstag with only one-third of the vote, and used this as a platform for an unconstitutional seizure of the media and other state powers, monopolizing one segment of the market was enough to be decisive in a splintered polity.(43)

Segmented markets of ideas in democratizing states are conducive to nationalism for several reasons. The most fundamental reason is that elites have an incentive to promote nationalist populism as a substitute for true democratization. Segmented markets allow elite oligopolists to carve out a market niche where their nationalist ideas are not held up to systematic scrutiny. This is true even if the segmental divisions do not follow linguistic, ethnic, or communal lines. The nationalist market niches in Weimar Germany and in contemporary Russia, for example, reflect segmental cleavages within the majority ethnic group, such as those between urban and rural groups, between large cities and small towns, between soldiers or veterans and civilians, between economic strata, or between age groups. In these cases, mobilization of support by nationalists and by nationalistic Russian communists relied more heavily on face-to-face ward-heeling, handbills, and pamphlets targeted to specialized constituencies than on open media debate aimed at a broader range of society.(44)

Market segmentation in newly democratizing states sometimes follows communal or linguistic lines. Language differences and exclusive face-to-face social networks may channel the dissemination of ideas along ethnic lines. These ethnic segments rarely start out with a highly developed sense of national political identity or nationalist political goals. On the contrary, ethnicity typically becomes politicized as nationalism only after the emergence of mass political discourse. Still less do ethnic market segments start out with consumer preferences for militant, xenophobic nationalism, based on "ancient hatreds." An ethnically differentiated market segment may, however, share a set of common experiences and a common, parochial discourse, which mythmakers can exploit.(45)

Propaganda is most effective when it taps into the audience's predispositions or when it can link a new idea to attitudes that the audience already holds.(46) Thus, Milosevic's success in mobilizing Serbian ethnic sentiment was due not only to his monopoly over Belgrade television but also to the historical legacy of ethnic conflict and the tense situation between Albanians and Serbs in Kosovo, which left his Serbian audience primed to accept his divisive and uncontested message. In this situation, there was positive feedback between supply and demand, in that segmented public opinion was ripe for nationalist appeals, which in turn increased nationalism and deepened the segmentation of the market.

Even those scholars who tout the rationality of public opinion attach two crucial qualifications, one on the supply side and one on the demand side: the public responds rationally to events within the limits of the information and analysis that it receives, and given its predispositions.(47) John Zaller, for example, shows that American voters rely for their opinions on perceived experts whom they believe share their own values.(48) In this view, experts do not tell people what to care about, but they do shape people's estimates of the costs and feasibility of various means for pursuing the ends that they value. Consequently, demand reflects not only the preferences of consumers but also the extent to which consumers with similar predispositions are isolated in separate market segments, each dominated by a single supplier.

Often the ethnic or communal segmentation of the market is not a spontaneous reflection of language or traditional social organization, but rather the modern artifact of elite strategies of divide and rule. For example, European colonial rulers - whether Stalin in Central Asia or Belgians in Rwanda - often highlighted or even created ethnic cleavages in order to split local populations and insure the dependence of native functionaries. Even in the heart of Europe, Bismarck and his successors concocted the segmentation of the German marketplace of ideas through their nationalistic agenda-setting and electoral propaganda, which divided the middle classes from socialists and Catholics, who were stigmatized as "enemies of the Reich." The belligerent tone of the bourgeois press, pressure groups and associations like the Navy League and Colonial Society, and political parties were all shaped by the nationalist themes around which elections were fought. Militarist ideas promoted in these campaigns and fostered by the middle-class, Protestant, patriotic organizations they spawned - including the notions of a need for Lebensraum, victimization by the encircling great powers, the superiority of German culture, and the spiritual benefits of war - became standard fare in right-wing thinking. In this way, electoral tactics erected high walls between segments of German society, which continued to shape political discourse and electoral strategies down through the Weimar period.(49)

Sometimes elites segment the marketplace in a way that inadvertently loads the dice in favor of nationalist ideas. Tito's decentralizing reforms of the 1960s, which were intended in part to assuage and defuse ethno-nationalism, put Yugoslavia's media in the hands of regional leaderships, which in the 1980s fell into the hands of nationalists like Milosevic. This federalization of power left pan-Yugoslav reformers like Ante Markovic with no instrument for transcending the Serb and Croat nationalists' media monopoly over their respective ethnic niche markets.

While governmental elites in democratizing states are segmenting the emerging marketplace of ideas, counter-elites and consumers are rarely passive. Often, elite manipulations produce unintended consequences, and nationalist mobilization spins out of control. In Wilhelmine Germany, for example, the strategy of popular nationalism became less and less manageable for the "iron and rye" coalition of heavy industrialists and landed aristocrats. Numerically, the working class grew faster than other constituencies, shrinking the base from which the government could mobilize a majority. Moreover, the conservative elites' mass allies increasingly tried to use nationalist issues to push the old elites aside. After Germany's supposed humiliation by France in the Moroccan Crisis of 1911, mass nationalist groups and middle-class military officers claimed that the old elites running the German state lacked the dynamism to meet the looming challenge from Germany's enemies. To stay ahead of this tide of popular criticism, even the Junker aristocrats leading the Conservative party felt compelled to slam the weak policy of the Bethmann Hollweg government.(50) The elites had unleashed a power that they were unable to control.

Segmentation of demand, in short, may be shaped by a number of factors: the pre-existing preferences or experiences of groups sharing a common outlook; differentiated preferences induced by targeted advertising; division of media markets by language or region; or divisions imposed by political boundaries, as in federal systems. Such factors may be overridden, however, if political discourse is channeled into a wider framework by strong catchall parties or non-partisan media institutions.

MEDIA INSTITUTIONS AND NORMS

Where markets are imperfect, increased freedom of speech will tend to exacerbate nationalist mythmaking unless institutions and norms correct the flaws in the market. A well-institutionalized marketplace of ideas requires anti-trust and equal time regulations guaranteeing media access, the training of journalists in the verification of sources and the separation of fact from opinion, and the development of expert evaluative institutions whose prestige depends on maintaining a reputation for objectivity. Without such regulatory institutions, free speech by itself will not guarantee that a range of voices is effectively heard, that competing arguments are forced to confront each other on the merits, that participants in debate are held accountable for the accuracy of their statements, that factual claims are scrutinized, that experts' credentials are verified, that hidden sources of bias are exposed, or that violators of the norms of fair debate are held up to public censure.(51)

Regulation entails some risk of abuse, the severity of which depends in part on how it is carried out. In centralized forms of regulation, a state official or governmental body decides who has access to the media and what are the ground rules for its use. In contrast, decentralized regulation is achieved through routines of professional behavior in institutions such as the professional media, universities, think tanks, and legislative oversight bodies. Both forms of regulation may be useful antidotes to market imperfections, and both may be used in combination. Decentralized regulation is generally preferable, since centralized regulation creates the risk that the state will exploit its regulatory power to establish its own media monopoly. However, where decentralized institutions are weak or lack the required professional norms, centralized regulation, especially if it is subject to democratic control or held accountable to international standards, may be preferable to an imperfect, unregulated marketplace.

Similarly, the regulation of the content of speech, such as the banning of hate speech, is more subject to abuse than the establishment of norms of debate, which set standards for how people are expected to argue their cases. The latter would include the professional journalist's norm of distinguishing facts from opinion, the scholar's norm of citing sources of alleged facts, and the League of Women Voters' norm of expecting candidates to debate issues in a common forum in front of a panel of disinterested expert questioners. Establishing strong norms of debate is generally preferable to regulating the content of speech, but when norms are weak, content regulation may also be needed. Like centralized regulation, content limits should be accountable to democratic oversight or international standards.

Regulation is not a panacea. Indeed, skeptics doubt how well media institutions structure public debate even in the most mature democracies. Nevertheless, there is substantial evidence that effective evaluative institutions do have an impact on public views. Studies show that, apart from the influence of a popular president, American public opinion is swayed most strongly by the media testimony of experts who are perceived to be credible and unbiased.(52)

If the marketplace of ideas is imperfect even in mature democracies, its flaws are still more grave in new democracies. An integrated public sphere, in which each idea confronts every other idea on its merits, does not get created overnight. Without the functional equivalents of institutions like the New York Times, the MacNeil-Lehrer News Hour, the Brookings Institution, and the Congressional Budget Office, discussion may be open, but an exchange and evaluation of contending views before a common audience may not occur. In many newly democratizing societies, press laws are biased and capriciously enforced.(53) The middlemen of the marketplace of ideas - journalists, public intellectuals, and public-interest watchdogs - tend to perform poorly in the initial stages of the expansion of press freedom. Instead of digging out the truth and blowing the whistle on fallacious arguments, journalists in emerging markets are often beholden to a particular party or interest group, make little attempt to distinguish between fact and opinion, and lack training in the standards of journalistic professionalism.(54) While Thomas Jefferson said that if forced to choose, he would rather have a free press than a democratic government, in assessing the actual state of the press in young America, he remarked that "a suppression of the press could not more completely deprive the nation of its benefits, than is done by its abandoned prostitution to falsehood. Nothing can now be believed which is seen in a newspaper. Truth itself becomes suspicious by being put into that polluted vehicle."(55)

Even if a new democracy has a responsible elite press, its ability to impose a coherent structure on discourse may not penetrate to the grassroots level. Weimar's liberal, Jewish-owned, mass circulation newspapers were objective and even erudite, but their ideas failed to penetrate beyond Berlin or Hamburg. Even in those urban centers, workers read the liberal press only for the sports, feature stories, and movie listings, ignoring the political views of the "class enemy." Today, India's elite English-language press has a laudable system of self-regulation and responsible coverage of communal conflict, but the populist vernacular press remains immune to these high standards. In newly democratizing states, the penetration of ideas to the grassroots often requires face-to-face contacts. In India, at least before Indira Gandhi's time, this was accomplished by the moderate, secularist political machine of the Congress Party. In contrast, at the dawn of Sri Lankan democracy in the 1950s, only Sinhalese Buddhist priests, who fiercely opposed toleration of the Tamil Hindu minority, had networks for persuading voters at the crucial village level.(56)

MARKET FORCES THAT PROMOTE NATIONALIST MYTHMAKING

In summary, under conditions of incipient democratization, the increased openness of public debate often fosters nationalist mythmaking and ethnic conflict because opportunistic governmental and non-governmental elites exploit partial monopolies of supply, segmented demand, and the weakness of regulatory institutions in the marketplace of ideas. We argue that the greater these market imperfections (that is, the greater the rivalry between oligopolistic elites, the greater the consumer segmentation, and the more dependent and partisan are media institutions), then the greater the likelihood for nationalist mythmaking to dominate public discourse, and the greater the likelihood for mythmaking to promote conflict. Conversely, the more perfect the marketplace and the more integrated the public sphere, the less effective is nationalist mythmaking. These hypotheses are probabilistic, not invariant relationships or sufficient causes. This article does not present a systematic test of all these propositions. As a first step towards evaluating them, however, we examine two hard cases, Yugoslavia and Rwanda, which are often invoked on behalf of the conventional wisdom that governments are largely responsible for nationalist mythmaking and that unconditional free speech is the best antidote.

Monopolizing Market Segments in the Former Yugoslavia

On the surface, the story of the media in the Yugoslav conflict may seem to fit the Human Rights Watch analysis quite well. Government officials in the republics of Serbia and Croatia used their near-monopoly control of the news media to fuel their publics' ethnic prejudices, mobilizing a popular nationalist constituency to support their rule while discrediting more liberal opponents. However, the media monopoly merely gave elites in the republics the tools to sell nationalist myths. The motive and the opportunity were created by the Serbian elite's fear of democratization, by the plausibility of these myths to consumers in a segmented market, and by the unevenness of journalistic standards. Under these highly imperfect market conditions, the weakening of the central Yugoslav state created a potential opening for increased political pluralism, which threatened the oligarchs who ruled the federal republics and also created an opportunity for political entrepreneurs - including politicians, journalists, and intellectuals - to exploit their media market power in the competition for mass support. Tito's dispersion of control over television to the republics in the 1960s and 1970s, under the theory that a federalist devolution of power would dampen underlying ethnic tensions, turned out to have been a grave mistake. By 1989, when Yugoslav Prime Minister Ante Markovic finally embarked on the creation of an all-Yugoslav television network, it was already too little, too late. This suggests that NGOs' standard prescription of reducing centralized state media power needs strong qualifications.

In 1987, Slobodan Milosevic, head of the Serbian Central Committee of the League of Communists, mounted a systematic campaign using his control over the Serbian state television monopoly to convince the Serbian people that Serbs residing in Kosovo province, the historic cradle of Serbdom, were suffering discrimination, repression, and rape at the hands of the Albanian majority there. He chose the television correspondent who would report to Belgrade from Pristina, the capital of Kosovo, and personally phoned the station almost daily to tell the editors what stories to highlight.(57) After Milosevic's April 1987 speech in Kosovo, Belgrade TV showed the local Albanian police clubbing the Serbian crowd, and Milosevic saying "From now on, no one has the right to beat you," but it left out the pictures of the crowd stoning the police.(58) Exploiting the wave of chauvinist sentiment touched off by this media campaign, Milosevic used the Kosovo issue as a pretext to purge anti-Milosevic journalists, charging them with issuing "one-sided and untrue reports," and to consolidate conservative domination in party circles in Belgrade. Thus, nationalist media manipulation was the centerpiece of Milosevic's successful strategy for defeating liberal reformers in the scramble for both mass and elite support as power devolved from the center in the post-Tito period. Milosevic never achieved an absolute monopoly over the Serb media, but he controlled its commanding heights, the state television station and Belgrade's three major daily newspapers. An independent TV station and the semi-independent Borba newspaper were prevented by low wattage and limited newsprint from reaching beyond the Belgrade suburbs into Milosevic's stronghold in rural Serbia.(59)

Because of Yugoslavia's decentralized federal structure, republican television stations were totally independent of the central government, but were monopolized by the republican Communist parties. The Yugoslav media, like most other aspects of Yugoslav life, had become by the 1980s "an alliance of regional oligarchies."(60) Republican television stations would not even show Prime Minister Markovic's speeches. To combat this, Markovic established an all-Yugoslav network, Yutel, in 1989. However, the central government's financial limitations, themselves a consequence of Yugoslavia's federal structure, left Yutel dependent on army surplus equipment and the sufferance of local broadcasters. After only four months on the air, Croatia pulled the plug on Yutel over a sensitive story on Slavonia, and most other republics followed suit. As the coup de grace, Serbian nationalist thugs trashed Yutel's Belgrade office.(61) Thus, the ability of republican government leaders to manipulate the mass media reflected the collapse of the multi-national Yugoslav state.

However, the ethnic segmentation of the media market cannot be blamed entirely on republican governments. Journalists and scholars also played the ethnic card, in some cases well before Milosevic. Many Serbian intellectuals were obsessed with the Albanian threat in Kosovo even before Milosevic began his media campaign on the issue. In 1986, for example, a large number of prominent members of the Serbian Academy of Sciences published a memorandum on the "genocide" being perpetrated against Serbs in Kosovo. This document was condemned by the Serbian Central Committee and the mainstream Belgrade press, still operating under traditional Yugoslav norms of comity between ethnic groups, though Milosevic urged them to keep the condemnation secret. Some of these nationalist intellectuals and a portion of the Serbian journalistic community may have been acting partly out of sincere concerns. But in the view of some analysts, these intellectuals saw the Kosovo issue as a vehicle for breaking down communist limitations on intellectual freedom and for press "liberalization."(62) This reflected the necessity for all Yugoslav elites to reposition themselves on a new foundation of ideological legitimacy in the context of the waning of centralized communist authority. In this setting, the professional journalistic community split, some choosing the nationalist route and energetically aiding the Milosevic takeover, some resisting it and ultimately being forced out.(63) Mark Thompson of the journalism NGO Article 19, though generally a strong partisan of Yugoslavia's independent journalists, describes the Milosevic takeover in the fall of 1987 as "a collusion among Serbia's Communist politicians, its bureaucracy, its intellectual class, and its news media."(64)

Thus, organized forces in "civil society," no less than in government, saw the benefits of the strategy of monopolizing media control within a market niche. They were sometimes even willing to conspire explicitly with the ethnic archfoe to accomplish it. For example, after Serb, Croat, and Muslim nationalist parties emerged as the winners of Bosnia's 1991 elections, the three nationalist foes tried to collude to divide up among them the assets of Bosnia's integrated, civic television service, and to exclude the moderate parties of their respective ethnic groups.(65)

The success of media propaganda depended both on monopoly of supply and also on the nature of demand, including the plausibility of the message in light of consumers' predispositions. Some propaganda campaigns were strikingly successful. For example, the Serbs enjoyed a six-month period of television monopoly in northern Bosnia, which they used to prime their population for the 1992 campaign of "ethnic cleansing" by repeatedly charging that Muslims were plotting to establish an Islamic fundamentalist state. Later, Serbs guarding prison camps accused their Muslim captives of precisely the charges that had been reiterated on the news. Similarly, as a result of Serb propaganda, 38 percent of Belgrade residents in a July 1992 poll thought that it was the Muslim-Croat forces who had recently been shelling the Bosnian capital of Sarajevo, versus only 20 percent who knew it had been the Serbs.(66) However, viewers refused to swallow every lie whole. When the popular nationalist Vuk Draskovic mounted a mass anti-war rally in March 1991, the government-controlled media's attempts to portray him as in league with the Croats and Albanians fizzled as too implausible. The following year, only 8 percent of Serbian respondents thought that state television kept them "well informed," versus 43 percent for the independent media.(67)

Thus, the impact of the supply of nationalist propaganda must be assessed in light of the demand for it. As Mark Thompson put it: "People's bedrock attitudes toward the wars in Croatia and Bosnia are not created by the state media; rather, the media play variations upon those attitudes, which derive from other sources (national history, family background, education, oral culture). Media did not inject their audiences with anti-Muslim prejudice or exploitable fear of Croatian nationalism. The prejudice and fear were widespread, latently at least; there was a predisposition to believe 'news' which elicited and exploited the prejudice; without the media, however, Serbia's leaders could not have obtained public consent and approval of its nationalist politics."(68)

The importance of underlying predispositions is demonstrated by comparing the propaganda strategies that Milosevic tailored for the Serbs and those Tudjman adopted for the Croats. Belgrade television portrayed the Serbs as always on the defensive, the perennial victims of every battle. Dead Serbs were favored imagery. This was thought to strike the right chord in a people who glorify a defeat at the hands of the Turks half a millennium ago in the battle of Kosovo. In contrast, government propaganda directives told Croatian television to soft-pedal defeats, never show footage of destroyed Croat towns, and "always finish such reports with optimistic declarations and avowals."(69) The government feared that Croats, lacking as firm a tradition of statehood as the Serbs had, might simply give up hope if they knew the odds they faced.

What was lacking in the Yugoslav case was not just free speech, but strong institutions to counteract market imperfections and to promote a professional, unbiased, pan-Yugoslav mass media. Standard antidotes to state power would have been of doubtful effectiveness in this case, even though it is true that media monopolies in the hands of republican governments caused most of the damage. Federalism, that standard remedy for constraining state exploitation of ethnic minorities, was in fact one of the main problems. Moreover, "consociational" power-sharing, which is often prescribed as a complement to federalism, was also troublesome. In the Bosnian media, for example, the practice of allotting equal time for each group's biases made the evening news a series of stories with different slants, while the true story of the Yugoslav army's role in the attacks on Sarajevo, for example, was suppressed as a violation of consociational comity.(70) Likewise, providing piecemeal subsidies to individual newspapers in the capital city, as the International Federation of Journalists did for Borba, failed to go to the heart of the problem, since the backbone of support for nationalism lay in the Serb countryside, where Milosevic's media monopoly was uncontested. Finally, simply prescribing maximum freedom of speech would have been unavailing, given the inclinations and the capacity of various elite strata, both inside and outside the government, to exploit the population's predispositions to ethnic anxiety.

Rwandan Hate Radio

The 1994 mass murder of some 800,000 Tutsi and moderate Hutu organized by extremist Hutu in top circles of the Rwandan government is another case that may seem to fit the Human Rights Watch analysis perfectly. Officials of the authoritarian regime of President Juvenal Habyarimana, feeling their power endangered, used their monopoly control of mass media and university appointments to create a "finely tuned propaganda machine" that played on Hutu fears of the former Tutsi elite and purveyed false, inflammatory versions of the history of relations between the two groups. In April 1994, the Hutu official clique unleashed militias trained in the techniques of genocide. Independent journalists were a special target in the first wave of the killings. At the same time, Radio-Television Libre des Mille Collines, a pseudo-private station established by Habyarimana's wife, spread the word that Tutsi rebels were about to rise up and kill Hutu, and consequently that all Hutu should join the militias in a campaign of preventive killing. Militias threatened to kill Hutu who did not participate in the genocide, so it is difficult to judge how much of the killing was triggered by the radio propaganda per se.(71) Nonetheless, all sources agree that the hate broadcasts played a significant role in the second phase of the killing, after the initial militia sweeps. Holly Burkhalter, the Washington director of Human Rights Watch, argued that jamming the hate radio was "the one action that, in retrospect, might have done the most to save Rwandan lives." The radios instead withdrew from the advancing Tutsi army into the safe haven of the French army zone, where they continued to broadcast.(72)

NGOs such as Human Rights Watch and Africa Rights, as well as many independent scholars, drew the lesson that the international community needs to encourage Rwanda and Burundi to democratize, to foster an independent press, and to bring the perpetrators of genocide to justice.(73) However, upon closer examination, their prescriptions are contradicted by their own highly persuasive analyses of the causes of the Rwandan genocide. After the genocide, NGOs continue to advocate precisely those measures that their analyses show to have triggered the killings: an increase in political pluralism, the prospect of trials of the guilty, and the promotion of anti-government media.(74)

In the late 1980s and early 1990s, falling coffee prices and economic disruptions caused by fighting with Uganda-based Tutsi rebels put the Habyarimana regime on the defensive. Under intense pressure from the domestic Hutu opposition and from international aid donors, the regime agreed under the 1993 Arusha Accords to a limited political opening, involving power-sharing with opposition groups, the legalization of numerous opposition political parties calling for democratic elections and, as an Africa Rights report puts it, "an explosion in the number of newspapers and journals" published by anti-government groups after the abandonment of the press monopoly in July 1990.(75) "A vibrant press had been born almost overnight," says Gerard Prunier, but its biased commentary was written "in terrible bad faith."(76) Hutu extremists attached to the regime continued to monopolize the radio, a key asset among a population that was 60 percent illiterate. After a Tutsi rebel attack on the capital in 1993 was parried only with the help of French troops, Habyarimana had no alternative but to accept the Arusha agreement, which provided for Tutsi participation in government, a rebel military unit to provide security for Tutsi politicians in the capital, and the exclusion of Hutu extremists from the joint Hutu-Tutsi government. Moderate Hutu from southern Rwanda, where "Hutu" and "Tutsi" were racially almost indistinguishable, began to mobilize politically against Hutu extremists in the government clique and their northern Rwanda social base.(77)

As part of the settlement, an international commission named names of highly placed Hutu extremists who were complicit in small-scale killings of Tutsi. "Individuals named were promised an amnesty," says Africa Rights's Alex de Waal, "but knew that their actions were under scrutiny," and so distrusted these guarantees. Human rights groups were active in this period of internationally sponsored power-sharing and pluralization. "Rwanda had one of the most vigorous human-rights movements in Africa," says de Waal. "Six independent human rights organizations cooperated in exposing abuses by government and rebel forces."(78)

In this setting, the clique around Habyarimana had every reason to fear democratization and calls for justice from the international community. To forestall a fall from power and judicial accountability, these officials developed the plan for a mass genocide. "The extremists' aim," says Africa Rights, "was for the entire Hutu populace to participate in the killing. That way, the blood of genocide would stain everybody. There could be no going back for the Hutu population."(79) But there was a flaw in this plan. Habyarimana, heavily dependent on foreign aid to prop up his system of official patronage, balked at implementing a bloodbath that he knew would cut him off from foreign funds. The president's extremist allies in the military and security services had no such qualms. From January to March 1994, their unofficial journal Kangura, an example of the "flowering" of Rwandan media in the period of pluralism and incipient power-sharing, warned Habyarimana not to flinch from the destruction of the Tutsi and predicted with astonishing accuracy the details of his assassination:(8) Habyarimana was killed in April by his own presidential guard upon returning from a meeting at Dar Es Salaam where he made renewed concessions to international donors, the UN, and the Organization of African Unity. As de Waal aptly states, "Habyarimana was a victim of the international peace industry."(81)

Despite the clear evidence that NGO analysts themselves recite, they fail to acknowledge that the very solutions they continue to promote are the same as the steps that caused the killings. Human Rights Watch correctly notes that the "free and fair" election of Burundi's first Hutu president in October 1993 set the stage for the killing of some 50,000 Hutu and Tutsi. The Tutsi military, fearing that the elected government's power-sharing scheme would neutralize the army as a security guarantee for the Tutsi minority, launched a coup to protect its monopoly of force, touching off a series of reprisals. Yet Human Rights Watch urges democratic accountability and prosecution of the killers "to deter further slaughter," despite the fact that it was precisely the threat of such accountability that provoked the slaughter in Rwanda and Burundi in the first place.(82)

Both the Rwanda and Burundi cases show that the ideals of democratic rights, uncompromising justice, and free speech must make pragmatic accommodations to recalcitrant reality. Recognizing this, Reporters sans Frontieres warns that the "error committed in Rwanda, which consisted of applying the rule of 'laissez faire' in the name of the principle of liberty of the press, must not be repeated in Burundi." While working to reconstitute the private news media in both countries and to bring journalists implicated in the genocide campaign to justice, the French NGO acknowledges that the thirteen newspapers that it is helping in Rwanda are short on personnel, paper, and facilities; have a circulation under 1,000 each; cost a day's wage to buy one issue; and consist primarily of opinions, not news. Realistically skeptical about some of the journalists it supports, Reporters sans Frontieres conditions aid on a pledge to forswear ethnic hate speech. In Burundi, Reporters sans Frontieres notes the paradox that many journalists working under a new law on press freedom are calling for an ethnic dictatorship that would shut down non-official expression of views. Since the invisible hand of the marketplace of ideas is so unreliable in such circumstances, Reporters sans Frontieres relies also on the visible hand of two international radio stations broadcasting into Rwanda and Burundi from Zaire.(83)

Conditions for Successful Liberalization of the Marketplace of Ideas

In numerous recent cases, such as South Africa, increases in press freedom and democratic participation in politics spawned no sanguinary outbursts of nationalism. Historically, Britain democratized and evolved a free press without developing German-style populist nationalism. These cases had better outcomes because their elites had weaker motivations to propound nationalist myths, because their markets were not as segmented, or because effective institutions of free debate were in place before the democratization of political participation. If elites believe that the expansion of free speech and democratic participation poses little threat to their interests, nationalism will be moderate. This pattern suggests that activists should target their efforts at patiently putting in place these preconditions of constructive public discourse, rather than clamoring for no-holds-barred press freedom across the board. Institutional foundations of free debate are achieved either by historical evolution or by conscious design, not instantaneously by the invisible hand of competition.

The paradigm-setting case for these conditions is Great Britain. England did fight an intense civil war in the seventeenth century shortly after a dramatic increase in the number of newspapers and the freedom of political expression in them.(84) However, by the dawn of the age of mass nationalism in the late eighteenth and early nineteenth centuries, Britain had already achieved a set of well-established norms of free speech among wide circles of the elite. The decisive move to end censorship in Britain came at the beginning of the eighteenth century, at a time when the political position of the Whig aristocracy was at its strongest and the proportion of the British population who could vote in parliamentary elections was actually declining. In that same era, Britain's integration of the Scottish and Welsh peoples into a centralized state before the era of mass democracy prevented ethnic segmentation of the market, at least in the core of the realm.(85) By the mid-nineteenth century, when the penny press and the expansion of the electoral franchise further widened the scope of political debate to include the middle class, existing journalistic institutions and norms of debate provided a structure to channel and regulate the exchange of ideas. Moreover, Britain's traditional ruling class shared many commercial interests with the rising middle class, and so had little reason to "play the nationalist card" to forestall democratic policies like the repeal of tariffs on imported food. In a limited way, some aristocratic demagogues, most notably Lord Palmerston on the eve of the Crimean War, succeeded in diverting public opinion away from political reform towards a foreign policy of nationalist expansion, but these adventures were moderate in scope. Though Britain conquered vast portions of the globe, it generally did so in a cost-conscious way: it appeased foes strategically, pulled back from overcommitments, and never placed its nationalism beyond the pale of rational discussion weighing the costs and benefits of imperial policies.(86)

Several contemporary success cases of relatively peaceful democratization and media liberalization share one or more of these characteristics of the British case: that is, guarantees of the interests of powerful elites, no ethnic segmentation of the market, or thorough institutionalization of the marketplace of ideas before democratization. Indeed, some of these moderate cases are former British colonies, which inherited along with the English language a tradition of professionalized journalism from the colonial period. One factor in the smooth South African transition, for example, was the well-established English-language opposition press, exemplified by the Rand Daily Mail and its successors, which for decades had been consistently more liberal than many of its readers.(87) The political opening in the 1990s permitted the English-language press to report more freely, and a new black press was funded by churches and through Danish and Norwegian subsidies. Television and especially print news, already staffed with a professional cadre, moved quickly toward international norms. This allowed the divisions between Afrikaner, English, and black media that had prevailed under apartheid to be overcome rapidly.(88) Moreover, Nelson Mandela's moderate rhetoric reassured whites about the consequences of free speech, as does their residual power to veto threatening developments.

India is another case in which a balanced marketplace of ideas was well institutionalized long before the transition to democratic politics, and in which the central ruling elite saw its interests as served by moderate, secular policies rather than divisive, ethno-communal ones. By the turn of the century, the English-language Indian press was already able to use the pressure of open public debate to constrain the non-elected British regime's policies.(89) By the time of independence, a number of highly professional, major urban newspapers had developed a voluntary press code for reporting on communal riots, which abjured inflammatory headlines, refrained from specifying casualty figures during the heat of the moment, scrupulously cited sources, and dug for accurate information on the causes of riots. These informal codes were institutionalized in the Press Council, modeled on its British forebear. The Council was given the same statutory powers to investigate violations as a civil court.(90) Smaller, partisan papers sometimes inflamed communal tensions, however, and during the 1947 riots a restrictive press ordinance with pre-publication censorship was temporarily adopted.(91) In provincial towns, publishers and journalists are highly dependent on the support of local business elites, and expedience often gets in the way of truth in reporting on communal tensions. The vernacular (i.e., non-English) press commonly circulates false reports, inflated death figures, and unevaluated statements by communal leaders. This gap between the restrained, professional, state-wide press and the inflammatory communal press has been growing over the past two decades, as a result of economic change and the growth of literacy in provincial areas.(92) Social change, sharply rising newspaper readership, and the emergence of a local intelligentsia has played a central role in reigniting communal conflict in Kashmir, for example.(93)

In the heyday of the Congress Party's centralized, secular leadership, local Congress notables kept a short leash on the expression of communal prejudices. In the 1970s, however, Indira Gandhi sought to free herself from Congress's grassroots organizational structure through direct appeals not only to religious groups, but also to increasingly politicized lower-caste and lower-class segments of Indian society. As Human Rights Watch argues correctly, these segmental appeals touched off an increasing communalization of Indian politics.(94) However, the problem nowadays is hardly that the central, secular elites, including the elite media, are too demagogic, but that their power was weakened vis-a-vis nationalist challengers by the de-institutionalization of the Congress Party.

In other contemporary cases, the nationalist dogs are not barking because old elites have found a safe haven in the new regime. In many of the democratizing post-communist states of Eastern Europe, former apparatchiks have profited from the privatization of industry, attracted a mass constituency based on appeals to economic security rather than nationalism, and still rule many of these countries. Likewise, in Latin America, former military elites have been given "golden parachutes," and journalistic institutions were already well established as the result of earlier periods of democratization.(95)

Prescriptions for an Integrated Marketplace of Ideas

Democratization and free speech can be made compatible with ethnic harmony and the moderation of nationalist sentiment only under favorable conditions of supply, demand, and institutional regulation. If these conditions do not exist, they need to be created before, or at least along with, the unfettering of speech and political participation.

On the supply side, the international community may be needed to help break up information monopolies, especially in states with very weak journalistic traditions and a weak civil society. In Cambodia, for example, the UN's successful media and information program was designed, according to the UN commander, to "bypass the propaganda of the Cambodian factions" by directly disseminating information about the elections.(96) The breakup of monopoly power over politics and discourse must coincide, however, with measures to reduce elites' incentives for nationalist mythmaking or to eliminate their capacity to make trouble. As our cases show, it is reckless for the international community to threaten elites with across-the-board exposure and prosecution of past crimes, unless there exists the will and capability to render harmless the likely backlash from elites that are pushed to the wall. Otherwise, elites that are potentially threatened by democratization and the end of censorship should be guaranteed a soft landing in the emerging open society. Many Latin American and East European countries have done well by keeping prosecutions limited. In contrast, fine moral declarations without effective actions are the worst possible policy.

On the demand side, ethnically segmented markets should be counteracted by the promotion of civic-territorial conceptions of national identity, as in Ukraine. Inclusive national identities can be fostered through an integrative press, which expresses a variety of outlooks on the same pages. All too often, international aid goes to the opposition press in democratizing countries regardless of its journalistic quality, on the grounds that creating a pluralism of voices is the essential objective. In Romania, for example, the U.S. Agency for International Development has subsidized anti-government newspapers that fail to meet even the most minimal standards of accuracy in reporting.(97) Instead, aid should go to forums that present varied ideas, not a single line, in a setting that fosters effective interchange and factual accuracy. In post-1945 Germany, for example, American occupiers licensing newspapers showed a strong preference for editorial teams whose members spanned diverse political orientations.(98) This approach extends Donald Horowitz's critique of Arend Lijphart's strategy of "consociational" representation of communal power blocs into the realm of public discourse. Whereas Lijphart's approach rewards politicians who mobilize support along ethnic lines, Horowitz advises electoral rules that reward vote-pooling, in order to promote cross-ethnic political alliances and to break down the communal segmentation of politics. Applying Horowitz's principles to the marketplace of ideas, we counsel idea-pooling through integrative public forums, to break down the intellectual boundaries between ethnically exclusive "imagined communities."(99)

For this reason, we urge NGOs and other aid donors to reconsider projects to provide ethnic minorities with their "own" media.(100) Instead, we suggest supporting media that strive to attract a politically and ethnically diverse audience, invite the expression of various viewpoints, and hold news stories to rigorous standards of objectivity. This can be done by expanding existing NGO programs to train journalists from newly democratizing countries, such as those of the International Press Institute in Vienna,(101) and by providing quality news organizations with equipment, subsidized newsprint, or other logistical support. Special efforts should be made to encompass the regional and local press in these efforts. In case after case - Weimar, Germany, India, Sri Lanka, and contemporary Russia - key vehicles of nationalist mythmaking have been face-to-face networks and rough-hewn periodicals. To provide an effective alternative to these, media projects should focus on the inclusion of local journalists in the activities of state-wide media associations, mid-career training sabbaticals for grassroots journalists, and financial subsidies to make a high quality local press independent and affordable.

Major efforts should be made to promote the institutionalization of effective norms of elite discourse, journalistic professionalism, and independent evaluative bodies before the full opening of mass political participation. Whenever possible, market imperfections should be counteracted by decentralized institutions, not centralized regulatory directives, and by the promotion of norms of fair debate, not by restrictions on the content of speech. In some cases, however, certain kinds of constraints on speech may be necessary in multiethnic societies while these institutions are being built. This may be ethically uncomfortable for Western liberals; moreover, it is politically difficult to design constraints on democracy and free speech that do not play into the hands of elites who want to squelch freedom entirely. When electoral polarization touched off communal riots in Malaysia in 1969, for example, Malay elites banned public discussion of ethnic issues and imposed a regime of ethnic coexistence that insured Malay political domination and economic prosperity for the Chinese business community. After a quarter-century of tight press controls, the uneasy communal peace still holds, but this interlude, which might have been used to prepare an institutional infrastructure for a more durable, democratic solution, has been squandered.(102)

Neither the ethnic strife unleashed by unchecked democratization in cases like Sri Lanka nor the temporary, repressive communal cease-fire in cases like Malaysia is desirable. One element of a better solution is for international donors to offer incentives to political and economic to elites to prepare the institutionalization of open discourse, while tolerating some limits on free expression, including limits on ethnic hate speech, in the short run. Another element is direct aid to professionalize those elements of the media that are attempting to create a integrated forum for responsible, accurate debate. But when these remedies are unavailing, those who value both unfettered speech and peace must, without illusions, assess the tradeoff between them.

We thank Fiona Adamson, Laura Belin, Mark Blyth, V.P. Gagnon, Sumit Ganguly, Robert Jervis, Arvid Lukauskas, Edward Mansfield, Anthony Marx, Helen Milner, Alexander Motyl, Anne Nelson, Bruce Pannier, Laura Pitter, Ronald Rogowski, Aaron Seeskin, Robert Shapiro, Kevin Smead, Stephen Van Evera, Robin Varghese, and Leslie Vinjamuri for helpful comments or other assistance, and the Harry Frank Guggenheim Foundation, the Institute for the Study of World Politics, and the Pew Charitable Trusts for financial support.

1. Human Rights Watch, Playing the "Communal Card" (New York: Human Rights Watch, April 1995), reprinted as Slaughter among Neighbors: The Political Origins of Communal Violence (New Haven, Conn.: Yale University Press, 1995); Human Rights Watch, "'Hate Speech' and Freedom of Expression," Free Expression Project, Vol. 4, No. 3 (March 1992); see also Leonard Sussman, Press Freedom 1996: The Journalist as Pariah (New York: Freedom House, 1996); Article 19, the International Centre Against Censorship, Guidelines for Election Broadcasting in Transitional Democracies (London: Article 19, August 1994).

2. Eric Hobsbawm, "Mass Producing Traditions," in Hobsbawm and Terence Ranger, eds., The Invention of Tradition (Cambridge, U.K.: Cambridge University Press, 1983), pp. 263-307; Eric Hobsbawm, Nations and Nationalism since 1780 (Cambridge, U.K.: Cambridge University Press, 1990), pp. 80-100, 141-142; Stephen Van Evera, "Hypotheses on Nationalism and War," International Security, Vol. 18, No. 4 (Spring 1994), pp. 26-33; Paul Kennedy, "The Decline of Nationalistic History in the West, 1900-1970," Journal of Contemporary History, Vol. 8, No. 1 (January 1973), pp. 77-100. V.P. Gagnon, Jr., "Ethnic Conflict as Demobilizer: The Case of Serbia," Institute for European Studies Working Paper No. 96.1 (Cornell University, May 1996), argues that propaganda does not mobilize popular nationalism, but rather that ethnic conflict and media control demobilizes opposition to the regime.

3. Van Evera, "Hypotheses," p. 37; V.P. Gagnon, Jr., "Ethnic Nationalism and International Conflict: The Case of Serbia," International Security, Vol. 19, No. 3 (Winter 1994/95), pp. 130-166, at 165.

4. Van Evera, "Hypotheses," p. 33. Human Rights Watch, Playing the "Communal Card," p. viii, points this out, but disregards it when drawing conclusions. According to Freedom House's ratings, civil liberties, which include freedom of speech, improved in Yugoslavia, Rwanda, and Burundi shortly before the recent outbreaks of massive ethnic violence there. Raymond Gastil and R. Bruce McColm, eds., Freedom in the World (New York: Freedom House, 1988-93), pp. 411-412 (1988), 272-274 (1989-90), 104-105, 313-315 (1990-91), and 153-155, 429-431 (1992-93).

5. Van Evera, "Hypotheses," p. 33. In mature democracies, government policy is made by officials chosen through free and fair elections on the basis of wide suffrage; the actions of officials are constrained by constitutional provisions and commitments to civil liberties; and government candidates sometimes lose elections, and leave office when they do. We define states as democratizing if they have recently adopted some of these democratic characteristics, even if they retain important non-democratic features. See Edward D. Mansfield and Jack Snyder, "Democratization and the Danger of War," International Security, Vol. 20, No. 1 (Summer 1995), pp. 8-10. On the gap between institutions and participation, see Samuel Huntington, Political Order in Changing Societies (New Haven, Conn.: Yale University Press, 1968).

6. R.H. Coase, "The Market for Goods and the Market for Ideas," American Economic Review, Vol. 64, No. 2 (May 1974), pp. 384-391, argues that imperfect markets for ideas are in no less need of regulation than imperfect markets for goods. On the benefits of gag rules in ethnically or religiously divided societies undergoing democratization, see Stephen Holmes, Passions and Constraint: On the Theory of Liberal Democracy (Chicago: University of Chicago, 1995), chapter 7, esp. pp. 206-213.

7. The impact of Hitler's oratory before his rise to power was mainly in motivating supporters who were already predisposed toward his outlook. Few people heard him on the radio until after he had been elected chancellor. Ian Kershaw, "Ideology, Propaganda, and the Rise of the Nazi Party," in Peter Stachura, ed., The Nazi Machtergreifung (London: Allen & Unwin, 1983), pp. 162-181; Thomas Childers, "The Social Language of Politics in Germany: The Sociology of Political Discourse in the Weimar Republic," American Historical Review, Vol. 95, No. 2 (April 1990), pp. 331-358.

8. T.C.W. Blanning, The Origins of the French Revolutionary Wars (London: Longman, 1986), p. 113; Jeremy Popkin, Revolutionary News: The Press in France, 1789-1799 (Durham, N.C.: Duke University Press, 1990), p. 41.

9. Rakiya Omaar and Alex de Waal, Africa Rights, Rwanda: Death, Despair and Defiance (London: Africa Rights, September 1994, exp. ed., 1995, all page references to 1994 ed.); Mark Thompson, Article 19, Forging War: The Media in Serbia, Croatia and Bosnia-Herzegovina (London: Article 19, May 1994).

10. Human Rights Watch, Playing the "Communal Card," pp. viii, xiv.

11. Thus, the Indian Congress government was outflanked by the nationalist Bharatiya Janata Party, the Israeli Labor government by Likud, South Africa's De Klerk government by irreconcilable Afrikaners, Romania's Iliescu government by the anti-Hungarian nationalist party, the Sri Lankan government by grassroots Buddhist organizations, Serbia's Slobodan Milosevic by Vojislav Seselj, the Syrian-backed Lebanese government by leaders of the various communal groups, Russian President Boris Yeltsin by Vladimir Zhirinovsky, the moderate Armenian government of Levon Ter-Petrosian by the nationalist Dashnaks, and the inert Azerbaijani government of Ayaz Mutalibov by the ethno-populist Abulfez Elchibey. Human Rights Watch's own study often acknowledges this phenomenon, but blames the government for weakly resisting these pressures. Playing the "Communal Card," p. 28.

12. Human Rights Watch, Playing the "Communal Card," xiv, xvii. For a more nuanced NGO view, see Bruce Allyn and Steven Wilkinson, Guidelines for Journalists Covering Ethnic Conflict (Cambridge, Mass.: Conflict Management Working Paper, January 1994).

13. Pierre Birnbaum, States and Collective Action (Cambridge, U.K.: Cambridge University Press, 1988), pp. 39-40; Peter Fritzsche, Rehearsals for Fascism: Populism and Political Mobilization in Weimar Germany (New York: Oxford University Press, 1990), p. 75. Robert Putnam, Making Democracy Work (Princeton, N.J.: Princeton University Press, 1993), fails to address the Weimar case.

14. Larry Diamond, "Rethinking Civil Society: Toward Democratic Consolidation," Journal of Democracy, Vol. 5, No. 3 (September 1994), pp. 5-6; Steven Fish, "Russia's Fourth Transition," Journal of Democracy, Vol. 5, No. 3 (September 1994), pp. 31-33.

15. Ernest Gellner, Nations and Nationalism (Ithaca, N.Y.: Cornell University Press, 1983), chap. 1.

16. Although narrower definitions are in principle preferable, we adopt a broader definition that includes both state-seeking ethnic nationalism and nationalism in comparatively homogeneous, established states, on the grounds that our theory illuminates a dynamic that occurs equally in both types of case. On the relative merits of broad versus narrow definitions of nationalism, see Alexander Motyl, "The Modernity of Nationalism," Journal of International Affairs, Vol. 45, No. 2 (Winter 1992), pp. 307-324.

17. Similarly, John Zaller, The Nature and Origins of Mass Opinion (Cambridge, U.K.: Cambridge University Press, 1992), p. 313, defines "elite domination" of opinion as "a situation in which elites induce citizens to hold opinions that they would not hold if aware of the best available information and analysis."

18. Quoted in Hobsbawm, Nations and Nationalism, p. 12.

19. Anthony D. Smith, The Ethnic Origins of Nations (Oxford, U.K.: Blackwell, 1986), pp. 15, 24; John Armstrong, Nations before Nationalism (Chapel Hill: University of North Carolina Press, 1982).

20. Miroslav Hroch, Social Preconditions of National Revival in Europe (Cambridge, U.K.: Cambridge University Press, 1985), pp. 11-13; Elie Kedourie, Nationalism (London: Hutchinson, 1960).

21. However, where democratic norms are intimately bound to national self-conceptions, nationalism may in fact support peaceful democracy by promoting loyalty to civic institutions. See Victor Zaslavsky, "Nationalism and Democratic Transition in Postcommunist Societies," Daedalus, Vol. 12, No. 2 (Spring 1992), pp. 97-122; Rogers Brubaker, Citizenship and Nationhood in France and Germany (Cambridge, Mass.: Harvard University Press, 1992), chaps. 2 and 5.

22. Van Evera, "Hypotheses," pp. 26-33; Jack Snyder, Myths of Empire (Ithaca, N.Y.: Cornell University Press, 1991), pp. 5-6.

23. We define perfect competition in the marketplace of ideas, which exists only as an ideal type, as a situation of no monopolies of information or media access, low barriers to entry, full exposure of all consumers to the full range of ideas, the confrontation of ideas in common forums, and public scrutiny of factual and causal claims by knowledgeable experts. On the benefits of a free market of ideas, see John Stuart Mill, On Liberty (Cambridge, U.K.: Cambridge University Press, 1989), part 2. Mill himself argued only that unconstrained debate was a guarantee that superior ideas would not be permanently suppressed; even he did not contend that the invisible hand of competition would automatically lead to the victory of the best idea, let alone of truth.

24. Stephen Van Evera, "Primed For Peace: Europe after the Cold War," International Security, Vol. 15, No. 3 (Winter 1990/91), p. 27.

25. Gary Becker, "A Theory of Competition among Pressure Groups for Political Influence," Quarterly Journal of Economics, Vol. 98, No. 3 (August 1983), pp. 371-400.

26. Our conception of the free marketplace of ideas includes features commonly ascribed to "civil society" and Jurgen Habermas's notion of "the public sphere." Like civil society, of which it is part, the free marketplace requires a legally sanctioned societal capacity for voluntary and autonomous organization. Like the "public sphere," a well-functioning marketplace is characterized by a public debate in which rational arguments are more decisive than appeals to tradition or the status of actors. However, whereas both civil society and the public sphere are typically conceived as exclusively non-state realms of activity, the marketplace of ideas embraces both state and non-state actors, focusing on their discursive and institutional interaction. Craig Calhoun, "Civil Society and The Public Sphere," Public Culture, Vol. 5, No. 2 (Winter 1993), pp. 267-280; Jurgen Habermas, The Structural Transformation of the Public Sphere (Cambridge, Mass.: MIT Press, 1989).

27. Becker, "A Theory of Competition," p. 392.

28. Ideas may also serve as signals of commitment, staking suppliers' reputation on their promise to deliver on a policy pledge. For example, nationalists may choose to enhance their credibility with their supporters through extremist rhetoric that burns bridges to alternative constituencies.

29. Paul Samuelson, Economics, 15th ed. (New York: McGraw-Hill, 1995), chaps. 9-10, esp. p. 152; Edwin Mansfield, Microeconomics, 4th ed. (New York: Norton, 1982), chaps. 11-12, esp. pp. 323, 344-346, 353-355; James W. Friedman, Oligopoly Theory (Cambridge, U.K.: Cambridge University Press, 1983), pp. 138-145; Robert Kuenne, The Economics of Oligopolistic Competition (Oxford, U.K.: Blackwell, 1992), pp. 469-476.

30. Richard Leftwich and Ross Eckert, The Price System and Resource Allocation, 9th ed. (Chicago: Dryden, 1985), p. 407. The same uncertainty that fuels rivalrous behavior in a "young industry" typically characterizes periods of democratization and, in the absence of shared norms or effective enforcement mechanisms, often produces the same results. See Adam Przeworski, Democracy and the Market (Cambridge, U.K.: Cambridge University Press, 1991).

31. Ellen Mickiewicz, Split Signals: Television and Politics in The Soviet Union (New York: Oxford University Press, 1988); Allyn and Wilkinson, Guidelines for Journalists, pp. 17-18; Ithiel de Sola Pool, "Communication in Totalitarian Societies," in Ithiel de Sola Pool Wilbur Schramm, et al., Handbook of Communication (Chicago: Rand McNally, 1973), pp. 462-511.

32. Stanislav Andreski, "On the Peaceful Disposition of Military Dictatorships," Journal of Strategic Studies, Vol. 3, No. 3 (December 1980), pp. 3-10.

33. Robert B. Holtman, Napoleonic Propaganda (Baton Rouge: Louisiana State University Press, 1950); Richard Smethurst, A Social Basis for Prewar Japanese Militarism (Berkeley: University of California Press, 1974); on the role of the Japanese press in promoting nationalism and empire, see John A. Lent, ed., Newspapers in Asia (Hong Kong: Heineman Asia, 1982), pp. 103, 106-108; on the nationalist rhetoric of the increasingly free Argentine press on the eve of the Falklands War, see Richard Ned Lebow, "Miscalculation in the South Atlantic," in Robert Jervis et al., Psychology and Deterrence (Baltimore, Md.: Johns Hopkins University Press, 1985), pp. 98-99.

34. Modris Eksteins, The Limits of Reason: The German Democratic Press and the Collapse of Weimar Democracy (London: Oxford University Press, 1975).

35. Laura Silber and Allan Little, Yugoslavia: Death of a Nation (New York: TV Books, 1996), pp. 38-39.

36. Brett Fairbairn, "Interpreting Wilhelmine Elections: National Issues, Fairness Issues, and Electoral Mobilization," in Larry Eugene Jones and James Retallack, eds., Elections, Mass Politics, and Social Change in Modern Germany (Cambridge, U.K.: Cambridge University Press, 1992), pp. 20-23; Robert J. Goldstein, Political Repression in Nineteenth Century Europe (London: Croom Helm, 1983), p. 39; Dirk Stegmann, "Between Economic Interests and Radical Nationalism," in Larry Eugene Jones and James Retallack, eds., Between Reform, Reaction, and Resistance: Studies in the History of German Conservatism from 1789 to 1945 (Providence, R.I.: Berg, 1993), pp. 173, 183; Geoff Eley, Reshaping the German Right (New Haven, Conn.: Yale University Press, 1980), pp. 140-147.

37. On the latter point, see Karen Barkey, "Consequences of Empire," in Karen Barkey and Mark von Hagen, eds., After Empire: Multi-Ethnic Societies and Nation-Building (Boulder, Colo.: Westview, forthcoming 1997). On the contending discourses of class and nation, see Ronald Suny, The Revenge of the Past (Stanford, Calif.: Stanford University Press, 1993).

38. Gerard Prunier, The Rwanda Crisis: History of a Genocide (New York: Columbia University Press, 1995), p. 140.

39. On market segmentation in the Russian case, see Laura Belin, "Russia: Wrestling Political and Financial Repression," ibid., pp. 59-63; Victor Davidoff, "Regional Press Fights Political Control," ibid., p. 65; Laurie Wilson, "Communication and Russia: Evolving Media in a Changing Society," Social Science Journal, Vol. 32, No. 1 (January 1995), p. 113.

40. For related arguments, see Edward S. Herman and Noam Chomsky, Manufacturing Consent (New York: Pantheon, 1988).

41. Eksteins, Limits of Reason, pp. 80-81; Thomas Childers, The Nazi Voter (Chapel Hill: University of North Carolina Press, 1983), pp. 157-159.

42. Richard Bessel, "The Formation and Dissolution of a German National Electorate," in Larry Eugene Jones and James Retallack, eds., Elections, Mass Politics, and Social Change in Modern Germany (Cambridge, U.K.: Cambridge University Press, 1992), pp. 404, 412-413; Peter Fritzsche, "Weimar Populism and National Socialism in Local Perspective," in Jones and Retallack, Elections, pp. 301-304; Wolfgang Mommsen, "Government without Parties," in Jones and Retallack, Between Reform, Reaction, and Resistance, pp. 359, 372.

43. E.J. Feuchtwanger, From Weimar to Hitler (London: Macmillan, 1993), pp. 298, 313-314. On the negative effects of unmodified proportional representation systems like Weimar's in fragmented polities, see Donald Horowitz, A Democratic South Africa? Constitutional Engineering in a Divided Society (Berkeley: University of California Press, 1991), pp. 166-176.

44. Childers, "The Social Language of Politics in Germany," pp. 331-358; Laura Belin, "Ultranationalist Parties Follow Disparate Paths," Transition, Vol. 1, No. 1 (June 23, 1995), pp. 8-12; Jerry F. Hough, Evelyn Davidheiser, and Susan Goodrich Lehmann, The 1966 Russian Presidential Election (Washington: Brookings, 1996), esp. pp. 2-13, 45-46, 69-70.

45. Susanne Hoeber Rudolph and Lloyd Rudolph, "Modern Hate," New Republic, Vol. 208, No. 12 (March 22, 1993), pp. 24-29; Milton Esman, Ethnic Politics (Ithaca, N.Y.: Cornell University Press, 1994), pp. 28-31.

46. Paul Keckemeti, "Propaganda," in Pool, Handbook of Communication, pp. 844-870; Donald Kinder and David Sears, "Public Opinion and Political Action," in Gardner Lindzey and Elliot Aronson, eds., Handbook of Social Psychology, Vol. 3 (New York: Random House, 1985), pp. 659-741; Robert Jervis, Perception and Misperception in International Politics (Princeton, N.J.: Princeton University Press, 1976).

47. Benjamin Page and Robert Shapiro, The Rational Public (Chicago: University of Chicago, 1992). Consumers' level of education - and presumably their sophistication about propaganda - seems to have mixed effects. Page and Shapiro, pp. 178, 203-205, 313-330.

48. John Zaller, The Nature and Origins of Mass Opinion (Cambridge, U.K.: Cambridge University Press, 1992).

49. Stegmann, "Between Economic Interests and Radical Nationalism," in Jones and Retallack, Between Reform, p. 170; Fairbairn, "Interpreting Wilhelmine Elections," in Jones and Retallack, Elections, pp. 22-23.

50. Eley, Reshaping the German Right, chap. 11; James Retallack, "The Road to Philippi: The Conservative Party and Bethmann Hollweg's 'Politics of the Diagonal,' 1909-1914," in Jones and Retallack, Between Reform, pp. 286-287.

51. For various approaches to regulating speech and media, see Judith Lichtenberg, ed., Democracy and the Mass Media (Cambridge, U.K.: Cambridge University Press, 1990), esp. pp. 52, 127-128, 144-145, 186-201.

52. Page and Shapiro, The Rational Public, pp. 339-354; for a skeptical view, see Robert Entman, Democracy without Citizens (New York: Oxford University Press, 1989).

53. See the periodic country studies by Article 19 and Freedom House.

54. Phillip Knightly, The First Casualty (New York: Harcourt, Brace, Jovanovich, 1975), pp. 21-25; Michael Schudson, Discovering the News: A Social History (New York: Basic, 1978); Belin, "Russia: Wrestling Political and Financial Repression," pp. 59-63; Lent, Newspapers in Asia, pp. 176, 211.

55. Letter to John Norvell, June 14, 1807, in Merrill D. Peterson, ed., The Portable Thomas Jefferson (New York: Viking, 1975), p. 505.

56. James Manor, "The Failure of Political Integration in Sri Lanka (Ceylon)," Journal of Commonwealth and Comparative Politics, Vol. 17, No. 1 (March 1979), pp. 21-44, esp. 21-23.

57. Thompson, Forging War, p. 20; Zdenka Milivojevic, "The Media in Serbia from 1985 to 1994," in Dusan Janjic, ed., Serbia Between the Past and the Future (Belgrade: Institute of Social Sciences, 1995), pp. 168-169.

58. Velko Vujacic, "Serbian Nationalism, Slobodan Milosevic and the Origins of the Yugoslav War," The Harriman Review, Vol. 8, No. 4 (December 1995), p. 29; Thompson, Forging War, p. 20.

59. Gagnon, "Ethnic Nationalism and International Conflict: The Case of Serbia"; Susan Woodward, Balkan Tragedy (Washington, D.C.: Brookings, 1995), pp. 99, 230-232, 293; Branka Magas, The Destruction of Yugoslavia (London: Verso, 1993), pp. 3-76; Thompson, Forging War, p. 56, 65-66, 114-116, 124.

60. Thompson, Forging War, pp. 6-7, 16.

61. Thompson, Forging War, pp. 38-43; Woodward, Balkan Tragedy, p. 230. In Croatia, only 600,000 people out of ten million get their news from media not controlled by the government, according to a survey conducted by Miklos Biro, "Is Anybody Out There?" War Report, No. 39 (February/March 1996), p. 17.

62. Milivojevic, "The Media in Serbia," p. 164; Thompson, Forging War, p. 54; Silber and Little, Yugoslavia, p. 33; see also Magas, Destruction, pp. 49-76.

63. Gagnon, "Ethnic Nationalism and International Conflict: The Case of Serbia," pp. 145-152; Vujacic, "Serbian Nationalism," p. 30; Thompson, Forging War, pp. 23-24, 52-53.

64. Thompson, Forging War, p. 55.

65. Thompson, Forging War, pp. 221-224; Woodward, Balkan Tragedy, p. 230.

66. Changes in media content were also used successfully to shift opinion in favor of peace. On April 9, 1993, 70 percent of Serbian respondents said they opposed the Vance-Owen peace plan, but on April 27, after a reversal of policy by the Serbian government and media, only 20 percent opposed it, and 39 percent were in favor. Thompson, Forging War, pp. 127-128, 209, 264.

67. Nonetheless, even the independent media found itself caught in the self-fulfilling prophecies generated by nationalist mythmaking. As Serbian journalist Stojan Cerovic said in May 1992, "Anybody who explains the truth can do so only at his own cost. Reality sounds like the blackest anti-Serbian propaganda, and anyone who describes it will frighten people and turn them against him." Thompson, Forging War, pp. 73-75, 127-129. For a dissenting view which stresses the limited success of appeals to Serbian nationalism, see Gagnon, "Ethnic Conflict as Demobilizer."

68. Thompson, Forging War, pp. 127-128.

69. Thompson, Forging War, pp. 105-111, 161.

70. Thompson, Forging War, pp. 225-225, 229-231.

71. Quotation from Africa Rights, Rwanda, p. 35; also pp. vi, 37-38, 63-64, 69-72, 150; Human Rights Watch, Playing the "Communal Card," pp. 7, 9.

72. Holly Burkhalter, "The Question of Genocide," World Policy Journal, Vol. 11, No. 4 (Winter 1994-95), pp. 44-54, esp. 51, 53.

73. Human Rights Watch, Playing, pp. 16-17; Africa Rights, Rwanda, p. 720; Reporters sans Frontieres, Rwanda: L'impasse? La liberte de las presse apres le genocide, 4 juillet 1994-28 aout 1995 (Paris: Reporters sans Frontieres, 1995), pp. 48-50; Alison Des Forges, "The Rwandan Crisis," paper prepared for a conference on Sources of Conflict in Rwanda (Washington, D.C.: U.S. Department of State, October 17, 1994).

74. Compare Africa Rights, Rwanda, pp. 32-34 with 720; also compare Des Forges, "Rwandan Crisis," pp. 1 with 9.

75. Africa Rights, Rwanda, p. 150.

76. Prunier, The Rwanda Crisis, chap. 4, "Slouching towards Democracy," esp. pp. 131-133, 157, on the low quality and extremism of these new entrants into public discourse.

77. Africa Rights, Rwanda, pp. 30-34, 44; Bruce D. Jones, "The Arusha Process," in Howard Adelman and Astri Suhrke, eds., Early Warning and Conflict Management in Rwanda (forthcoming); Alan J. Kuperman, "The Other Lesson of Rwanda: Mediators Sometimes Do More Damage Than Good," SAIS Review, Vol. 16, No. 1 (Winter-Spring 1996), pp. 221-240.

78. Alex de Waal, "The Genocidal State," Times Literary Supplement, July 1, 1994, pp. 3-4; see also Africa Rights, Rwanda, pp. 30-32.

79. Africa Rights, Rwanda, p. v; also pp. 568-596; Prunier, The Rwanda Crisis, p. 170; Jones, "Arusha."

80. Africa Rights, Rwanda, pp. 66-68.

81. De Waal, "The Genocidal State," p. 4; also Jones, "Arusha." On Habyarimana's death, see Prunier, Rwanda Crisis, pp. 213-229.

82. Human Rights Watch, Playing, pp. 16-17.

83. Reporters sans Frontieres, Rwanda, pp. 6, 41-42; 52-53; Reporters sans Frontieres, Burundi, le Venin de la Haine: Etude sur les medias extremistes, 2d ed. (Paris: Reporters sans Frontieres, July 1995), pp. 63, 68-69; quotation on p. 69.

84. Tim Harris, "Propaganda and Public Opinion in Seventeenth-Century England," in Jeremy Popkin, ed., Media and Revolution (Lexington: University Press of Kentucky, 1995), pp. 48-73.

85. Geoffrey Holmes and Daniel Szechi, The Age of Oligarchy (London: Longman, 1993), chap. 13, esp. pp. 194-195; Martin Daunton, Progress and Poverty (Oxford: Oxford University Press, 1995), p. 482; Linda Colley, Britons (New Haven, Conn.: Yale University Press, 1992), esp. pp. 25, 220.

86. Snyder, Myths of Empire, chap. 5; Paul Kennedy, Strategy and Diplomacy (London: Allen and Unwin, 1983), chaps. 1 and 8; for a more critical view, see Charles Kupchan, The Vulnerability of Empire (Ithaca, N.Y.: Cornell University Press, 1994), chap. 3.

87. Elaine Potter, The Press as Opposition: The Political Role of South African Newspapers (Totowa, N.J.: Rowman and Littlefield, 1975).

88. On the segmented market, see William Finnegan, Dateline Soweto (New York: Harper and Row, 1988), pp. 24-33, and Potter, The Press as Opposition, p. 164; on the transition, see Timothy Sisk, Democratization in South Africa (Princeton, N.J.: Princeton University Press, 1995).

89. S.N. Paul, Public Opinion and British Rule (New Delhi: Metropolitan, 1979).

90. R.C.S. Sarkar, The Press in India (New Delhi: S. Chand, 1984), pp. 190-193, 295-296; Allyn and Wilkinson, Guidelines for Journalists, appendix.

91. Moti Lal Bhargava, Role of Press in the Freedom Movement (New Delhi: Reliance, 1987), pp. 336, 341-342.

92. S.B. Kolpe, "Caste and Communal Violence and the Role of the Press," pp. 342, 349, in Asghar Ali Engineer, ed., Communal Riots in Post-Independence India (Hyderabad: Sangam, 1984); and Asghar Ali Engineer, "The Causes of Communal Riots," ibid., pp. 36-38; Hamish McDonald, "Paper Tigers," Far Eastern Economic Review, Vol. 158, No. 40 (October 5, 1995), pp. 28-30; Zenab Banu, Politics of Communalism (Bombay: Popular Prakashan, 1989), p. 21.

93. Sumit Ganguly, Between War and Peace: The Crisis in Kashmir (Cambridge, U.K.: Cambridge University Press, forthcoming); Ganguly, "Explaining the Kashmir Insurgency: Political Mobilization and Institutional Decay," International Security, Vol. 21, No. 2 (Fall 1996), pp. 76-107.

94. Human Rights Watch, Playing the "Communal Card," p. 21.

95. Article 19, Guidelines for Election Broadcasting, p. 5.

96. Michael W. Doyle, UN Peacekeeping in Cambodia: UNTAC's Civil Mandate (Boulder, Colo.: Lynne Rienner, 1995), pp. 54-55. For other cases, see Dan Lindley, "Collective Security Organizations and Internal Conflict," in Michael E. Brown, ed., The International Dimensions of Internal Conflict (Cambridge, Mass.: MIT Press, 1996), pp. 562-567.

97. Thomas Carothers, Assessing Democratic Assistance: The Case of Romania (Washington, D.C.: Carnegie Endowment for International Peace, 1996), pp. 80-89.

98. Richard L. Merritt, Democracy Imposed: U.S. Occupation Policy and the German Public, 1945-1949 (New Haven, Conn.: Yale University Press, 1996), pp. 291-315, esp. 296, emphasizes the effectiveness of this strategy.

99. Horowitz, A Democratic South Africa? chaps. 4 and 5; Arend Lijphart, Democracy in Plural Societies (New Haven, Conn.: Yale University Press, 1977). On common media as a precondition for an integrated national consciousness, see Benedict Anderson, Imagined Communities (London: Verso, 1983).

100. Stephen Harold Riggins, Ethnic Minority Media (Newbury Park, Calif.: Sage, 1992).

101. Larry Diamond, Promoting Democracy in the 1990s (New York: Carnegie Corporation, Report to the Carnegie Commission on Preventing Deadly Conflict, December 1995), pp. 24-25.

102. Gordon P. Means, Malaysian Politics: The Second Generation, 2d ed. (Oxford: Oxford University Press, 1991), esp. pp. 137-138, 313-314, 418-422, 439; Karl von Vorys, Democracy without Consensus: Communalism and Political Stability in Malaysia (Princeton, N.J.: Princeton University Press, 1975), pp. 394-412.

Jack Snyder is Professor of Political Science and Director of the Institute of War and Peace Studies, Columbia University. Karen Ballentine is a Ph.D. candidate in political science at Columbia University and in 1996-97 will be a Research Fellow at the Center for Science in International Affairs, Harvard University.

International Security, 1996, 21 (2)