² Presumably the salience of compromise in Belgian public life can be traced to the fact that major social decisions, such as the content of the programme of a national government or the pluri-annual deal between trade unions and employers have to be made jointly by people and organizations who have previously made, publicly and often forcefully, mutually incompatible demands. However, compromise politics is by no means confined to countries with proportional representation (and hence coalition governments) or corporatist institutions. It is, for example, sometimes identified as a core feature of social democracy (see Alain Bergounioux and Bernard Manin, La Social-démocratie ou le compromis, Paris, P.U.F., 1979).

³ Narrower definitions are possible, such as those proposed by Bellamy, Richard and Hollis, Martin, ‘Consensus, Neutrality and Compromise’, Critical Review of International Social and Political Philosophy, 1: 3 (1998), pp. 54–78 CrossRefGoogle Scholar, at pp. 54–5, 62–3, which distinguishes a compromise (by negotiators) from a bargain (by traders); and by Dominique Leydet, ‘Pluralisme et compromis’, in M. de Nanteuil and M. Nachi (eds), Éloge du compromis. Pour une nouvelle pratique démocratique, Louvain-la-Neuve, Academia Bruylant, 2005, pp. 81–106, at pp. 84–8, which also distinguishes between compromise (positively connoted) and bargaining (negatively connoted). In the present context, however, it seems better to see this as a basis for a hypothesis (discussed below) as to what constitutes a good compromise. Other definitions of compromise include a process of negotiation through mutual concessions, rather than the agreement that this process produces (see Mohamed Nachi, ‘Esquisse d’une théorie du compromis', in Nanteuil and Nachi, Éloge du compromis, pp. 143–72, at pp. 154–5. A definition is not true or false, but more or less useful. The one adopted for this article is offered as the most useful for my purposes.

⁴ In other words, the terms of a feasible compromise must be such that each party feels (ex ante) better off under the compromise than under its best alternative to a negotiated agreement (BATNA), see Roger Fisher and Bill Ury, Getting To Yes: Negotiating Agreement Without Giving In, Boston, MA, Houghton Mifflin, 1981, and the subsequent literature in negotiation theory. What a compromise enables a party to do is to avoid the loss (in terms of whatever it cares about) that corresponds to the difference between what it gets under the compromise and its BATNA.

⁵ In the interesting analysis of Avishai Margalit, On Compromise and Rotten Compromises, Princeton and Oxford, Princeton University Press, 2010, the focus is on this first case: compromises that avoid conflict, violence or war. The second case plays only a minor role, and the third one is ignored.

⁶ The third-party arbitrator, who decides in the absence of agreement, must be distinguished from the third-party mediator, who assists the parties in finding a compromise. The person mandated to assist a company's management and its workers to reach an agreement that would avoid or end a strike, is a third-party mediator. The government that threatens to legislate if the social partners fail to reach an agreement is a third-party arbitrator.

⁷ The balance of power thus determined affects both the possibility and the contents of the compromise as far as the extent the parties are aware of it. But the parties may also have an inaccurate idea of what may happen in the absence of an agreement, and unrealistic expectations can both facilitate the reaching of a compromise and make it very difficult. If expectations on either side are overly optimistic about what will happen in case of conflict, exit or arbitration, compromise may prove elusive. If both sides are overly pessimistic, on the other hand, it should generally be quite easy to reach an agreement. Margalit, On Compromise, p. 51, argues that exaggerated optimism is one of the reasons why war prevails over a compromise that would have saved huge costs for both parties. Even with realistic expectations, however, war will prevail either because at least one party does not sufficiently trust that the other will respect the terms of the agreement, or because any compromise would necessarily involve the sacrifice of something so valuable to at least one party that any probability above zero of obtaining it by war justifies the refusal to compromise.

⁸ In order to characterize concessions, Margalit, On Compromise, pp. 47–50, refers to the dream points of the two parties. However, this reference fails to distinguish sufficiently between what the parties would like and what they feel entitled to.

⁹ This distinguishes my purpose here from Avishai Margalit's. Whereas I want to identify the difference between good and bad compromises, he wants to characterize ‘rotten’(as opposed to acceptable) compromises. A rotten compromise, he proposes, is one that is both humiliating and cruel (On Compromise, pp. 54, 62). Not all acceptable compromises are good, and not all bad compromises are rotten.

¹⁰ Whether a compromise is honourable or not in this sense is a matter of plausible presentation to the outside world. It is therefore very sensitive to a specific subsequent cooperation between the two parties: in order to save its own face, each party must downplay its concessions, but in order not to threaten the other party's face, it must not downplay them too much. To the extent that each party's declarations reaches a common audience, triumphalism must be avoided. The very feasibility of a compromise therefore depends on the negotiators managing to be able to trust each other sufficiently not to run the risk of being publicly humiliated by the other party's triumphalist exhibition of the concessions they made.

¹¹ An honourable compromise in this sense corresponds to what Leydet, ‘Pluralisme et compromis’, p. 103, simply calls ‘compromise’ in a narrower sense than the one adopted here: it results, by definition, from a process of negotiation that is halfway between deliberation (leading to a robust consensus) and bargaining (leading to a precarious modus vivendi).

¹² The use of financial compensation to achieve a (good?) compromise is obviously nothing new. Emmanuelle Tixier du Mesnil, ‘L'exemple andalou: le compromis érigé en mode de gouvernement’, in M. Nachi (ed.), Les sociétés islamiques sont-elles hostiles au compromis? Paris, Karthala, 2011, pp. 113–25, shows, for example, how in the fifteenth century, the Sultan of Granada, Abdullah, made this a centrepiece of his statecraft.

¹³ See the discussion of compromise as multi-dimensional ‘trading’ in Richard Bellamy, Liberalism and Pluralism. Towards a Politics of Compromise, London, Routledge, 1999, pp. 103–5.

¹⁴ In particular, adequate mutual understanding and a joint exploration of the consequences of different possible compromises goes beyond the mutual respect required for an honourable compromise. In the Belgian context, the development of mutually beneficial compromises has been adversely affected by the vanishing of intellectual interaction between members of the two communities (following the splitting up of the national political parties and the devolution of all competences in matters of education and research to the linguistically defined communities) and the virtual disappearance, therefore, of any common reflection on the consequences of different reform options for each of the two communities. The Re-Bel initiative (http://www.rethinkingbelgium.eu) aims to remedy this situation and thereby to facilitate the emergence of win-win reforms. In other places, where the conflict is more acute, the lack of intellectual interaction is even more striking. During a visit to Jerusalem in April 2010, I discovered that the colleagues I met at lectures at Hebrew University and Al Quds University had never set foot on each other's campus, although they are but a few kilometres apart. One month later, when attending a conference on ‘bi-nationalism’ organized by the Israel Democracy Institute, I was dismayed to discover than not one Arab featured among the participants.

¹⁵ The British historian Henry Kamen, The Duke of Alba, New Haven, CT, Yale University Press, 2004, p. 134, identifies what makes the Pacification of Ghent unprecedented as follows: ‘nowhere in Europe had any state gone so far as to permit freedom of belief to subjects. The most that had been done, for example, was to permit a limited freedom to certain nobility and the regions they controlled.’ The Edict of Nantes was not published until 1598. Note, however, that in the same period, the Mogul emperor, Akbar, could be considered to have gone further: see Amartya Sen, The Idea of Justice, Cambridge, MA, Harvard University Press, 2009, pp. 36–9.

¹⁶ For a critical perspective, see W. P. Blockmans and P. Van Peteghem, ‘La Pacification de Gand à la lumière d'un siècle de continuité constitutionnelle dans les Pays-Bas’, in R. Vierhaus (ed.), Herrschaftsverträge, Wahlkapitulationen, Fundamentalgesetze, Göttingen, Vandenhoeck & Rupprecht, 1977, pp. 220–34.

¹⁷ See Hans Cools, ‘De Pacificatie van Gent’, in J. Tollebeek and H. te Velde (eds), Het geheugen van de Lage Landen, Rekkem, Ons Erfdeel, 2009, pp. 19–25, at pp. 19–22.

¹⁸ John Rawls, Justice as Fairness: A Restatement, Cambridge, MA, Harvard University Press, 2001, pp. 192–5, stresses that this stability is an essential difference between what he calls an overlapping consensus (which allows for a shared conception of justice) and what he calls a modus vivendi (which corresponds, when it is the subject of an explicit agreement, to a compromise in the sense defined here).

¹⁹ See the notion of ‘burdens of judgement’, as introduced by John Rawls, Political Liberalism, New York, Columbia University Press, 1993, pp. 54–8; Rawls, Justice as Fairness, pp. 35–7; and discussed, for example, by Richard Bellamy, Liberalism and Pluralism, chapter 2.