Skip to main content Accessibility help

The Security Dilemma in Alliance Politics

  • Glenn H. Snyder (a1)

The concept of the “security dilemma” is applied to alliance relations in multipolar and bipolar systems. The dilemma involves a choice between support or nonsupport of allies, and tension between fears of entrapment and abandonment. It interacts with the adversary security dilemma in which the choice is between firmness and conciliation toward the opponent. The multipolar interaction is illustrated by a survey of the 1904–1914 period, the bipolar by reference to the contemporary crisis in NATO. The alliance security dilemma is more severe, and places more constraints on allies' policies toward adversaries, in multipolar than in bipolar alliances. The weakness of the dilemma in the contemporary system is a major reason for the current persistence of conflict in NATO.

Hide All

1 The best recent treatment is Jervis, Robert, “Cooperation under the Security Dilemma,” World Politics 30 (January 1978), 167214. The concept of the security dilemma was originated by Herz, John, in his Political Realism and Political Idealism (Chicago: University of Chicago Press, 1951).

2 This is, of course, an idealized model based on certain assumptions from which the empirical world will deviate more or less, from time to time. The basic assumptions are that: (1) no state is aggressive, but none can know the intentions of others; (2) the states are roughly equal in military strength; and (3) military technology is such that there is no time to form a successful defense alliance after war begins. Uncertainty about the aims of others is inherent in structural anarchy. If a state clearly reveals itself as expansionist, however, the alliance that forms against it is not “self-defeating” as in the prisoners' dilemma (security dilemma) model. Or, if some states are weaker than others, their motives to ally will be different from the incentives of the prisoner's dilemma. The third assumption has been valid since about 1870. Before then (when the pace of warfare was slower), the compulsion to ally in peacetime was much weaker than suggested by the model. Despite these qualifications and possibly others, the model does capture some essential dynamics of multipolar alliance formation between 1870 and 1939.

3 “Being in” the most powerful coalition does not necessarily mean that states join the most powerful coalition that is already in existence. Indeed, they will more likely join the weaker one which then becomes the most powerful as a consequence of their joining, because this gives them leverage to bargain for a maximum share of the alliance's payoff. Thus the logic of N-person game theory is consistent with Waltz's argument that states “balance” rather than “bandwagon.” See Waltz, Kenneth N., Theory of International Politics (Reading, Mass.: Addison Wesley, 1979), 125–26.

4 Despite the importance of internal politics, this reference will be the only one in this essay. For reasons of theoretical parsimony and space limitations, the analysis is based entirely on what in recently popular academic terminology is called the “rational actor model,” the actors being states.

5 The concepts of abandonment and entrapment were first posited, I believe, by Mandelbaum, Michael, in The Nuclear Revolution: International Politics Before and After Hiroshima (New York: Cambridge University Press, 1981), chap. 6.

6 The adversary dilemma presented here is a secondary security dilemma, analogous to the secondary alliance dilemma. In both cases, the primary security dilemma, which is a prisoner's dilemma, has already been resolved by mutual defection. That is, alliances have formed and adversaries have adopted a general posture of power/security rivalry in the “DD” cell of the primary game, which we might also call the international “supergame.” Once adversaries are in this cell they may be able to reduce their conflict; on the other hand, they may sink deeper into conflict and competition. That is, having protected themselves against the worst, they are now able to consider whether they might not improve their situation by conciliating the opponent—although they must also guard against exploitation of such cooperation. Whether conciliation or continued (or greater) firmness is the better policy for any state will depend on its adversary's preferences—which the state does not know, although presumably it knows its own. The adversary dilemma in Table 1 simply assumes the extremes—the opponent is either status-quo-oriented or expansionist—even though he may have mixed motives or be expansionist in different degree in different situations. Formally speaking, adversaries play a series of sub-games within the general context of supergame “DD.” Some of these may be prisoner's dilemmas, but some may be other games that are either more or less conflictual than the prisoner's dilemma. The dilemma for the state, then, arises from its uncertainty as to what game is being played, which in turn stems from its uncertainty about the adversary's preferences. Choosing the best strategy thus requires an estimate of the opponent's preference rankings for possible outcomes. Space forbids a discussion of the various possible sub-games: in Table 1, they are implicitly “Stag Hunt” (when the opponent prefers the status quo) and a variant of “Chicken” (when the opponent is expansionist but prefers peace over expansion by war). For a comprehensive treatment, see Snyder, Glenn H. and Diesing, Paul, Conflict Among Nations (Princeton: Princeton University Press, 1977), chap. 2.

The secondary adversary dilemma is similar to the dichotomy between the axioms of deterrence theory and those of the “spiral model” as described by Jervis, Robert in Perception and Misperception in International Politics (Princeton: Princeton University Press, 1976), chap. 3. Jervis makes the point that the choice between deterrence and spiral-model axioms—between firmness and accommodation—depends essentially on one's estimate of the adversary's ultimate aims. Jervis does not apply the term “security dilemma” to this broad choice, but uses it in its traditional sense as an explanation of why the search for security among status-quo states may be self-defeating—i.e., as the structural basis of the spiral model.

7 There are some interesting analogies between the dynamics of the two games. A strategy of conciliation in the adversary game may produce a “falling domino” effect: the opponent interprets one's overture as weakness and pushes harder on both present and future issues. The alliance analogue is the entrapment effect. As an opponent may be emboldened if he is appeased, the ally may become more intransigent and aggressive if he is supported. The alliance analogue to deterrence of the opponent is restraint of the ally. Deterrence involves a threat of force against an adversary; restraint may be accomplished by threatening not to use force in support of the ally. In the alliance game, strengthening one's commitment to the ally tends to foreclose one's alternative alliance options; in the adversary game, firming up one's commitment against the opponent may foreclose compromise settlement options. Bargaining power over the ally is enhanced by a weak, ambiguous commitment; leverage over the adversary is strengthened by a firm, explicit one.

8 “Conflict” and “tension” are analytically separable. Conflict is incompatibility of interest; tension may be defined crudely as the felt likelihood that the conflict will produce war in the immediate future. A high degree of conflict may be accompanied by low tension; the opposite is also possible, though less likely. The two are linked here because of their similar effect on alliance dependence.

9 Austria also had a strategic interest in Germany's not being absorbed by Russia, but this interest was abstract—i.e., non-operational—since Germany was quite capable of holding her own against Russia, provided Russia had not previously conquered Austria. Thus, the situation was analogous to the Franco-British relationship. The general point is that the stronger ally will be less directly dependent than its partner but more indirectly dependent. Asymmetry of this kind is especially pronounced in a bipolar world because of the wide disparity of power between the superpowers and their allies. Thus the European members of NATO are highly dependent on the U.S. in the direct sense, but the U.S. is not similarly dependent on them; on the other hand, the U.S. is indirectly dependent on its allies because of its strategic interest in containing Soviet power.

10 The best treatment of this series of crises, beginning with Morocco (1905) is still that of Albertini, Luigi, The Origins of the War of 1914 (London: Oxford University Press, 1952), Vol. I.

11 The term “abandonment” is used here in the sense of reneging on one's specific alliance obligation, not as realignment or de-alignment. On Franco-Russian relations between 1912 and 1914 and the strengthening of the French commitment, see Gooch, G. P., Before the War: Studies in Diplomacy (London: Longmans, Green, 1938), chap. 2.

12 On Anglo-French-German relations during this period, see K. A. Hamilton, “Great Britain and France, 1905–1911,” and Sweet, D. W., “Great Britain and Germany, 1905–1911,” in Hinsley, F. H., ed., British Foreign Policy Under Sir Edward Grey (Cambridge: At the University Press, 1977), chaps. 5 and 11.

13 What I call the insecurity spiral is sometimes referred to as the security dilemma. I believe this is a mislabeling, however, since the spiral is an outcome of both adversaries' choosing one of the options in the security dilemma, not the problem of choice itself. The defining feature of any dilemma is the difficulty of choosing between two options, each of which will have more or less unsatisfactory consequences.

14 Monger, George, The End of Isolation (London: Nelson, 1963), chap. 12.

15 Schmitt, Bernadotte, The Annexation of Bosnia (Cambridge: At the University Press, 1937).

16 Barlow, Ima C., The Agadir Crisis (Chapel Hill: University of North Carolina Press, 1940).

17 Williamson, Samuel R. Jr, The Politics of Grand Strategy (Cambridge: Harvard University Press, 1969), provides an excellent account of these French-British negotiations; see chaps. 11 and 12.

18 It is interesting that France tightened her alliance with Russia by increasing her own commitment, but strengthened her British connection by extracting a firmer commitment from England. The explanation for the difference is that the “balance of dependence” between France and Russia favored Russia, whereas between France and England, it favored France. In other words, France enjoyed more bargaining power over England than over Russia.

19 The classic study of the Balkan Wars is Helmreich, E. C., The Diplomacy of the Balkan Wars (Cambridge: Harvard University Press, 1938). For a more recent interpretation emphasizing Anglo-German relations, see Crampton, R. J., The Hollow Detente (London: George Prior Publishers, 1979).

20 Historians are divided as to whether Grey could have averted the war by taking a clear stance one way or the other. Albertini's opinion (fn. 10, Vol. II, 514), is that he could have. For a contrary view, see Taylor, A.J.P., The Struggle for Mastery in Europe (London: Oxford University Press, 1954), 525. For a discussion of Grey's dilemma, see Eckstein, Michael J., “Great Britain and the Triple Entente on the Eve of the Sarajevo Crisis,” in Hinsley (fn. 12), chap. 18.

21 It is true that Grey felt constrained by domestic and constitutional considerations from issuing an unequivocal commitment. The fact that eventually he did issue a clear warning to Germany that England could not remain neutral—although it came too late to affect German decision making—indicates that these constraints were not absolute.

22 These remarks can no more than suggest the complexity of the straddle problem. One aspect of this complexity is that one is trying to optimize among four objectives: restrain the ally, but avoid alienating him; deter the opponent but avoid provoking him. Just what mixture of communications is likely to optimize will be difficult to estimate, even if one could assume that they will be interpreted as desired by both other parties. Another aspect is the necessary ambiguity of one's messages, which makes it unlikely that they will be interpreted exactly as desired. The possible porosity of diplomatic communication channels limits the extent to which signals sent to the ally and adversary can be inconsistent with each other—e.g., signals to the ally that lean toward restraint and messages to the opponent that emphasize firmness. The “pure” and consistent strategies of restraining the ally while conciliating the adversary, or supporting the ally while warning the opponent, are less likely to be misread, and either one may well keep the peace, but they risk alienating the ally or increasing the opponent's hostility. By minimizing these latter risks, a mixed strategy promises a better outcome than either of the pure ones, but also risks a worse outcome, especially when the background conditions approach those in 1914.

23 I believe the present system should still be classified as bipolar, even though there has been some movement toward multipolarity. Although some of my theoretical statements in this section apply to bipolarity in Europe in general, they are much more relevant to NATO than to the Warsaw Pact, as my examples clearly indicate. Bipolar alliances outside Europe have somewhat different dynamics, which cannot be explored here.

24 Some might argue that the West European countries are allied with the United States out of cultural and ideological affinities (and disaffinities with the Soviet Union), not because of structural compulsion. However, even if these affinities did not exist, alignment with the U.S. would be dictated by their security interests.

25 For a similar argument, see Waltz, Kenneth N., Theory of International Politics (Reading, Mass.: Addison Wesley, 1979), esp. 169–70. This essay has been considerably influenced by Waltz's impressive work.

26 Kissinger, , The White House Years (Boston: Little, Brown, 1979), 382.

27 Ibid., 382.

28 Kissinger, , Years of Upheaval (Boston: Little, Brown, 1980), 145–46.

29 Brandt, Willy, Begegnungen und Einsichten. Die Jahre 1960–1975 (Hamburg: Hoffmann & Campe, 1976), 348, cited in Kaiser, Karl and Schwarz, Hans-Peter, America and Western Europe: Problems and Prospects (Lexington, Mass.: Lexington Books, 1977), 229.

30 Conversely, it ought to be easier for the bipolar superpowers than for multipolar adversaries to dampen or avoid insecurity spirals by conciliating each other, since their conciliation is so much less constrained by alliance concerns. In the pre-1914 decade, Sir Edward Grey empathized with Germany's insecurities and fears of encirclement; he realized that the steady tightening of the Entente was provoking her, but he could do little about it because he feared abandonment by Russia and France if he tried. The U.S. and the Soviet Union do not have a similar problem. This is not to say that it is easy to escape from or weaken insecurity spirals in a bipolar system—it is merely easier because the attempt involves significant risks in the adversary game only.

31 The terms “European” and “American” are intended to refer to the apparently dominant views among European and American elites. Obviously, there is a wide spectrum of opinion on both sides of the Atlantic.

32 The 1914 analogy is poorly taken, since that spiral fed on factors peculiar to a multipolar system—notably high and equal interdependence among allies. Also, recent historiography has suggested that World War I was a war that “somebody” (Germany) did want. See, for example, Fischer, Fritz, Germany's Aims in the First World War (New York: W. W. Norton, 1967).

33 A similar point is made by Art, Robert J. in “Fixing Atlantic Bridges,” Foreign Policy, No. 46 (Spring 1982), 77.

34 These remarks are, in effect, an elaboration of Kenneth Waltz's axiom that flexibility of alignment in multipolarity produces rigidity of strategy, while rigidity of alignment in bipolarity promotes flexibility of strategy. Waltz (fn. 25).

35 Tucker, Robert W. and Wrigley, Linda, eds., The Atlantic Alliance and Its Critics (New York: Praeger, 1983), chap. 6.

* I wish to thank Robert Art, David Goldfischer, Robert Jervis, Robert Osgood, Dean Pruitt, Jack Snyder, and Kenneth Waltz, for helpful comments; and the Woodrow Wilson International Center for Scholars and the Institute of War and Peace Studies, Columbia University, for financial support and office space.

Recommend this journal

Email your librarian or administrator to recommend adding this journal to your organisation's collection.

World Politics
  • ISSN: 0043-8871
  • EISSN: 1086-3338
  • URL: /core/journals/world-politics
Please enter your name
Please enter a valid email address
Who would you like to send this to? *


Altmetric attention score

Full text views

Total number of HTML views: 0
Total number of PDF views: 0 *
Loading metrics...

Abstract views

Total abstract views: 0 *
Loading metrics...

* Views captured on Cambridge Core between <date>. This data will be updated every 24 hours.

Usage data cannot currently be displayed