INTERNATIONAL RELATIONS

II [: 1. '

' • II I

'I' ':! I

i , I , II

I II I I•

1 ,, Iii 'II

i!

l l i I

I: I•

!·'

I' I

II .'' I,

:'I II 1

Lt

?i~

ONE

Realism, Neorealism and the Study of World Politics

ROBERT 0. K EOHAN E

W ORLD POLITICS today is a matter of life and death-not just for soldiers or citizens caught in the path of war, but for the whole human race. Nuclear holocaust remains a continual threat. At a

less apocalyptic level, world politics has a daily impact on the lives of

people throughout the globe: political forces and decisions affect patterns

of international trade , investme nt, and production. Whether OPEC oil

will be cheap or dear; whether China will export textiles to the Unite d

States; how many Americans will work in the auto industry: all depend

as muc h on political dec isions as on patterns of comparative costs and effici encies.

The complexities of security in a nuclear age, and of economic viability

in an era of interdependence and rapid technological change, have created

dema r\ds by policymakers for expertise on a myriad of international po-

litical issues. Thousands of experts, in and out of governme nts around

the world, analyze the relative military forces of the superpowers, fluc-

tuations in supply and demand in oil markets, or shifts in international

patterns of comparative advantage . These observers interpret the signif-

icance of c hanges in policies of governments and nongovernmental actors

such as transnational corporations and international organizations, and

they seek to de termine their impact-on Soviet or American security,

Saudi or Mexican oil revenues, inflation in the United States, or the

viability of high-tech industry in Europe.

Little of this immense application of brainpower to world affairs takes

any explicit account of theories about world politics. Indeed, a nai"ve

Note: For comme nts on earlier drafts of this essay I am indebted to Nannerl 0. Keohane, Andy Moravcs ik, Jose ph S. Nye , John Gerard Ruggie , and Howard Silverman.

I I I

I

I I

I :i1· i 1 1,1 ·I

:I!

I i~

'Iii I! II

'I I il

!\, I II

ill 'I I' ·'

2 Robert 0. Keohane

observer might conclude that all of the relevant theories come from other

. , disciplines: physics, laser engineering, petroleum geology, electronics, the

economics of comparative advantage or exchange-rate determination.

Foundations and governments display little interest in supporting theo-

retical work in international relations, compared with their zest for

"policy-relevant" research or the incorporation into international relations

of findings from other fields.

In view of this widespread lack of interest, the reader may well ask:

"Why should I care about theoretical debates among scholars of inter-

national relations? If I want to understand world politics or make policy,

shouldn't I devote my efforts directly to studying nuclear deterrence,

OPEC, or the international financial system?" In other words, is the

enterprise represented by this volume-abstract discussion of theoretical

issues in international relations-relevant to the practical tasks of inter-

preting the world and seeking to change it? Does the theory of inter-

national relations have implications for practice? Should it be studied by

practitioners or by those who see k to become policymakers in the future?

The contributors to this volume believe that theory does have impli-

cations for practice and that it should be studied by those who see k to

influence events. In the first section of this introductory essay I present

my reasons for holding su ch belief'>. After making my argument, I will

turn to the issues rai sed by the tradition of political realism and its

contemporary offshoot, "neorealism;' and then conclude with a brief

overview of the contributions that constitute this volume.

I. THEORY AND PRACTICE IN INTERNATIONAL RELATIONS

One reason for an aspmng pract1t1oner to learn international relations

theory would be to absorb valuable maxims or propositions that would

prove useful in specific situations . If the conditions for applicability of

these maxims are sufficiently well understood, and if the practitioner

knows enough history to be able to place them in context, they can

furnish useful guides to the interpretation of events. Even a limited, partial

theory-with only a few propositions and a number of interpretive

guides- can be useful. For instance, the r ealist theory of the balance of

power, discussed by Waltz, could have alerted American policymakers in ' .

NEOREALISM AND WORLD POLITICS 3

the 1950s (who were excessively imbued with an ideological view of

world politics) to the likelihood of an eventual Sino-Soviet split. Realist

maxims would have couns eled the United States to be in a position to

make an alliance, or at least an accommodation, when feasible, with the

weaker Chinese to counterhalance the Soviet Union-as Henry Kissinger

and Richard Nixon eventually did.

Theory can therefore be useful: it is often better than unconsc ious

adherence to the prejudices of the day. Nevertheless, it is necessary for

any practitioner using international relations theory to remain skeptical;

indeed, the more seriously the maxims are take n, the more important is

the task of critical analysis . If the maxims apply only under certain

conditions, or if th e theory underlying them is fundamentally erroneous

in its understanding of the forces affecting cooperation and discord, peace

and war, they will be dangerously misleading. For example, the popular

maxim in Western foreign policy after World War II, that appeasem ent

(making concessions to others' demands) should be avoided since it breeds

aggression, is by no means universally valid. Whether making concessions

to others' demands breeds aggression depends, among other things, on

the nature of the demands, the capahilities of the demander, and the

willingness of the responding power to use force effec tively, in a sustaine d

way, as an alternative to making con cessions . A policy of appeasem ent,

disastrous whe n pursued by Britain and France toward Hitler, would not

necessarily have bee n inappropriate for those two countries in the ir

relations with Egypt in the mid-19 50s or even for the United States in

its relations with North Vietnam in the 1960s.

Practitioners cannot judge the validity of theories offered to them, or

the conditions under which theoretical maxims might apply, without

studying theory. It may appear, however, that practitioners could avoid

the pitfalls of misguided or misapplied theory simply by shunning theory

altogether; and indeed, most commentators on international relations

avoid explicit discussion of theories of world politics. Yet to purge oneself

of all traces of theory would be impossible, since even our intuitions about world politics are deeply affected by how the subject has been

thought about in the past. For over 2000 years, thinkers have sought to

understand, more or less systematically, the most basic questions of world

politics: the sources of discord and of war and the conditions of coop-

eration and of peace. As Keynes said in another context, practitioners

I

II;_

::1

li·:

il'l ~~~~·, !:I !

lit

'I: fill

1

!. ·, 1 !I ,I I

!1:1 i I I

Ill!

1lr

il!l' I

if l'i

4 Robert 0. Keohane

are prisoners of "academic scribblers;' whose views of reality profoundly

'affect the contemporary actions of practical people. The choice for prac-

titioners is not between being influenced by theory or examining each

case "on its merits": it is rather between being aware of the theoretical

basis for one's interpretation and action, and being unaware of it.

Even if one could eradicate theory from one 's mind, it would be self-

defeating to try. No one can cope with the complexities of world politics

without the aid either of a theory or of implicit assumptions and prop-

ositions that substitute, however poorly, for theory. Reality has to be

ordered into categories, and relationships drawn between events. To prefer

atheoretical activity over theoretically informed practice would be to seek

to be guided by an unexamined jumble of pre judices , yielding conclusions

that may not logically follow from the assumptions, rather than by beliefs

based on systematic attempts to specify one's assumptions and to derive

and test propositions.

The inescapability of theory in studying world politics suggests a

second reason for exploring what ar~ labeled here political realism and neorealism. Whatever one's conclusion about the value of contemporary

neor ealism for the analysis of world politics in our time, it is important

to understand realism and neorealism because of their widespread ac-

ceptance in contemporary scholarship and in policy circles. Politica l re-

alism is deeply embedded in Western thought. Without understanding

it, we can neither understand nor critic ize our own tradition of thinking

about international relations. Nor could we hope to change either our

thinking or our practice. All people who are interested in having a

sustained profess ional impact on world affairs should study international

relations theory at some time, if only to examine prevailing assumptions

and evaluate the basic propositions that they might otherwise take for

granted.

The danger that one will become the prisoner of unstated assumptions

is rendered particularly acute by the value -laden nature of international

relations theory. This does not mean that observers simply see what they

want to see: on the contrary, virtually all serious students of world politics

view it as a highly imperfect realm of action in which wrongdoing is

common and unimaginable evil is threatened. Nevertheless, it is hard to

. doubt that scholars' values, and their own personal experiences and

temperaments, affect which aspects of world politics they emphasize and

NEOREALISM AND WORLD POLITICS 5

how strenuously, or desperately, they search for ways to mitigate the

injustice and cruelty that is so evident. For this reason, as well as those

mentioned above, interpretations of world politics need to be scrutinized

critically-a task for which a certain degree of sophistication about the

subject is essential. ·

An obj ection could be raised to the above argument. If a theory

provides sufficiently accurate guidance about cause-effect relationships,

and if its propositions about these relationships remain valid over time

and under differ ent conditions, practitioners may not need to study it

deeply. They can learn its major theorems without being too concerned

about how they were derived, or about the range of their theore tical

application. A structural engineer, engaged in building bridges , can as-

sume the validity of Newtonian physics and can apply its propositions

without be ing able to derive them from basic premises. It is of no concern

to the engineer that this same assumption would be terribly mislead ing

for a scientist trying to understand black holes in outer space or the

behavior of quarks at the subatomic level. Similarly, if the theories of

world politics on which policymakers and commentators rely provided

powerful, value -free explanations of outcomes in world politics, which

wer e sure to remain valid throughout the time -horizon of policymakers,

it would be unnecessary for interpreters of contemporary world politics

to concern themselves with theor etical controversies . The methodological

presuppositions of international political theories would be of no more

concern to practitioners than are investigations of quarks or black holes

to the bridge -building engineer.

But theories of world politics are not at all like those of physics. No

careful analyst believes that our theories of world politics have attained

either the explanatory quality or the practical usefulness of Newton's

system, much less of quantum mechanics; and there is general skepticism

that they will ever approximate the rigor and accuracy even of seven-

teenth-century physics. Furthermore, since both world politics and our

values keep changing, there is no guarantee that even a well-tested theory

will remain valid in the future. Each proposition of any theory of world

politics should therefore be scrutinized carefully to ascertain the range

of its applicability, its robustness under different conditions, and the

likelihood of its being overtaken by events .

What this suggests sounds paradoxical. The problematic character of

I IIIII I

1111

i 1

1

,II

Ill

'!II :I

::!' II' r

11

1

''I il,

jill

I;

1:

I

I : ~ • I

II

:I

IIIIi ' I I jll i ·. ~ I' 1,: 1.11

II

1~1 !

6 Robert 0. Keohane

international relations theory increases the importance of studying it

., deeply. If international relations theory were as generally valid as New-

tonian physics is for ordinary events, practitioners could learn only its

theorems, or maxims, without exploring carefully the guestion of on

what assumptions, and under what conditi·::ms, they will continue to be

applicable. But since it is neither so generally valid nor so unchangingly

applicable, we must be on our guard. Critical reevaluation is continually

important in international relations-more so than copybook learning.

Even if one accepts these arguments for studying international relations

theory, it remains an open gu estion whether political realism and nco-

realism constitute valuable ways to enhance our understanding of inter-

national affairs. Indeed, many of the disagreements expressed in this

volume revolve around this point. It is to be anticipated that readers who

are primarily interested in understanding contemporary world affairs, or

in giving policy advice, will answer this gu estion differently for them-

selves, reacting to this volume's debate over theories of world politics in

at least three ways:

I. Some may conclude that neore-a li st th eory provides a so lid basis lor

und erstanding contemporary events, and that we are unlikely to make sub-

stantial additiona l th eoretical progress in th e ne ar futur e . That is, t>ven if

th e contribution of neorealist theory is relatively sma ll compared with that

of well -d eveloped th eories such as thos e of physics, th e th eory is th e best

we can constru ct, givt'n the diffi culties of th e sub ject. Such a judgm ent

would imply that students of world politi cs shou ld lea rn th e essentials of

neorealist th eory in order to avoid reli an ce on c rud e oversimplifi cat ions or

car icatures; but that th ey should not dwell too long on th eory, but go on

to study spec ifi c asp ec ts of contempor ary intf" rnational affairs in de tail.

2. A different conclusion would be that ne orealist theory has value in

ce rtain situations, hut that for analyzing other situat ions it is of quite lim-

ited us e-or even misleading-and cou ld be improved through modifica-

tion. For thos e who beli eve this, continuing to work on th eory holds more

promise than for thos e in th e first camp: it is worthwhile to see k to mod-

ify and extend neorealist theory in order to increase its validity and us eful-

ness. Practitioners who adhere to this vi ew have two tasks: to distinguish

carefully th e areas where neorealist th eory is applicable from those in

which it is irrelevant or confusing; and to reflect periodically on new at-

NEOREALISM AND WORLD POLITICS

tempts at international theory, asking themselves whether these could be

helpful in interpreting contemporary events .

3. A third view would hold that neorealist theory is fundamentally

flawed and misleading to the core- an "orrery of errors;' as Richard Ash-

ley claims. From this perspective, theoretical work that fundamentally re in-

terprets the nature of world politics is crucially necessary; and in the

meantime, policymakers and analysts of particular issues should strive to

rid themselves of the pernicious biases of neorealism. From this perspec-

tive, studying neorealism is necessary in order to "know thy enemy:'

' 7

V/hichcver view is taken, some value will have been derived frotn

studying realism and neorealism, even if exactly what has bee n gained

will depend on the reader's own evaluation of the theoretical contentions

put forward in this volume.

II. POLITICAL REALISM AND NEOREALISM

Even as long ago as the time of Thucydides, political realism (as desc ribed

in more detail in my own essay later in this volume) contained three key

assumptions: (I) states (or city-states) are the key units of action; ( 2) they

see k power, either as an end in itself or as a means to other ends; and

(3) they behave in ways that are, by and large , rational, and therefore

comprehe nsibl e to outsiders in rational terms . These premises do not,

by themselves, constitute the basis for a scie nce : they do not establish

propositions linking causes with effec ts. Yet they have furnished a usable

inte;-pre tative framework for observers from Thucydides onward. Indeed,

all three elements can be found in Thucydides' discussion of the causes

of the Peloponnesian War in the first chapter of his book. Having de -

scribed the complaints and legal violations leading up to the war, Thu-

cydide s observes that "the real cause I consider to be the one which was

formally most kept out of sight. The growth of the power of Athens,

and the alarm which this inspired in Lacedaemon, made war inevitable"

(Thucydides, ca. 400 B.C./19 51, chapter I, paragraph 24 ). That is, the

Lacedaemonians rationally feared that Athens would at some point direct

its growing power against their interests, and decided to act while they

could still exert some influence over the course of events.

One of the appeals of realist thinking is its applicability to practical

Jl l

IJ

Ill: :::1 li!i

l j: I I I I I

!I;

I' I i1

_ lli.l

8 Robert 0. Keohane

problems of international relations: it provides a readily comprehensible

. "' set of steps to be followed by those seeking to understand and deal with

potential threats to the security of their states. Analysts working within

the realist tradition will proceed as the Lacedaemonians did. They will

focus on states that could constitute effective threats, alone or in coalition

with one another, given the power at their disposal. They will interpret

the actions of those states not on the basis simply of their announced

policies or on the assumption that they will behave morally, but rather

on the premise that they are seeking rationally to increase their power.

And they will devise policies that would protect their own society by

amassing or maintaining sufficient power, alone or in coalitions, to main~

tain their essential security interests.

In view of its easy applicability to a competitive interstate system, and

its prudential maxims, it is not surprising that realism was regarded as

intuitively plausible by statesmen in post-medieval Western Europe. After

the Peace of Westphalia in 1648, with its legitimation of the state system,

political realism became the generally accepted conventional wisdom,

particularly in continental Europe. Critics of power politics who offered

alternative plans based on loose federations of republican governments-

such as Immanuel Kant, in Perpetual Peace ( 179 5 )--did not exercise de-

cisive influence on the thoughts of those in power. The response of Frederick the Great to the earlier, more utopian, plan of the Abbe de

Saint Pierre was rather typical : "The thing is most practicable; for its

success all that is lacking is the consent of Europe and a few similar

triAes" (Hinsley 1963 : 45).

In Great Britain, however- and even more in the United States-

there has been a greater tendency to envisage alternatives to power

politics and to question the premises of political realism (Wolfers and

Martin 1956). The arbitration movem ent, Woodrow Wilson's speeches

during World War I, and the Kellogg-Briand Pact of 1927 to "outlaw

war" all exemplify this institutional and legalistic approach to interna-

tional relations. Yet by the late 1930s this liberal theory was in decline:

1939 marked the publication date of E. H . Carr's classic attack on the conceptions of harmony of interest, or of morality unrelated to power in

world politics. As Carr stated in the second edition ( 1946) of The Twenty

Years' Crisis, I 9 I 9-1939, "in the international order, the role of power is

greater and that of morality less" (p. 168).

NEOREALISM AND WORLD POLITICS 9

World War II elevated this realist perspective to the new orthodoxy

in Anglo-American thinking on international affairs. The struggle with

Nazism cast doubt on the efficacy of international law and emphasized

the role of power in world politics. This shift in perspective was partic-

ularly rapid in America, since the collapse of the European balance of power meant that the United States, no longer merely an observer of

European foibles, had accepted the burdens, along with the heady priv-

ileges, of becoming a hegemonic power-one with both the willingness

and the ability to make and maintain rules for world politics.

It is therefore not surprising that during and immediately after World

War II the tradition of power politics was r ev ived and reinvigorated in

the United States. John Herz , George F. Kennan, Walter Lippmann, and Hans J. Morgenthau articulated what Morgenthau called "political real-

ism;' in contrast to the "utopianism;' "legalism;' or "idealism" that they

associated with liberal writers on international affairs. These self-styled

realists sought to reorient United States foreign policy so that American

policymakers could cope with Soviet attempts at domination without

either lapsing into passive unwillingness to use force or engaging in

destructive and quixotic crusades to "make the world safe for democracy"

Their ideas were greeted warmly by the policymakers, who sought, in

Stanley Hoffmann's words, to " exorcise isolationism, justify a permanent

and global involvement in world affairs , [and] rationalize the accumulation

of power.''' During the postwar years, political realism swept the field in the

Un1ted States. Its opponents may have been overwhelmed as much by

the exigencies of the Cold War as by the rhetorical brilliance of the

leading realists or the power of their arguments. Yet for the most part,

discussions of foreign policy have been carried on, since 194 5, in the

language of political realism-that is, the language of power and interests

rather than of ideals or norms . In public discourse in the United States

today, foreign policy prescriptions are rarely justified directly by reference

to universal moral principles or utopian aspirations. When commentators

wish to justify policy prescriptions on ethical grounds, they smuggle their

ethics into the ambiguous and elastic concept of "the national interest?' 2

For commentators such as Kennan and Lippmann, the triumph of

realism as a way of thinking about foreign policy was sufficient. They

were more interested in practice than in abstract theory Like them,

IIIII '!''

1 ~~~~ .. I .I

I.· II 1,!

I

Li ~·

~ II I

I f

,;''1:

il l: I !I

I ~~ ~~~ 11 1

j'

I~ ' ~~ 1'1 . I I til

'!''

l

fi:

'1 :.'1, !,· I, ,,

::1·'

~~r ,'Ill ·11·:: I I~

II.'. IIi!, II' ~~~~~ ;1:1

,f

i1 1!1 i

'I i 4.1

IO Robert 0. Keohane

Morgenthau wished to influence contemporary foreign policy, and much

. ' of his writing critically assessed American actions in light of his concep-

tion of the "national interest:' Morgenthau did not aspire to be a value-

free scientist, detached from the world of power. On the contrary, he

viewed realism largely as an interpretive guide, which would help us to

"look over the shoulder" of a statesman, enabling us "to read and an-

ticipate his very thoughts" (Morgenthau 1948/ 1978:5; Ashley 1981 ).

Yet Morgenthau also sought to use Realism to create what he called

a "science" of international politics . He was, after all , a scholar who

sought to extend knowledge as well as to apply it. Furthermore, he was

well aware that political realism was at odds with the American political

tradition and wou ld therefore come under attack; when this happened,

its :;tatus as a science would make it eas ier to defend. Thus it is not

inappropriate for us to consider Morgenthau's bold attempt to create

such a science, as long as we do not commit the error of assuming that

this was his only aspiration, or that he ca n be categorized simply as a

prec ursor of later ana lysts. He was, as Stanley Hoffmann points out, the

founding father, if ther e is one, of our discipline (Hoffmann, 1977:44 ).

Neither Kenneth N. Wa ltz's Theory cj' Internationa l Politics, four chapters from which are reproduced below, nor the res ponses of Waltz's critics,

can be fully understood without some comprehension of Morgenthau's

attempt to construct a theory of international politics.

Some of Morgcnthau's arguments, such as his reliance on objective

national interes ts, have been thoroughly discussed in the literature. His

failure to explore problems of misperception systematically has been

noted, and appropriate modifi ca tions have been introduced into the realist

canon. 3 Other conce ptions of his, however, require more attention. Like

all r ealists, Morgenthau relied heavi ly on the concepts of power, rationality

and the balance of power; accordingly, analysis of his treatment of these

concepts offers insights into the strengths and weaknesses of realist

theory, and therefore into the nature of the task that Waltz set for himself.

Morgenthau characterized international politics as a struggle for power

and argued that it could be understood by assuming that statesmen

"think and act in terms of interest defined as power" (Morgenthau 1948/

1967:5). International politics is a struggle for power not only because

of the inherent logic of a competitive realm such as world politics, but

also because of the "limitless character of the lust for power [which]

~r NEOREALISM AND WORLD POLITICS II

reveals a general quality of the human mind" (Morgenthau 1946:194 ).

As Waltz (1959:34ff.) points out, Morgenthau is not content to see power

as an instrument for the attainment of other ends in a competitive world,

but regards it also as an end in itself, due to the nature of human beings.

Explanations of international conflict as r es ulting from human nature

are vulnerable to severe criticism. As Waltz (1959:39) argues, they blame

a small number of behavior traits for conflict, ignoring more benign

aspects of human nature that point in the opposite direction. Further-

more, such theories fail to ex plain variations in warfare over time and

space: if human nature is constant, why .shouldn't warfare be constant

as well?

If Morgenthau's reasons why world politics is a struggle for power are

not entirely convincing, neither is his treatment of the conce pt of power

itself. His definition of power was murky, since he failed to distinguish

between power as a resource (based on tangible as well as intangible

assets) and power as the ability to influence others' behavior. If the latter

definition is adopted, any effect ive action in world politics will necessarily

involve power; but since this is a tautology, we will have lea rned nothing

about the capabilities that create such influence. Is others' behavior

affected more by greater numbers of tanks, superior economic produc -

tivity, or hy an attractive id eology ? If, on the other hand, power is defined

in tern1s of specific resources , we avoid tautology and ca n begin to

construct and tes t theory. Unfortunately, however, theories based solely

on definabl e power capabilities have proven to be notoriously poor at

accounting for political outcomes (Baldwin 1979; March 1966; Keohane

1983, reprinted below in article 7).

Morgenthau' s conception of rationality is clearer than his view of

power. Although he does not offer a formal definition in Politics Amona

Nations, he seems to accept the conception that is standard in neoclassical

economics. To say that governments act rationally in this sense m eans

that they have consistent, ordered preferences , and that they calculate

the costs and benefits of all alternative polic ies in order to maximize

their utility in light both of those preferences and of their perceptions

of the nature of reality.• Morgenthau explicitly acknowledged that the assumption of rationality

was not descriptively accurate-indeed, one of his purposes was to

instruct leaders in order to enable them to act more rationally-but he

1

1

11 ,11 1'1

'i·

·I I f

i'.l' il,

'I

!II I

: 111 ,,,il 1.!!

II : 11

I 1 ' 1

:I! I!

IIi 1:,

;I I

1 ~~ I

12 Robert 0. Keohane

believed that it could be used as a baseline, which could be "tested

'~gainst the actual facts;' making a theory of international politics possible

(Morgenthau 1948/1978:5). That is, even though such an assumption is

not always descriptively correct, it serves a valuable theoretical function.

With it, the analyst can infer actions from interests, and thereby construct

an explanatory theory of behavior. Against the baseline provided by the

theory's prediction, we can ask how "imperfections" caused by misper-

ceptions, a lack of information, bargaining perversities, or even sheer

irrationality could have made actual patterns of behavior diverge from

our expectations.

Morgenthau's sophisticated use of the rationality assumption was con-

sistent both with that of Thucydides and those of later realists and

neorealists, including Waltz. Yet there are alternative approaches that do

not sacrifi ce the possibility of systemic theory. One such research strategy

would be to follow the lead of Herbert Simon, who introduced the notion

of bounded rationality into the study of decisionmaking (Simon 1979,

1982). Actors subj ect to bounded rationality cannot maximize their util-

ities, since they find it diffi cult to use available information to calculate

the costs and benefits of every alternative course of action. They therefore

use shortcuts such as rules of thumb in order to "satisfice"-achieving

a satisfactory level of performance rather than an optimal one. In the

terms of Akerlof and Yellen (1985), they may be "near rational?' That is,

their deviations from rationality may not be so costly as to lead them to

c hange their behavior.

This approach seems intuitively to describe governmental behavior

better than does maximizing rationality; and it is also more plausible as

a description of how business firms behave (Allison 1971; Snyder and

Diesing 1977; Nelson and Winter 1982). Although Waltz is content to

make theoretical assumptions about units that deviate sharply from their

known patterns of behavior, this is not, pace Milton Friedman (1953), a

universally accepted practice in the natural or social sciences (Cyert and

Simon 1983:101; McKeown 1986). Indeed, the correctness of an empir-

ically testable theoretical assumption is by no means irrelevant to the

adequacy of a theory. As Cyert and Simon say about neoclassical theory,

there are great difficulties in "attributing, to firms, decision processes

_more sophisticated than those yet invented by economists" (ibid., p.I03).

This object ion is reinforced by recent findings that satisficing or near

NEOREALISM AND WORLD POLITICS 13

rational behavior at the unit level can produce substantially different

system level outcomes than those characteristic of maximizing rationality.

This suggests that systemic models based on maximizing assumptions

may seriously distort our understanding of what happens in world pol-

itics. Students of international politics should be wary of becoming too

attached to the assumptions of neoclassical microeconomics, just when

these assumptions are under attac k on behavioral grounds (Kahneman

and T versky 1984) and when imaginative economists are seeking to build

rigorous models based on satisficing (Nelson and Winter 1982) or on

near rationality (Akerlof and Yellen 1985). Conceptions of satisficing or

near rationality open up the possibility of constructing systemic theories

of world politics that do not rely on the implausible unit-level assumption

of perfect rationality built into classical microeconomics (Keohane 1984,

chapter 7 ). 5

Morgenthau's third major concept was that of the balance of power,

whic h he referred to as a "necessary outgrowth" of power politics (Mor-

genthau 1948/1967: 161 ). The balance of power is for Morgenthau a

"universal concept:' His desire to demonstrate the universality of the

balance of power may be what led him to use the term so broadly as to

lead to inconsistency, as I. L. C laude ( 1962 :25-37) showed. Morgenthau used "the balance of power" to refer to a situation of equilibrium as

well as to any situation in which power struggles take place; but since

Morge nthau did not regard equilibrium as inevitable , this double usage

of the phrase led him into unresolvable contradictions.

Wi~hout coherent definitions of "power" and "balance of power;'

Morge nthau was unable to create a consistent and convincing theory.

Subsequent attempts to construct better theory were made by others,

notably Morton Kaplan (19 57), Stanley Hoffmann ( 19 59, 1965 ), and

Richard Rosecrance (I 96 3 ). All three of these theorists sought to use

what Waltz refers to as "systemic theory" to attempt to account for state

behavior. That is, they abandoned reliance on the nature of human beings

to account for discord and cooperation in world politics, but focus ed

instead on the competitive, anarchic nature of world politics as a whole.

Furthermore , unlike Morgenthau-who sought general principles-they

were more interested in explaining variations : alternations of war and

peace, stability and instability. They asked not so much about an alleged

· "essence" of international politics as why international systems changed.

Iiiii I

111

1

:

II ' 'I

I.' I'·. I''•'' 'Iii!' I

.1!1 I

lilt I

:!!·!·: I

:I

l4 Robert 0. Keohane

In chapter 3 of Theory ?[ International Politics (not reprinted in this

. ,, volume) Waltz agrees with these writers that a good theory of interna -

tional politics must be systemic, since how the relationships among states

are organized strongly affects governments' behavior toward one another.

A system, fo r Waltz, consists of a set of interacting units exhibiting

behaviora l reg ularities and hav ing an identity over tim e. Yet Wa ltz is

severely critical of Hoffmann, Kaplan, and Rosecrance for not having

gone beyond the descript ion of internat ional systems to identify their

structures, defined sepa rately from the attributes of and relations between

the units. "In order to turn a systems approach into a theory, one has

to move from the usua l vague identifi cation of systemic forc es and effects

to their more prec ise specifi cation, to say what un its the system com-

prises, to indicate the comparative weights of systemic and subsystemic

causes, and to show how forces and effects change from one system to

another" (Wa ltz 1979:40-41 ). And it is crucia l that the ana lyst " carefu lly

keep the attributes and interact ions of the system's un its out of the

defin ition of its structure;' if the theory is to avoid the peril s of reduc-

tionism (relying for explanation on unit- level rather than system- level

attributes) or tautology.

In chapters 4-6 of Theory ?[International Politics, reprinted as art icles 3-

5 of this volume , Waltz prese nts his own systemic theory.6 The chapter

on "reductionist and systemic theories" defends the ana lytica l priority of

systemic theory and inJicates how such a theory cou ld meaningfu lly

explain state behavior. "Political structures" argues that structure "de-

fines the arrangement, or the ordering, of the parts of a system" (p. 73).

Structures vary along three dimensions: by their ordering prin ciples, the

specification of func tions of formally differentiated parts, and the relative

capabilities (or power) of the units themselves. International relations is

an anarchic rather than hierarchic realm, populated bv units (states)

performing simi lar functions. Thus any international systems that we

analyze are "ordered" by the principle of anarchy. And in such systems,

we need not be concerned with the functions performed by the units,

since they are functionally alike. Thus the dimension of differentiation of

units "drops out:'

This characterization of the first two attributes of international systems

enormously simplifies the analyst's task, since it means that structures

of international systems differ only along the third dimension, that of the

NEOREALISM AND WORLD POLITICS r5

distribution of power. In analyzing actual international-political struc-

tures, we therefore "abstract from every attribute of states except their

capabilities:' What emerges Waltz calls "a positional picture;' which

portrays the placement of the units-where they stand relative to one

another- rather than their intrinsic qualities. The key changes that we

are to look for, in international politics, are changes in the distribution

of capabilities across un its.

Waltz uses his defin itions of system and structure in "Anarchic orders

and balances of power" to develop a theory of the balance of power that

will not be vu lnerable to th e criticisms leveled against Morgenthau. For

Wa ltz, a good theory wi ll not merely point to the importance of power

and the ba lance of power, as Morgenthau did, but will account for the

recurrent formation of balances of power in world politics, and tell us

how changing power configurations affect patterns of alignments and

conflict in world politics. From the anarchic nature of the international

system, and the assumption that states "are unitary actors who, at a

minimum, see k their own preservation and, at a maximum, drive for

universa l dominat ion;' Wa ltz deduces that balances of power must nec -

essarily emerge. Furthermore, as states compete with each other "they

wi ll imitate eac h other and becom e socialized to their system" (p. 129).

In the next threC' chapters of Th eory (not reprinted in this volume ),

Wa ltz applies his stru ctural theory to issues of economic interdependence,

military relationsh ips, and what he ca ll s "the managem ent of international

affairs:' He argues in chapter 7 that discussions of interdepe ndence ex-

agge~ate the extent to which great powers, including the United States,

are dependent on others. In c hapter 8 he contends that military power

remains as usefu l as ever-even that its usefulness has increased-and

that military bipolarity is remarkably stable. Chapter 9 emphasizes the

role of great powers in maintaining order: the problem "is not to say

how to manage the world, including its great powers, but to say how

the possibility that great powers will constructively manage international

affairs varies as systems change" (Waltz 1979:21 0).

The significance of Waltz's theory, as elaborated in the selections

reprinted here , lies less in his imtiation of a new line of theoretical

inquiry or speculation than in his attempt to systematize political realism

into a rigorous, deductive systemic theory of international politics. The

Waltzian synthesis is referred to here as neorealism, to indicate both its

lill!i I ~ I :

I i! ,, I ,I ,I .,

II

1.

i,l[l

'ill ''II ill I'

,, I• I 11·1

I, I

li'i: .p, 1 ii 'l•l

J

r6 Robert 0. Keohane

intellectual affinity with the classical realism of Morgenthau and Herz

. ' and its elements of originality and distinctiveness. 7

III. CRITICISMS AND REBUITALS

The importance of neorealism has been widely recognized, and it has

been subjected to critical scrutiny from a variety of perspectives, and

with varying concentration on Waltz's work. Of the articles reprinted

here, only one (that of John Gerard Ruggie) actually constitutes a review

essay of Theory ?[International Politics. The others all had broader purposes;

but for each author, neorealism (a term coined, so far as I know, by

Robert W Cox) constitutes a central issue for discussion, and in every

case Waltz's work is recognized as a major statement of neorealist doc-

trine. Since these articles appeared in a wide variety of publications

(three different journals and one collection of works on political science

as a discipline), they have often not' been considered together, as major contributions to a sustained, focus ed debate. Thus a major purpose of

this volume is to collect the debate on neorealism within one set of

covers, so that students and other observers of world politics can focus

coherently on the issues that it raises.

Until publication of this volume, Waltz had not responded in print to

the criticisms expressed by Ruggie, Cox, Richard K. Ashley, and myself.

A second purpose of Neorealism and Its Critics, therefore, is to provide an

appropriate forum for Waltz to do this, giving readers the opportunity

to judge the validity of his theory not only in light of the critics' argu-

ments but in view of his defense and reconsideration of his views. Waltz's

response clarifies his arguments, in some cases reducing the differences

between himself and his critics, in others making the nature and signif-

icance of those disagreements more readily apparent. The intellectually

honest and constructive character of his response both advances the

debate and illustrates how genuine scholarly discourse should be carried

out.

This volume begins with the principal theoretical chapters from Theory

if International Politics, reprinted with only minor editorial changes so that each reader will have the opportunity to judge Waltz's theory, and the

various criticisms of it, for himself or herself. The first commentary

NEOREALISM AND WORLD POLITICS I7

consists of Ruggie's review essay; the articles by Keohane, Cox, and Ashley

are presented in order of their increasing divergence from Waltz's prem-

ises. Following Ashley's essay, the volume continues with a response to

Ashley's critique of neorealism by Robert G. Gilpin, and Waltz's con-

cluding essay. 8

John Gerard Ruggie embeds his criticisms of Waltz in a larger appre-

ciation of his work, declaring Theory if International Politics to be "one of the most important contributions to the theory of international relations"

since Waltz's last book on the subject, Man, the State and War ( 1959). Yet

Ruggie also concludes that Waltz does not fully succeed on his own

terms. By truncating his own concept of structure, Waltz has (according

to Ruggie) made it impossible for his theory to account for major changes

in world politics, such as that between the medieval and modern period.

To show this, Ruggie focuses on the concept of sovereignty, arguing that

the medieval system differed from the modern one "by the principles on

the basis of which the constituent units are separated from one another"

(p.142). After correcting what he regards as Waltz's misleading definition

of differentiation, Ruggie uses this concept to depict what he calls an

"institutional transformation" from a medieval set of arrangements to

the modern conception of sovereignty. He argues that the differentiation

of units does not drop out of a structural theory of world politics, but

provides a way of understanding change--from the medieval to modern

system, and perhaps to a future international system . Ruggie also criticizes Waltz for overlooking Durkheim's concept of

"dynamic density"-the quantity, velocity, and diversity of transac-

tions-as a determinant of change in world politics. Such changing pat-

terns of interdependence could, according to Ruggie, affect world politics

even without changes in the structure of the system, narrowly defined.

Ultimately, for Ruggie, structural change has no source other than unit-

level processes. Waltz's exclusion of such processes prevents him from

developing a theory of change in the form of a "generative" model that

would analyze how changes in interdependence, or "dynamic densitY,'

affect the relationship between the structure of an international system

and the degree of order observed within it. Robert 0. Keohane focuses on what he calls the "research program"

of Structural Realism-referred to in the rest of this volume as nco-

realism. Keohane accepts Waltz's emphasis on system-level theory and

11

1

I·

.lij

!i! •, ,, I

I

::1111

I Iii, ' iP 1:'1 I

llo.

r8 Robert 0. Keohane

his acceptance of the rationality assumption as starting-points for theory

. ,.in international relations. Keohane argues that systemic theory is essential

both because we have to understand the context of states' actions before

accounting for the actions themselves; and because a good structural

theory can be simpler (more parsimonious) and more readily testable

than theories that rely on variations in the internal attributes of states.

Yet Keohane is dissatisfied with Waltz's theory. He seeks to show that

Waltz's theory of the balance of power is inconsistent with his assumption

that states seek to "maximize power.' He points to difficulties in relying

heavily on the ambiguous concept of power, especially when the contexts

within which power is exercis ed are not sufficiently specified and dis-

tinguished from one another. He claims that Waltz fails to test his theory

in accordance with the standards that he set up in chapter I . He argues

that Waltz's theory does not explain change well and agrees with Ruggie

that more attention needs to he paid to connections between the internal

attributes of states on the one hand, and the international system on the

other.

Keohane suggests the need for a revised theory that incorporates

Waltz's notion of structure but that takes seriously, as explanatory factors,

elements of the international system not included in Waltz's limited

conception of stru cture. Such a theory would follow the lead of earlier

work by Keohane and Nye ( 1977), which emphasized the significance of

economic processes and of international political institutions. Keohane

calls for "systemic theories that retain some of the parsimony of Struc-

tural Realism, but that are able to deal better with differences between

issue -areas, with institutions, and with change " (p. 197). Foreshadowing

his later work (Keohane 1984 ), he also calls for more serious attention

to the ways in which international systems may facilitate or inhibit the

flow of information, thereby affecting the behavior of actors and their

ability to cooperate with one another.

Robert W Cox also focuses on the alleged inability of neorealism to

comprehend change in world politics. In Cox's view, Morgenthau and

Waltz infelicitously transformed realism from the category of historically

based critical theory to what he calls "problem-solving theorY,' which

regards contemporary institutions and power relations as permanent.

Such a theory is flawed by the error of "taking a form of thought derived

from a particular phase of history (and thus from a particular structure

NEOREALISM AND WORLD POLITIC S 19

of social relations) and assuming it to be universally valid" (p. 214 ).

Critical theory is needed to understand change : the analyst needs to

stand outside the prevailing order and inguire into how that order came

about and what forces are at work to change it. The most promising

form of critical theory, for Cox, is historical materialism , which sees

conflict as a possible source of structural change rather than as "a

recurrent conseguence of a continuing structure" (p . 215). By analyzing

the interplay of ideas, material capabilities, institutions, and social forc es,

critical theorists can understand how world orders, such as the Pax

Americana , came about and can ther efore dis cern some of the possible

ways in which they might change .

In a postscript written especially for this volume, Cox contrasts the

research programs of positiYism (with which he associates Waltz's nco-

r ealism) and of historicism, or historical materialism (vvith which he

identifies hims elf). Cox argues that neorealism does not suffi ciently take

account of human ideas and practices, and that its search for general

laws prevents it from accounting for change. The research program of

historic ism, by contrast, involves finding connections between people 's

ideas and the material world, revealing the historical structures charac -

teristi c of parti cular eras, and ultimately explaining transformations from

one stru cture to another. In contrast to neorealism, which is "ideologically

at the service of big-power management of the international system;'

Cox's nond eterministic Marxism foc uses on class struggle as "the heu -

ristic model for the understanding of structural change" (p. 248 ).

Like, Cox, Richard K. Ashley sharply distinguishes neorealism from

the classical realism of Morgenthau and Herz. Neorealism, like structur-

alist analysis in other fields, see ks to "transcend empiric ist fixations" by

comprehending deeper levels of social reality through systematic social

scientific investigation; yet according to Ashley, this attempt leads to

contradictions and ideological narrowness . Neorealism is statist, yet the

atomistic utilitarian epistemology adopted by its proponents undermines

"the ideal of the state-as-actor upon which the ir distinction among 'levels'

and their whole theory of international politics depend" (p. 279). Fur-

thermore , the positivist commitment to technical rationality and the

dichotomy between scientific knowledge and values "ties positivism to

an ideology of its own" by endorsing an unguestionable faith in scientific-

technical progress (p. 282). Positivism becomes committed to what Ash-

I .I 'I' I I :l i! II

li I i! I' : I

I

'·1.

'

20 Robert 0. Keohane

ley calls an "actor model" of social reality, within which science cannot

question the constitution or the ends of social actors, but can only provide

advice about means . In the end, therefore, "politics in neorealism be-

comes pure technique~' Neorealism thus combines the superficiality of

positivistic atomism and structuralism's inability to account for change,

with an ideological aversion to critical thinking about values .

Ashley's alternative is his dialectical competence model, which has some

obvious affinities with Cox's project for the analysis of changing social

forces . Ashley 's model would account for the emergence of balance of

power politics and explore the conditions for its continuation or trans-

formation. In partic ular, it would explore the subjective aspects of the

balance of power, focusing on the manipulation of symbols and on learn-

ing. It would seek to provide an account of crisis and consider alternatives

to "the mod ern global hegemonY,' in the hope of transce nding its con-

ditions of dominance. And it would reinterpret neor ea lism as a perverse

r esearch program contributing to "the impoveri shment of politica l imag-

ination and the reduction of international politics to a battleground for

the self-blind strategic clash of technical reason against technical reason

in the service of unquestioned ends" (Ashley, p. 297).

The criticisms of neorealism made by Ruggie, Keohane, Cox, and

Ashley can be summarized with the aid of an architectural metaphor.

Those who accept the found ations of neorealism, and the overall shape

of the building, can still argue about the exact des ign. If the concept of

international structure is valuable, how much should it include: what

should be the boundaries between unit-level and system-level variables?

To what extent should system-level analysis have priority over analysis of

foreign policy, and how should the analysis of these two levels be linked?

More fundamental questions about the edifice itself can be raised even

by scholars who accept neorealism's utilitarian, positivistic m ethodology.

How well-crafted are the cornerstone concepts of power, the balance of

power and sovereignty? Shouldn't neorealist theory take better account

of institutions and the role of information? How could neorealism be

reformulated to account for transformation as well as continuity? Finally,

the deeper foundations of the structure can be questioned . Neorealism,

in the view of its more severe critics, ignores both history and human

subjectivity. It does not investigate how the order that it analyzes came

~~- ~- ~~

,'i. ~~-:

II" I>

NEOREALISM AND WORLD POLITICS 21

about, nor does it consider the production relations on which it depends.

Philosophically, according to the critics, neorealism does not rest on the

solid rock of a coherent epistemology but rather is sinking into a swamp

of a state-as-actor theory contradicted by its utilitarian premises .

In the symposium that accompanied "The Poverty of Neorealism" in

International Organization, Robert G. Gilpin replied to many of Ashley 's

criticisms. In his essay reprinted below, Gilpin describes realism as em-

phasizing that "the final arbiter of things political is power,' and that

"the essence of social reality is the group;' represented preeminently in

our own time by the state. Gilpin concentrates his critique on Ashley's

distinction between classical realism and neor ealism . For Gilpin, conti-

nuity in the tradition of political realism is more important than differ-

ences between contemporary neor ealists and their intellectual ancestors.

Gilpin contends that the greatest realist writers have always appreciated

"the intimate connection between international politics and international

economics" (Gilpin, p. 309). Nor do neor ealists adopt determinism in

contrast to the classical realists' view that statesmen can change the

international environment. Finally, the neorealists, like classical rea lists,

adhere to moral values that they seek to promote : they are not guilty of

"moral neutrality:' On the contrary, "moral skepticism joined to a hope

that reason may one day ga in g reater control over passions constitutes

the essence of realism and unites rea lists of every generation" (p. 321).

Waltz, in his concluding essay, fo cuses principally on three issues that

have attracted the attention of hi s critics: ( 1) what he ca lls his "spare"

definition of international-politica l structure, critic ized es pecially by Rug-

gie; (2) his view on how theories should be tested, ques tioned by Keohane ;

and (3 ) his attempt to construct a problem-solving international structural

theory of international politics rather than a critical theor y of the state-

which in his view is the focus of Cox's and Ashley 's principal complaints.

In defending himself from Ruggie's charge of truncating the concept

of structure and overlooking "dynamic densitY,' Waltz argt~es -that dy-

namic density reflects unit-level processes rather than system structure.

In Waltz's view, elegant definitions of structure (which Ruggie regards

as truncated and insufficiently rich) have the great virtues of precision

and parsimony: they "enable one to fashion an explanatory system having

only a few variables" (p. 330). They do not tell us everything; but they

1

:.,

1

·,',.

I

:-:'1 :1,1: 'I

I:

l ·.ll!i: .,,

I : . ; ~ I

l:,,i

'' I' I::' I

I

I:· ['.!:,

jili ,

.,j 'I ;.1

'I· 1,1,

22 Robert 0. Keohane

do help us understand fundamental continuities in world politics that

would otherwise be obscured by inclusion of too many other factors that

vary from one era to another. In responding to Ruggie's objection, Waltz is attempting also to rebut

my own claim that a good systems-level theory would take into account

"the institutional context of action as well as the underlying power

realities and state position upon which Realist thought concentrates"

(p. 195, below). Ruggie and I both seek to make systemic theory account

for more of what we see by taking into account aspects of international

systems that are excluded from Waltz's spare concept of structure. Waltz

regards changes in the intensity of interactions among the members of a

system, or in the international institutions that connect them, as unit-

level phenomena becaus e patterns of interaction are not included as part

of his definition of system structure. Yet these patterns do not depend on differences among the internal

attributes of states. On the contrary, changes both in the intensity of

interactions and in international institutions can take place even when

states continue to be similar in their internal attributes. These changes,

furth ermore, may affect actors' behavior in classic systemic ways by al-

tering the incentives and opportunities that face them . Such processes

and institutions should therefore, in my view, be consid ered system-level

rather than unit-level changes. By contrast, a unit -level theory relies on

cross -national variations to explain variations in outcomes. The interna-

tional system includes systemwide processes and institutions as well as

a stru cture in Waltz's sense: this should not be forgotten when we

construct systemic theories of international politics. 9

In contrast to the continuing debate over the boundaries of systemic

and unit-level explanations, Waltz's response to my points on testing

seem s to suggest convergence between our views. I had interpreted Waltz,

in chapter I, as believing that social scie nce theories could be tested

rigorously, and the n criticized Waltz for not following his own m eth-

odological strictures. In his response, Waltz explicitly adopts a more

complex and sophisticated view of social science theory-testing, one that

is close to that of Lakatos, with which I am sympathetic. By renouncing

the na'ive falsificationism that he had seem ed to adopt in chapter 1 of

Theory, Waltz clarifies his views and removes the apparent inconsistency

in his earlier argument.

NEOREALISM AND WORLD POLITICS 23

In reply to Cox, Waltz affirms that he set out to construct what Cox

refers to as problem-solving theory, and to do so at the level of the

international system. In Waltz's view, it is these choices to which Cox

objects: Cox believes in critical theory and thinks that without a theory

of domestic politics and state-society relations, a theory of international

structure is of limited value. To Waltz, Ashley's obje ctions run along

similar lines. Associating himself with Gilpin's response to Ashley, Waltz

goes on to deny that he assumes, ahistorically, that states will always exist

as they do now. "I find it hard to believe;' he says, "that anyone would

think that states will remain fixed in their pres ent condition" (p. 339).

He holds, however, that in building a systems-level theory, it is theoret-

ically useful to assume unitary, purposeful states, whose internal char-

acteristics do not vary. "A separate theory dealing with the politics and

policies of states" would be valuable; but Waltz did not set himself this

task.

According to Waltz, this separation of theories of international politics

and theories of the state would be problematic only if one sought to

combine international and domesti c politics in one theory; but Waltz

doubts that this is possible .

Waltz's response to Ashley takes us back to the issu e that I raised

earlier in thi s essay when dis cussing Morgenthau's conception of ratio -

nality. Since Waltz accepts the separation of international and domestic

political theory, he is content to follow th e lead of neoclassical economists,

who do the same. Waltz argues that " economists get along guite well

with s~parate theories of marke ts and firms" (p. 340). But a number of

economists arc dis contented with the disjunction betwee n our knowledge

of firms' behavior and orthodox mi croeconomic theory. Richard Nelson

and Sydney Winter view the neoclassical assumption that "economic man

is a perfect mathemati cian" as an "affront to reason [that) is not innocu-

. ous" ( 1982:66). Richard Cyert and Herbert A. Simon regard this as-

sumption as empirically indefensible and not in accord with the canons

of natural scie nce (Cyert and Simon, 1983:101). One of the issues high-

lighted by the debate between Waltz and his critics is precisely how

serious is the disjunction between the assumptions of our systemic theory

and what we know about unit-level behavior. Some of us seek eventually

· to build an integrated theory of world politics, linking the domestic and

international levels of analysis, rather than being content with unit-level

II :I

11[ ~ ;!!: II;,

I

,,,

il'l

I ll

L,l (l:il. i.'l!'

''I

I

I I!Jdl 1'::1:

1

.·1: II·

II I, :I!·

ili ,l' l ,~1 ! .11

I

_ __ if1r i

24 Robert 0. Keohane

and system-level theories that are inconsistent with one another. Insofar

as this is our goal, we should hesitate before following the neoclassical

economists into what may be an intellectual dead-end. 10

CONCLUSION

The issues on which the neorealist debate explicitly focus es are familiar

ones to socia l scientists, since they have to do with levels of analysis, the

precision of concepts,· and philosophical underpinnings of theory. Un-

derneath these conundrums, how<>ver, lurk guestions of purpos<"s and

valu<"s, theory and practice. The reality of domination-of certa in states

over others, and of elit<>s ov<>r nonelites-continues. Mor<" novel is the

shadow of atomic weapons, which hangs over all students of world politics

today, giving urgency to their attempts to discover how confl ict cou ld

be limited and cooperation facilitated. The widespread, if varied, sense

of dissatisfaction with Waltz's versi;n of neorealism has its roots not only

in the critical, idealistic tradition of commentary on world politics, but

also in the enormity of nuclear war. Although the criti cs reject the

answers of Wilsonian liberalism, they share th e traditional liberal, and

Marxist, sense of unease about Realpolitik . Th <>y seek in some sense to

move, in Ernst Haas's phrase , "beyond the nation-stat<>" (Haas 1964 )--

by devis ing new international institutions or regimes, by thinking about

c hanges in tlw principles of sovereignty on the basis of which units are currently separated from one another, or by fundamentally guestioning

the validity of the "state as actor" model on which neorealism relies.

Thus the critics of neorealism also chall enge , more or less profoundly,

the adeguacy of the interstate system in the contemporary era; and this

challenge raises once again the guestions of theory and practice discussed

in the first part of this essay. If neorealism as articulated by Waltz remains

an adeguate theory, efficaciou s action in world politics will have to take

account of the limitations of anarchy to which it points. But if it is

seriously incomplete , as Ruggie and I argue, more attention will need to

be paid to aspects of world politics that it downgrades or ignores: eco-

nomic and ecological interdependence, changes in the functional capa-

bilities of governments, variations in the availability of information, and

the role of international institutions and regimes. If it is erroneous to its

NEOREALISM AND WORLD POLITICS 25

positivist core, as Cox and Ashley believe, the theory will need to be

restructured in conjunction with massive changes in political practice.

Readers' judgments about these theoretical issues must necessarily affect

how they think about foreign policy and world politics.

None of these issues can be resolved here; they can only be raised .

Since this volume does not enunciate an orthodoxy, its contributors can

only hope that you, the reader, learn from our disagreements:

Like or find fault, do as your pleasures are, Now good or bad, 'tis but the chance of war. 11

NOTES

I. Hoffmann (1977), pp. 47- 48 . See Herz (195 1); Ke nnan ( 195 1) ; Lippmann

(1943); Morgenthau (1946); and Morge nthau ( 1948/ 1967).

2. On t he ambig uit y of th e "national inter est;' see Arno ld Wolfe r s, "The Na-

tional Interest as an Ambiguous Symbol;' in Wolfers ( 1962: 14 7-165 ). The ideo logical hegemony of r ea li sm is regarded as a mixed blessing by som e of its or igina l pro-

pone nts. Herz has recently emphasized that he never regard ed r ea lism and idea lism as in compat ible : "'Rea li st li beralism' r e nd ers realism 'more hum ane' and id ealism

'less c himer ical'" (H erz 198 1 :203) . Eve n Morge nthau, in Politics Among Nacions, ac -

knowl edged that intt-rnational po litics ope rates within the fram ework of rules and institutions, although he stressed that it " cannot be re du ced to legal rul es and

institutions" (Morge nthau 1946/ 1967: 16) .

3. On th e con cept of th e national interest, see Ros e nau ( 1968) and George ( 1980), c hapter I 3. For a classi c study of mispcrceptions, sec Jervis ( 1976 ). For an

attempt to incorporate the concept of misperce ption into realist th eory, see Kin -

derm <t:nn ( 19!! 5 ). 4 . For an en lighte ning dis c ussion of "rat iona le exp lanation s" in political science ,

see Moon ( 197 5 ), especia lly pp. 156-166. A pathbreaking although controversia l attempt to use the assumpt ion of maximizing rationality is Bue no d e Mesquita ( 1981 ). For an inc isive c ritique of this assumption, see McKeown ( 1986 ).

5. Read e rs of my article "Theory of World Po litics: Struc tural Realism and Beyond" will r ecogn ize that l have become more skeptica l of th e valu e of assuming

rationality than l was wht>n th e a rticl e was written . See below, pp. 190- 197. 6. C hapter I of Theo~r. whi c h indicates how Waltz views soc ial theory and the

tests he r egards as appropr iate for it, is also r e printed below, as article 2 of this

volume . 7. Since l am c ritical of Wa ltz on a numb er of points, it follows that I am

unwilling to classify myself as a neorealist, despite my sympathy w ith a number of Waltz's positions and Richard Ash ley's inclusion of my name as one of the contem-

porary neorealists. As readers of my essay reprinted in this volume, or my subsequent

book, Afier Hegemony ( 1984 ), will recognize, l admire the clarity and parsimony of

II

!I ill :11

I iii ,,

'I :rl II ,il ill

26 Robert 0. Keohane

Waltz 's systemic th eo ry without subscribing to many of th e inferences that he draws from it.

'·' 8. When I refer to myself in this essay with refere nce exclusively to my 1983 essay on theories of world politics, reprinted as article 7 below, I use the third p erson, but whe n I refer to my current views, I us e th e first person form .

9. The theory of regime change prese nted in Power and Interdep endence ( 1977) by JosephS. Nye and myself is a system- level rather than a unit-level th eory by th e criteria outli ned in this paragraph . So is my fun c tional theory of internationa l r egimes in chapter 6 of After Hegemony ( 1984 ).

I 0. As Timothy Mc Keown (! 986) has emphasized, th e question is how mu ch we should rely on a th eor y whose premises are markedly inco nsiste nt with what we know about how choi ces are ac tually made. Admittedl y, as Wa ltz emphas izes, ever y theory emp loys simplifying assum p t ions; but if a theo ry neither makes acc urate predictions about state be havior nor helps us und erstand th e polit ical processes of c hoi ce , it may be inferior to pote ntial altern atives. Il lustrating his argument, Mc Keown points out th e diffi c ulti es enco untered by sev('ral generations of stru c tural th('orists who have sought to explain states' trade po licies hy employ ing th e unitar y ac tor, uti lity-maximizing assumptions of mi c ro economics.

I I. Wi lli am Shakesp eare , Th e Hiswry r:{ 7roi lus and Crcssid" (1603), prologue, lines 30-31.

:;•

TWO

Laws and Theories

KENNETH N. WALTZ

I WRITE THIS BOOK with three aims in mind: first, to examine theories of international politics and approaches to the subject matter that make some claim to being theoretically important; second, to con-

struct a theory of international politics that remedies the defects of

present theories ; and third , to examine some applications of the theory

constructed. The rcCJuired preliminary to the accomplishment of these

tasks is to say what theories are and to state the requirements for testing

them.

Students of international politics use the tem1 "theory" freely, often to

cover any work that departs from mere description and seldom to refer

only to work that meets philosophy-of-science standards. The aims I

inte nd to pursue requ ire that definitions of the key terms theory and law

be ca refully chosen. Whereas two definit ions of theory vie for acceptance,

a simple definition of law is widely accepted . Laws establish relations

between variables, variables be ing concepts that can take different values.

If a, then b, where a stands for one or more independent variables and

b stands for the dependent variable: In form, this is the statement of a

law. If the relation between a and b is invariant, the law is absolute . If

the relation is high ly constant, though not invariant, the law would read

like this: If a, then b with probability x. A law is based not simply on a

relation that has been found, but on one that has been found repeatedly.

Repetition gives rise to the expectation that if I find a in the future ,

then with specified probability I will also find b. In the natural sciences

even probabilistic laws contain a strong imputation of necessity. In the

ELEVEN

Reflections on Theory of International Politics: A Response to My Critics

KENNETH N . WALTZ

D !CHARD ASHLEY says that older realists, des pite some limitations, ~et a high standard of political reasoning from which I and other neor ealists have regressed. Robert Keohane says that I merely reformu-

lated rea lism and made it more systematic without developing "new ways

of seeing" international relations (1983:515; reprinted above, p. l75).

Ashley and Robert Cox are highly critical of structural approaches. John

Ruggie complains that having started down the structural path, I failed

to follow the path to its end (1983, part V; reprinted above). Facing a

variety of criticisms, perhaps I can best beg in by saying what I tried to

accomplish in Theory qf International Policies. 1 My aim was to do the follow- ing:

I . Develop a more rigorous theory of international politics than ea rli er real-

ists had don e.

2. Show how one can distinguish unit-level from structural elements and

then make connections betwee n them .

3. Demonstrate the inadequacy of the prevalent inside-out pattern of think-

ing that has dominated the study of international politics.

4. Show how state behavior differs, and how expected outcomes vary, as

systems change.

5. Suggest some ways in which the theory can be tested and provide some

examples of its practical application, largely to economic and military problems .

A RESPONSE TO MY CRITICS 323

STRUCTURES AND UNITS

Anyone who believes that a systemic theory is required for an adequate

understanding of international politics has to distinguish between struc-

tural and unit levels. In making the distinction, whether certain com-

ponents belong at one level or the other is not immediately apparent. By

my definition, national political structures are spare, and international-

political structures even sparer. The second term of the dl'finition-

"specification of the functions of formally differentiated units"-appears

in domestic but not in international structures as I conceive of them .

This leaves only the first and third tern1s, the ordering principle of the

system-anarchy-and the distribution of capabilities across its units-

the states. Ruggie argues impressively, but in the end unconvincingly, for

restoring the second term. He tries to show that beca use I omitted it,

both a dimension of change and a determinant of change are missing

from my model (p. 148). He seeks to capture both the dimension and

the determinant by subtly redefining "differentiation:'

Ruggie draws a distinction between differentiation meaning "that

which denotes differen ces rather than that which denotes separateness"

(p.l42). The sociologically "proper" definition of differentiation tells us

on what basis the segmentation of an anarchic realm is determined. The

second component of international-political structure is thus rescued

from oblivion. He then argues that "dynamic density;' as defined by

Durkheim, should have been included in my model as well, since vari-

atiOT)S in density may be determinants of systemic transformation.

In our discourse, saying who was more faithful to Durkheim is less

important than finding the theor etically most tenable and practically most

useful definition of structure. Durkheim is nevertheless an able guide to

the elusive prey. He distinguishes between societies of mechanical and of

organic solidarity, corresponding respectively to the anarchic order of

international politics and the hierarchic order of domestic politics. Dur-

kheim describes a mechanical society as "a multitude of little centres,

distinctive and alike" ( 1893:257). They have their own needs and inter-

ests, but they do not interact through their special characteristics in such

a way as to become entangled in one another's affairs and dependent on

one another's efforts. Each unit does for itself roughly what all of the

others are doing. Their lives are characterized by a duplication of effort

324 Kenneth N. Waltz

rather than by a division of labor that would produce their integration.

Interactions and exchanges among segments are variable and sporadic.

Exchange of products, even if more or less regular, gives rise only to

"simple relations of mutualism having nothing in common with the di-

vision of labor" ( 1893:282). 2 Like units interact only marginally because

of the ir pervasive resemblance. The more nearly units are alike , the less

they can gain by cooperating with one another.

The segm ents of a mechanical society may grow or may shrink. They

may range in size from the clans of what anthropologists call segmentary

lineage societies, to villages and cities , to such extensive territorial or-

ganizations as nations. The distinction between types of society is not

one of size. A society may attain grea t size, but so long as it is composed

of similar segments its unity remain s m echanica l. Beca use the parts

remain little dependent on one another, any of them ca n be severed from

the whole with little effect on the cons ciousness , happiness, hea lth, and

well - being of the remaining parts (189 3: 148- 149, 26 1 ).

A mechanical society res ts on the similarity of the units that compose

it; an organic society is based on their differences. An organic soc iety

promotes the sharpening of individua l talents and skills . Different parts

of the society make their particular contributions to its general welfare.

Units become closely linked because they do special jobs and then ex-

c hange goods and servi ces in order to mee t their common requirements.

The divi sion of labor inc reases effi ciency and promotes the ge neral pros-

perity. More important still, the division of labor makes for social soli-

darity. As Durkheim says, "when men unite in a contract, it is because ,

through the division oflabor .. . they need each other" ( 189 3:21 2 ). The

parts of a highly developed society are tightly integrated . Som e parts

depe nd on others for services and supplies that they cannot easily, if at

all, provide for themselves. 3 Mechanica l societies are loosely linked

through the resemblance of their members. Organic societies become

closely integrated through the differences of their members.

The division of labor renders "societies possible which, without it,

would not exist:' Social structure is transformed as society progresses

from mechanical to organic unity. From differ ences of structure predic-

tions can be made. Understanding this enabled Durkheim to make one

of the most striking predictions to be found in social-science literature.

The sexual division of labor, he writes, establishes "a social and moral

A RESPONSE TO MY CRITICS 325

order sui aeneris." The division of labor brings unlikes together. Unlike

individuals depend on each other precisely because their different activ-

ities, the distinct tasks they perform, contribute to their mutual satis-

faction and benefit. Unlikes become strongly glued together because they

depend on one another's different abilities and skills. It follow s that if

the division of labor were to lessen, for example if the roles of the sexes

became less distinct, marriage would become less stable, and "conjugal

society would eventually subsist in sexual relations preeminently ephem-

eral" ( 1893:61 ). The union of likes is brittle beca use one's own efforts

can replace the other's contributions .

The transition from mec hanical to organic society proceeds as more

and more individuals come " suffi ciently in contact to be able to act and

r eact upo n one another. If we ag ree to ca ll this relation and the active

[social and economic] commerce res ulting from it dynamic or moral

density, we ca n say that the prog ress of the division of labor is in direct

ratio to the m ora l or dynamic density of society" ( 1893 :257). Durkheim's

view seems much in accord with Ruggie 's interpreta tion . Dynamic den-

sity acts as a force that may transform mechanical socie ties or produce

a transition from an anarchic to a hierarchic order. In Ruggie's words,

"the principles on the basis cf which the constituent units are separa te d from one another" have changed (p. 142, his italics). But for Durkheim this is

only beca use the units them selves have become different. They are no

longer similar segments weakly united by their resemblance. They have

instead become dissimilar parts of a society strongly united by their

differences. Durkheim's transformation of society is not rooted in dif-

ferentiation defin ed as a princ iple of separation; it is rooted in the dif-

ferences of the parts.

How do simple societies become complex ones? Simple and compl ex

societies are organized according to oppos ite principles. Complex soci-

eties grow out of simple socie ties and must overcome them . In Dur-

kheim's view, the "segmental arrangement is an insurmountable obstacle

to the division of labor" (1893:256; if 182-185). A structural transfor- mation, the replacem ent of one principle of order by another, is nothing

short of a revolution in social life. How might such a revolution be

produced and conducted? Revolution breaks out when the "growth and

·condensation of societies ... necessitate a greater division of labor." Pop-

ulation grows; the struggle for existence becomes acute; war breaks out

326 Kenneth N. Waltz

and becomes more violent the harder the population presses on the

resources available. The more acute the struggle, the greater the social

progress. The stronger segments of the old society forge ahead at the

expense of the weaker. The "triumphant segmental organs" take on the

vaster tasks of society; the losers are left with the lesser ones ( 189 3:262- 272). The division of labor develops in the struggle and flourish es within

the newly integrated society. Durkheim fac es up to what revolutions

ordinarily entail-fierce and bloody struggle with risks high and out- comes always uncertain .

A few points r equire emphasis. First, the social segments of the old

society must begin to break down before the division of labor can appear

(1893:256). Because the new social type rests on such different princi - ples, "it can develop only in proportion to the effacement" of the pre-

ceding society (1893:182). The transformation of the old structure begins, as Durkheim emphasizes , only when the segments of a m echanical

society are thrown upon one another, only when their fates become

tightly entangled beca use of the intense pressures of the struggle for

survival ( 1893 :266). Dynamic density consists not only in economic transactions but also in social ones. It is "a function of a number of

individuals who are actually having not only commerce but also social

relations, i. e. , who not only exchange services or compete with one

another but also live a common life:' The paragraph is worth completing.

"For, as purely economic relations leave men es tranged from one another,

ther e may be continuous relations of that sort without participation in

the same collec tive ex iste nce. Business carried on across the frontier s

which separate peoples does not abolish these frontiers" (1895:114). The transformation of social structure is not produced by the mere mutualism of international trade.

Second, the change in property relations that Ruggie identifies with

the shift from medieval to modern politics is significant. It is not for

Durkheim, nor I should think for most systems theorists, a transfor-

mation of structure. The change Ruggie identifies does not move inter-

national society from a condition in which like units are weakly held

together by their similarities to one in which unlike units are united by

their differences. Only a structural transformation can abolish the inter-

national imperative-take care of yoursel£1-and r eplace it with the domestic one-specialize!

A RESPONSE TO MY CRITICS 327

The redefinition of property relations, however important it may be,

cannot produce a transformation of the international system. Dynamic

density may have reshaped the structure of property rights, but that took

place within different societies, as Ruggie himself says (pp. 148-150). The point is important since it is often thought, wrongly, that any change

having widespread repercussions must be a structural change or even a

systemic transformation. Ruggie says that he would be surprised if the

changes he alludes to--demographic trends, resource constraints, and

the like-"do not adversely affect the managerial capacity of bipolarity

and, thereby, alter systemic outcomes " (p. 151 ). So would I. I would be surprised if many different sorts of unit-level changes did not alter

systemic outcomes. Ruggie says that I omit such forces. Yet I define a

system as consisting of a structure and of interacting units. The question

is not one of omission but of the level at which one sees such forces

operating. Changes in some or in all of the units will make their relations

harder, or easier, to manage. I might add to Ruggie 's cogent example a

still stronger one: the nuclear revolution in military weaponry. In my

view, the two biggest changes in international politics since World War II

are the structural shift from multi- to bipolarity and the unit-level change

in the extent and rapidity with which some states can hurt others . Surely,

the second change , like the first one, has system-wide effects. Wars that

might bring nuclear weapons into play have become much harder to start.

One must be struck by the fact that over the centuries great powers

have fought more wars and minor states have fought fewer. The frequency

of wali has correlated less closely with the attributes of states than with

their international standing. Standing, of course , is a structural charac-

teristic. Ye t because of a change in military technology, a change at the

unit level, waging war has more and more become the privilege of poor

and wea k states. A unit-level change has much diminished a structural

effect.

Ruggie has described some forces that "adversely affect the managerial

capacity of bipolaritf' Unfortunately, in doing so he mingles structural

and unit levels. Greatly reducing the odds that war will occur among

the great and the major powers represents a profound change in the

· quality of international life , such as only a structural transformation would

ordinarily bring. Just what is, and what is not, changed by the nuclear

revolution in weaponry? What has changed is this: The international

328 Kenneth N. Waltz

system has become more peaceful, at least at the top . Since the nuclear

reyplution, states relate to one another differently, yet each state still has

to take care of itself as best it can. Nuclear states continue to compete

militarily. The continuity of the system, lodged in its structure, accounts

for the latter. Nuclear weapons in the hands of some of the states help

to account for the former. Nuclear states are loathe to use their most

powerful weapons except in the service of peace , that is, for the sake of

deterrence. The latter part of Ruggie 's otherwise penetrating essay mixes unit and structural matters.

Third, Ruggie is right in saying that for m e international structure is

not fully generative (pp. 13 5-136; 148-15 2 ). A "generative model of

structure" should not be expected to generate all that goes on within a

system. Ruggie thinks that structure should contain a logic that accounts

for its own transformation and believes that Durkheim's dynamic de nsity

r eflects "structural effects and aggregated unit-level processes " (fn.

45). Ruggie compounds unit-level and structural forces, thus illustrat-

ing how difficult it is to keep the level s .of a system consistently distinct

and separate. Careful though Ruggie has been, his attempt to explain

more through structure by increasing its content proves to be the first

step down the slippery slope toward reduction. I admire Ruggie 's fine

and rich account of the historica l transition from the medieval to the

modern state. The account, however, tells us nothing about the structure

of international politics. Durkheim did not confound the internal con-

dition_ of states with their external environment . Ourkheim did not think

of dynamic density as part of a system's structure. Dynamic density is

a unit-level condition that may burst the bonds of the old system and

break its structure apart. Far from thinking of unit-level processes as

"all product .. _ and ... not at all productive " (p. 151 ), I, like Durk-

heim, think of unit-level processes as a source both of changes in systems

and of possible changes of systems, hard though it is to imagine the

latter. Neither structure nor units determine outcomes. Each affects the other.

In defining structure, what then are the criteria of inclusion and

exclusion? Since not much goes into the definition of structure, a negative

injunction is appropriate. Asking whether something is important cannot

tell us whether it should be included or excluded . If all that is important

for a system were in its structure, then we could ignore the units of the

A RESPONSE TO MY CRITICS 329

system. We would be saying that structures are determinant, even while

knowing that politics is a problematic and uncertain arena of action.

Ruggie widens the criteria of inclusion by arguing that property relations,

because they affect the way states relate to one another, should be in-

cluded. His reasoning makes the criteria of inclusion infinitely expansible.

Nuclear weapons, as I have said, strongly affect how states relate to one

another. So do national ideologies. Surely totalitarian and democratic

states relate to one another differ ently than did the old monarchic states.

And one could go on and on.

Still, one might ask why the distribution of capabilities across states

should be included in the definition of structure and not other charac-

teristics of states that could be cast in distributional terms . The simple

answer is that an international - political system is one of self-help. In a

self-help system, states are differ ently placed by their power. States are

self-r egarding units. State behavior varies more with differences of power

than with differences in ideology, in internal structure of property re -

lations, or in governmental form. In sel f-help systems, the press ures of

competition we ig h more heavily than ideological preferences or internal

political pressures.

In effect, Ruggie is saying that power does not tell us enough about

the placem e nt of states in the system. He is rig ht, but he draws the

wrong conclu sion. Structures never tell us all that we want to know.

Instead they tell us a small number of big and important things. They

fo cus our attention on those components and forces that usually continue

for lopg periods. Clean and simple definitions of structure save us from

the pernicious practice of summoning new systems into being in response

to every salient change within a system. They direct our attention to the

units and to unit-level forces when the parti cularity of outcomes leads

us to search for more idiosyncratic causes than are found in structures.

The world of states is older than any state in it. Thinking only of the

modern state system, conventionally dated from 1648, today's states are

hardly recognizable when compared with their originals even where their

names survive from a distant time. Through all of the changes of bound-

aries, of social, economic, and political form, of economic and military

activity, the substance and style of international politics r emain strikingly

constant. We can look farther afield, for example, to the China of the

warring states era or to the India of Kautilya, and see that where political

330 Kenneth N. Waltz

entities of whatever sort compete freely, substantive and stylistic char-

acteristics are similar. Ruggie would elevate qualities of the units shared

by some or by all of them to the level of structure. When the units of

an anarchic system develop new qualities through changes of "property

rights;' of "social formation;' and of "state/society relations" (pp. 149-

152), or presumably through changes in the quality of weaponry, or

whatever, he would have us say that the system has been transformed.

Structures would the n no longer show us a purely positional picture of

society. Ruggie would lower the level of abstraction by adding to struc-

tures more information about the characteristics of units and of unit-

level processes. Structure, properly defined, is transposable (cf. Nadel 19 57: I 04-1 09). If we follow Ruggie 's advice, structure will no longer

apply to different realms, even where the arrangement of the ir parts is

similar. We sha ll also lose another advantage of structural approaches.

Elegant definitions of structure enable one to fashion an explanatory

system having only a few variables. If we add more variables, the ex-

planatory system becomes more complicated, as one sees in Ruggie 's

essay. Especially in its last part, theoretica l acuity gives way to rich and dense description .

PREDICTION, POWER, AND TH E TESTING OF TH EORIES

With Robert Keohane I have on ly a few disagreements, largely on ques-

tion s of emphasis. Contrary to his statement, I do not differ with him

over rationality, exce pt semantically. I prefer to state the rationality as -

sumption differently. My preference is based partly on fear that "ratio-

nality" carries the wrong connotations. Since making fore ign policy is

such a complicated business, one cannot expect of political leaders the

nicely calcu lated decisions that the word "rationality" suggests. More

significantly, my preference is based on the importance I accord, and

Keohane denies , to the process of selection that takes place in competitive

systems. In structural-functional logic, behaviors are selected for their

consequences (Stinchcombe 1968:85). I fail to understand why Keohane

thinks that selection does not work if the death rate of a system's units

is low (p . 173). Selection does take place more swiftly and surely

when death rates are high, as in a sector of the economy populated by

A RESPONSE TO MY CRITICS 331

small economic units . In oligopolistic sectors, the survival rate of firms

is higher. Their fortunes nevertheless rise and fall. The market shapes

behavior by rewarding some firms and penalizing others. Selection lessens

the importance of the rationality assumption, but because selection works

less well in oligopolistic than in competitive sectors, we need to know

more about oligopoli stic firms and are able to predict less surely from

market theory alone.

In the international-political system, states wax and wane even as the ir

death rates r emain low. In the international-political system, great powers

come and go, although not with great frequency. We should keep the

notion of "selection" in a position of central importance. Even though

constrained by a system 's structure, a unit of the system can behave as

it pleases. It will, however, fare badly if some of the other parties are making reasonably intelligent decisions. That some states imitate the

successful practices of others indicates that the international arena is a

competitive one in which the less skillful must ex pect to pay the price

of their ine ptitude. The situation provides enough incentive to cause

most of the actors to behave sensibly. Actors become "se nsitive to cos ts"

to use Shai Feldman's apt phrase, which for convenience ca n be called

an assumption of rationality ( 1982).

Keohane is surely right to emphasize, as I did, that with the aid of a

rationality ass umption one still cannot, from national interest alone, pre-

dict what the policy of a country might he. Any theory of international

politics requires also a theory of domestic politics , since states affect the

system's structure even as it affects them . This is why Snyder and Diesing

in their excellent study explore information process ing and decisionmak-

ing ( 1977). To do so is fully in accord with, rather than a departure

from, realist assumptions. Realist theory by itself can handle some, but

not all, of the problems that concern us. Just as market theory at times

r equires a theory of the firm, so international-political theory at times

needs a theory of the state. Yet some successful predictions can be made

without paying attention to states . We do not always need to hop quickly

from structure to states when looking for explanations. For example,

contrary to the expec tations of such an experienced statesman as Pres-

ident Franklin D. Roosevelt, realist theorists would surely have predicted

the collapse of the allied coalition upon the morrow of victory. Whether

alliances cohere or collapse depends more on external situations than on

332 Kenneth N. Waltz

internal characteristics of allies, as their contrasting wartime and peace-

ti;me behaviors indicate. The prediction follows from balance-of-power

theory. The absence of constraints on American policy for three decades

following the war, seen in the light of the same theory, was hardly the

anomaly that Keohane takes it to be. Overweening power gives a state

the opportunity to act beyond its narrowly defined interests and provides

incentives for others to try to catch up. Predictions can be made when

we can answer this question with some confidence: How would we

expect any state so placed to act? In 1950, the Peopl e's Republic of China

intervened in the Korean War, to the surprise of General MacArthur

among others. Yet a~ C hinese government see ing the forces of a great

power approach ing the Yalu border would almost surely have moved

militarily if it were able to. Keohane wrongly emphasizes the failure of

realist predictions while rightly emphasizing the limitations of the theory

when standing alone (pp . 18 2- 1 8 3 ).

Even when we have fail ed to predic t, theory st ill helps us to understand

and ex plain some things about the b6havior of states. States who lost

the ir great-power status in the course of the Second World War no

longer behave as they used to. We tend to think of the ir postwar preoc -

cupation with self, of their inclination to take a free or at leas t a chea p

ride by spend ing disproportionately little o n defense, of their pusillani -

mous behavior in the oil crisis following the Middle Eastern War of 1973,

as deriving from their political attributes . Instead, suc h behavior follows

mainly from the new structure of international politics. Not unexpec tedly,

the English historian A. J. P Taylor assigne d str uctura l effects to the unit

level and saw them as ca use. "For years after the second world war;' he

wrote, "I continued to believe that there would be another German bid

for European supremacy and that we must take precautions against it.

Events have proved me totally wrong . I tried to lea rn lessons from history,

which is always a mistake. The Germans have changed the ir national

character" Qune 4, 1976, p. 742). More perceptively, Roy Macridis saw

the importance of changed position. "To specu late about a Franco-Ger-

man war in the 1950s or 1960s;' he wrote , "is just as boring as it would

have been to contemplate a war between Sparta and Athens under the

Roman Empire" ( 1971 :14 3 ). If their national characters, and ours, have

changed since the war, it is largely because their and our international

positions have become profoundly differ ent.

A RESPONSE TO MY CRITICS 333

Not only have some states sunk in the international rankings while

others have risen but also a great power now is one of two instead of

five or more, as was true earlier. The United States and the Soviet Union

behave differ ently from such countries as Germany and Japan because

the latter are no longer great powers. The behavior of the United States

and of the Soviet Union is also different from the behavior of ea rlier

great powers. A grea t power that is one among many learns how to

manipulate allies as well as adversaries. Great powers have to accom-

modate some of their number in order to gain strength vis-a-vis others.

In dealing with near equals , they des ig n their policies to influe nce the

actions of others. In a crowded field, those who play the great-power

game well flourish; those who do not risk falling by the wayside. The

situation of the United States and of the Soviet Union is markedly

different. Their field is not crowded. The most telling illustration of the

difference is seen in the mutual depende nce of allies before and dur ing

World War II and in the relative ind ependence of the two alliance leaders

since , along with the dependence of their associates on them.

Keohane raises the question of properly defining powfT and holds my

definition to be insuffi cient. He is right. To define power in terms of who

affects whom more strongly, is, I think , a move in the right direc tion;

but I did not carry the definition ver y far. Although power is a key

conce pt in realist theory, its proper definition r emains a matter of con-

troversy. On the fungibility of power, however, Keohane and I differ.

Obvio usly, power is not as fungibl e as money. Not much is. But power

is ml;J ch more fungibl e than Keohane allows. As ever, the distinction

betwee n strong and weak states is important. The stronger the state, the

greater the variety of its ca pabilities. Power may be only slightly fungible

for wea k states , but it is highly so for strong ones. Did, for example ,

America 's failure to respond to economic pressures with economic or

military force show that the United States was unable to translate its

capab ilities into effective power, as some thought during the oi l embargo

of the early 1970s, or did it indica te that the United States, more nearly

self-sufficie nt than most countries, was not pressed hard enough to make

the effort seem worthwhile? Moreover, in many of the examples Keohane

adduces as evidence that Canada, a weak state, prevailed over the United

States, a strong one, I suspect that American officials hardly cared about

the outcomes or even noticed what they might be. The United States

334 Kenneth N. Waltz

has more levers to pull than other states do but need not always pull

them.

Keohane rightly criticizes some realists for assuming that states seek

to maximize power (pp. 173-174 ). He wrongly associates me with them

because I point out that a balance-of-power system works whether we

find states seeking only the minimum of power needed for security or

whether some of them strive for domination. Because the belief that

states do or should try to maximize power is guite widespread among

realists, I emphasized the error in a paragraph containing these sentences:

"Only if survival is assured, can states safely seek such other goals as

tranquility, profit, and powe r. Beca use power is a means and not an end,

states prefer to join the weaker of two coalitions. They cannot let power,

a possibly useful means, become the end they pursue" (p. 127).

How one should tes t theories is the only guestion on which Keohane

and I are far apart. In my simple, and perhaps simplistic, recipe for the

testing of theories, given in chapter I of Theory cf Internationa l Politics, I may have sounded like a "na·ive falsifi cationist ." I should like to correct

the impression.

I~ criticizing my comments on testing, Keohane says that I find it diffi cult to "state precisely the conditions under which coalitions will

change" (p. 172). I think it is impossible to do that. Beca use of the

impossibility of precise specification, balance-of-power theory, like most

theories in the social sciences, is diffi cult to tes t . We should therefore

apply a variety of tests. Keohane is bothered by my urging that we seek

to confirm theories as well as to falsify them . I do indeed depart from

Karl Popper, who insists that only efforts to falsify theories count as

legitimate tests. That may be a suitable way to go about testing, but it

is not the only way. Errol Harris argues that among natural scientists it

is a little used method . 4 Attempts to falsify theories are as problematic ·

as attempts to confirm them. Because of the interdependence of theory

and fact, we can find no Popperian critical experiment, the negative

results of which would send a theory crashing to the ground. The

background knowledge against which to test a theory is as problematic

as the theory itself. Popper understood the problem but passed over it

in various ways. In science there are no ultimate, or certainly true ,

statements. Therefore, no test is conclusive; in principle we should test

theories ad ir>jinitum. One way to avoid an infinite regress is to reguire

1

A RESPONSE TO MY CRITICS 335

not tests but testability. Another way is simply to limit tests on the

ground that we are not trying to prove a theory true, but only to disprove

a hypothesis (if. Popper 1935:47-50, 105, 278). Popper's famous example of falsification is the one black swan that disproves the thousands of

instances seeming to confirm the proposition that all swans are white.

But if the positive statement-this is a black swan-is not proved, the

hypothesis is not shown conclusively wrong. As Harris puts it: "We

already know that there is and can be no external body of fact-external,

that is, to all theory" ( 1970:35 3 ). As Popper once said:

Science does not rest upon solid bedrock. The bold structure of its

theories rises , as it were, above a swamp. It is like a building erected on

piles . The piles are drive n down from above into the swamp , but not

down to any natural or "give n" base; and if we stop driving the piles

deeper, it is not because we have reached firm ground. We simply stop

when we are satisfied that the piles are firm e nough to carry the struc-

ture, at least for the time being (1935:111).

A theory may help us to understand and explain phenomena and

events yet not be a useful instrument for prediction . Darwin's theory of

evolution predic ted nothing. It did help mightily to explain a changing

world. That theories are not merely instruments for pre diction opens the

way for confirming tests. The inconclusive sta tus of fal sifi cation invites

us to try other means. Keohane chides me for advocating efforts to

confirm theories and at the same time admitting that confirming in -

stances can always be found (p. 172). Beca use they ca n be, I insisted

on the importance of making tests diffi cult. This strikes me as being

wise counsel since the lower the prior probability of a new piece of

evidence, the higher the increased confirmation. -Moreover, sound testing

does not reguire one to examine "a universe of cases;' as Keohane would

have it (p . 172). A small number of cases well studied may be worth

hundreds cursorily treated. Insofar as the accumulation of a number of

cases is "mer e repetition ... it does not serve to enlighten:' Each ob-

servation, as Harris has written, is valuable only if it "supplies new

information, offers a fresh clue to the form of the total structure"

( 1970: 348). One experiment well designed, one demonstration well con-

ducted, one case carefully examined, may add more to one's confidence

in a theory than hundreds of instances looked at hastily.

336 Kenneth N. Waltz

Testing theories is difficult; interpreting the results of tests is a subtle

task.' Since results are always problematic, some part of the scientific

community has to decide whether enough of an empirical warrant exists

to give a theory credibility. Theories gain credibility in a variety of

ways-by unsuccessfully attempting to falsify, by successfully attempting

to verify, by demonstrating that outcomes are produced in the way the

theory contemplates, and by the intellectual force of the theory itself.

Keohane laments realism's lack of a theory of peaceful change, and

he calls for more emphasis on norms, institutions, and non-state actors.

The structure of a self-help system is defined in terms of its principal

actors . The definition does not exclude other components, but m erely

sets the context of their existence. Empirical and theoretical work often

proceeds without consideration of how the structure of the system affects

institutions and actions within it. Some states sometimes want to work

together to secure the be nefits of cooperation. Cooperative projects in

the prese nt may lead to more cooperation in the future. But self-help

systems do make the cooperation of parties diffic ult. As Gilpin puts it,

"the traditional insights of realism .. . helps us to explain ... the ongoing

retreat from an interdependent world" (p. 312). Rules, institutions, and

patterns of cooperation, when they develop in sel f-help systems, are all

limited in extent and modified from what they might otherwise be .

Whether Keohane's conclusion, that Th eory if International Politics offers nothing new, is valid depends on one's view of the old (p. 17 5). The

behavioral mode of thinking is deeply ingrained in students of interna -

tional politics. Whether by liberals, Marxist s, rea lists, or behavioralists,

the attempt has usually bee n to ex plain outcomes through the varying

attributes of the acting units. System and structure have becom e fairly

common terms in political science discourse . Only in the most general

way, however, had sys temic approaches been us ed to show how a structure

shapes and shoves the units of a system. The effects on units of changes

in structure had not been identified and examined . I developed a way of

thinking that had not been widely familiar. Ashley is partly right; in

certain important ways I did break with earlier realist thinking. Keohane

is partly right; ther e is more continuity between earlier realists and me

than Ashley noticed. Earlier realists thought of international anarchy

simply as setting problems for statesmen different from those to be coped

with internally and as altering standards of appropriate behavior. While

A RESPONSE TO MY CR ITICS 337

earlier r ealists talked about the anarchy of international politics as mark-

ing their field of inquiry, they continued to explain international political

outcomes in terms of the aims and policies, the actions and interactions,

of states and of non-state actors . This remained the dominant pattern of

explanation even for those political scientists who began to use the terminology of systems theory.

"PROBLEM - SOLVING" TH EORY

I find Ric hard K. Ashley difficult to deal with. Reading his essay is like

entering a maze. I never know quite where I am or how to get out. He

is sometimes elusive, shifting from one view to another. In one essay, he

quotes m e as saying that "the behavior of states and statesmen ... is

indeterminate :' He then correctly observes "that there is room for prac-

tica l action: Practical realism has partial autono"!Y" (1981 :220, 222). In an-

other essay, neor ealists, apparently including me, are said to "grant to

the 'international politica l system absolute predominance over the parts'"

(p. 288). In the ea rlier essay, he lumps Morge nthau and me pretty

much together and tars us with the same brush . In the later essay, he

find s more virtue in Morgenthau and less in me. Some of Ashley's com -

me nts bewilder me. Like John Herz, I often fail to recognize myself in

what he writes about me (Herz 1981 :237). In preceding pages , I attended

to some of Ashley 's complaints. Others are covered in Robert Gilpin's

spirit~d respons e to Ashley 's indictment of neor ea lism. I shall avoid rep- etition .

In Ashley 's alternative model of international politics, the balance -of-

power regime "produces sovereign states, who as a condition of their

sovereignty, embody the regime" (above, p. 294, his italics). What can

this m ean? It may mean that states as we know them behave in certain

ways because self-help systems strongly encourage some modes of be-

havior and discourage others. States develop along certain lines and ac-

quire certain characteristics in order to survive and flourish in the system.

In a differ ent regime states would be different and would behave differ-

ently. But that seems unlikely to be his meaning because it would bring

Ashley into agreement with me on a fundamental proposition. Yet I can

see only one other way to interpret the sentence: namely, that the balance-

338 Kenneth N. Waltz

of-power regime antedates the units that engage in the balancing! I find

this baffiing, yet it does seem to be his meaning. Thus a major charge

he brings against me is that I understand international structure

not as a deep, internal relation prior to and constitutive of social actors but as an external joining of states-as-actors who have precisely the bound- aries, ends, and self-understandings that theorists accord them on the basis of unexamined common sense. In tum-and here is the coup-- Waltz grants the structure a life of its own independent of the parts, the states-as-actors; and he shows in countless ways how this structure limits and disposes action on the part of states such that, on balance, the structure is reproduced and actors are drawn into conformity with its requisites. But how is the independence of the structural whole estab- lished? It is not established independent of the parts taken together, for it is never anything more than the logical consequence of the parts taken together (p. 287).

This is a pretty fair summary, requiring only a few qualifications. The

main one is this : The structure is not "independent of the parts, the

states as actors;' but constantly interacts with them . Neither the structure

nor the units determine , as Ashley seemed to realize in 1981 and had

forgotten by 1984.

The root of our differences is exposed in Robert Cox's nice distinction

between problem -solving theory and critical theory. Critical theory deals

with the "continuing process of historical change" ( 1981: 128; r eprinted

above, p. 209). Both Cox and Ashley seem to think that my big mistake

was to concentrate on the first kind of theory instead of on the second.

I have no quarrel with Cox 's concern with counter and latent structures,

with historical inquiry, and with speculation about possible futures . Ash-

ley and Cox would transcend the world as it is; meanwhile we have to

live in it. At the end of his essay Cox speculates about emerging world

orders. The likelihood of their realization will vary not only with changing

production processes and social forces, which he emphasizes, but also

with distributions of capability across states, which I emphasize.

Next to my whole enterprise being misconce ived , what bothers Cox

and Ashley most is my assumption that states are the units of international

politics . They see this as enshrining the state, as freezing the system,

and as making it static and eternal. Even though I made abundantly clear

that I take the state to be a unit by assumption, Ashley says this: "For

A RESPONSE TO MY C RITICS 339

despite its statism, neorealism can produce no theory of the state capable

of satisfying the state-as-actor premises of its political theory"

(p. 279). I have not tried, but surely some neorealist is capable of

producing a theor y of the state. It would reveal, among other things,

one that we already know : The state in fact is not a unitary and purposive

actor. I assumed it to be such only for the purpose of constructing a

theory. In r eality, as I put it: "States pursue many goals, which are often

vaguely formulated and inconsistent . They fluctuate with the changing

currents of domestic politics , are prey to the vagaries of a shifting cas t

of political leaders, and are influenced by the outcomes of bureaucratic

struggles" (p. 118).

Because I concentrated on "problem-solving" theor y and left "critical

theor y" aside, I had to introduce theoretica l concepts. Neither Cox nor

Ashley likes the assumption that states are the units of the system. Yet

if one is to develop a problem-solving theory about anything, assumptions

of this sort have to be made. The alternative is simply to esc hew such

theories altogether. Would we then know more or less about the social

and the natural worlds? In developing a structural theory of internationa l

politics , I was mos t influenced by economists and anthropologists; spe-

cifi cally by microcconomic or market theory and by Emile Durkheim.

Obviously, economi sts in assuming that firm s act as maximizing units

know that firm s in fact do not conform closely to the conception . They

know further that forms and modes of economic organization change

over the years. The distinction between an assumption and a statement

striving for descriptive accuracy should be easy to grasp.

Cox 's and Ashley 's main objection seems to be that I did not write a

theory of domestic politics. Ashley notices that I allow for considerable

variety among states but complains that the variety is not a concern of

my theor y (p. 266). That is so, but only because I essayed an inter-

national-political theory and not a domestic one. Not everything need

go into one book and not ever ything can go into one theor y. 5 Realizing

that I did not write a theory of the state, Ashley and Cox cannot from

a theoretical assumption rightly infer what notions I might entertain

about the historical origins and development of states and about their

possible fates. I find it hard to believe that anyone would think that states

will remain fixed in their present condition. But Ashley seems to believe

that I hold that odd view. He infers what I believe about states from

340 Kenneth N. Waltz

what I did not say about them. In the book he criticizes I abstracted

"fr0m every attribute of states except their capabilities" (p. 94 ). The

behavior and practice of states and of statesmen are omitted from inter-

national-political theory not because of the ir unimportance but because

their exclusion from the system's structure requires a distinct theory

dealing with the politics and policies of states. I see something problem-

atic about this only for those who think that domestic and international

politics must be combined in one theory. Someone may one day fashion

a unified theory of internal and external politics. Critical theory appar-

ently aspires to do so, hut I read in Ashley and Cox only what such a

theory might do rather than what the theory is. The theor etical sepa-

ration of domesti c and international politics need not bother us unduly.

Economists get along quite well with separate theories of markets and

firms . Students of international politics will do well to concentrate on

separate theories of internal and external po litics until someone figures

out a way to unite them. Ashley accuses me of excluding history from the study of international

politics and of immunizing a part of my theory from falsifying tests. In

his words: "Despite neorea lism's much ballyhooed emphasis on the role

of hard falsifying tests as the measure of theoretical prog ress , neor ealism

immunizes its statist commitments from any form of falsifi ca tion"

(p. 270). To exclud e hi story from a problem-solving theory is hardly

to enjoin the historical study of politics. How can one incorporate history

into the type of theory I constructed? Ne ither Ashl ey nor Cox gives an

answer, but again apparently I should simply not have done what I did

but something quite different. Nor does a theoretica l assumption about

states enjoin anyone from studying them empirically or from theorizing

about them. Ashley finds in neorea lism "a historicism qf stasis. . . . a historicism that freezes the politica l institutions of the current world

order" (p. 289). How can any theory have these effects? Ashley has a

higher regard for the power of theories than I have. A theory applies

only so long as the conditions it contemplates endure in their essentials.

If the anarchy of international politics were to give way to a world

hierarchy, a theory of international politics would become a theory about

the past. Ashley thinks I am reluctant to contemplate "a hierarchy cen-

tering power within the grasp of a singular hegemon" because such a

concentration would overturn the "fundamental organizing principle of

A RESPONSE TO MY CRITICS 341

international politics" (p. 277). How could he know this was my reason

without asking me? It is true that some of Ashley's neorealists like

hegemony better than I do, but that simply illustrates Gilpin's statement

that we are a mixed bunch. The influence behind my preference is partly

Immanuel Kant and partly Reinhold Niebuhr. Kant feared that a world

government would stifle liberty, become a terrible despotism, and in the

e nd collapse into chaos. Niebuhr drew the conclusion from his dim view

of human nature that domestically and internationally the ends of security

and decency are served better by balanced than by concentrated power.

I distrust hegemonic power, whoever may wield it, because it is so eas ily

misused.

Ashley offers an alternative to a neorealist model (pp. 294-297). It

looks quite a bit like mine, except that it is wrapped in a ca pitali st

blanket . Does he mean to imply that "a balance-of-power regime" would

not exist in, say, a world of socialist states? If this is his m eaning, I flatly

disagree. Or does he mean that "a balance-of-power regime" would ex ist

but with some differences in the attributes and behaviors of states? No

doubt ther e would be some differences, but not ones of much systemic

importance. Balance-of-power politics in much the form that we know

it has bee n practiced over the millennia by many different types of

political units, from ancient China and India, to the G reek and Italian

c ity states, and unto our own day.

Critica l theory seeks to interpret the world histo rica lly and philo -

sophica lly. Problem-solving theory seeks to understa nd and explain it.

Ashler.:'s critica l essay revea ls to me no clue about how to write an

improved theory of the latter sort. I am sorry that it does not.

CONCLUSION

Systems theory is frequ ently criticized for being static. It is in one sense

but not in others. Self-ordered systems, as Michael Polanyi wrote , are

subject to dual control, that is, "control in accordance with the laws

that apply to its elements in themselves , and . .. control in accordance

with the laws of the powers that control the comprehensive entity formed

by these elements" ( 1968:1311 ). The latter are the constraints of the

system, with the organization of the parts affecting the way the parts

342 Kenneth N. Waltz

function. International-political systems exhibit dual control. Behaviors

aRd outcomes change as interactions among a system's units become

sparser or denser, as alliances shift, as nations adapt their policies to one

another: These are changes within the system, and often systems dynamics

are identified with, and limited to, such changes. What really matters,

it seems, are changes in the behavior of states and in their alignments.

That is the whole of the story only if dynamic is defined as energy in

motion. This is the dynamic of the units. Another part of the story is

revealed if c!ynamics are thought of as in physics-the action of forces on

bodies in motion or at rest. These are the dynamics of the sy~tem.

Structur~l changes alter a system's dynamics. Systemic effects cannot be reconstructed from the system's interacting parts since the parts behave

differently because they are parts of a system. The constraints and in-

centives of a system, its dynamics, change if its structure changes or is

transformed. To explain outcomes, we have to look at a system's dynamics

and not just at the c haracteristics and the strategies of the units .

Self-help systems are transformed if' their organizing principle shifts from anarchy to hierarchy. Establishing a world government would do

this. Either all states pooling their sovereignty or some states conguering

the others would transform the system. Self-help systems change through

conseguential variation in the distribution of capabilities among their

m embers. States can more readily change their system than transform

it. States fighting wars have immediate offensive or defensive aims . If the

aim is to red uce the number of great powers significantly through con-

guest, then from a system's perspective the aim is to change the system.

The result may be produced aside from the intentions of states. The

victors in World War II thought of themselves as fighting a defensive war.

In doing so, they nevertheless fought a war that changed the system from

one of multipolarity to one of bipolarity. Shortly after helping to produce

that change, the United States began to promote another one by en-

couraging Western European countries to unite. From the unit's per-

spective, one easily understands why. A united Western Europe would

be a bulwark against the Soviet Union. From a system's perspective, the

aim is an odd one. Why should one of two great powers wish to move

the system from bipolarity to tripolarity?

By rewarding behavior that conforms with systemic reguirements and

punishing behavior that does not, a system's structure works against

A RESPONSE TO MY CRITICS 343

transformation. This is fortunate only if we are content with the system

we have or if we are pessimistic about the costs and consequences of

transformation. Changes in, and transformation of, systems originate not

in the structure of a system but in its parts. Through selection, structures

promote the continuity of systems in form; through variation, unit-level

forces contain the possibilities of systemic change. The possibilities of

rising in the international system, and the costs and benefits of doing so,

vary as systems change; but states decide whether making the effort to

advance is worthwhile. Japan has the capability of raising herself to great-

power rank, but has lacked the inclination to do so. Systems change, or

are transformed, depending on the resources and aims of their units and

on the fates that befall them. Structures condition behaviors and outcomes, yet explanations of be -

haviors and outcomes are indeterminate because both unit-level and

structural causes are in play. Systems are stable if they endure for long

periods. They are impressively stable if they survive the disruption of

large-scale wars. The bipolar world has been both stable and peaceful-

if peace is defined as the low incidence of war among great and major

powers, that is, among those states most immediately affected by the

structure of global politics. The bothersome limitations of systemic ex-

planations arise from the problem of weighing unit-level and structural

causes. To what extent is an effect to be ascribed to one level or the

other? One may believe, as I do, that both bipolarity and nuclear weapons

promote peace. But one cannot say for sure whether the structural or

the unit-level cause is the stronger. The difficulty of sorting causes out

is a serious, and seemingly inescapable , limitation of systems theories of

international politics. Structures shape and shove. They do not determine behaviors and

outcomes, not only because unit-level and structural causes interact, but

also because the shaping and shoving of structures may be successfully

resisted. We attribute such success to Bismarck when we describe him

as a diplomatic virtuoso. The unification of Germany, fashioned in the

fighting of three short wars, shifted the balance of power in the center

of Europe. Could that be done without igniting a general war? Given

the structure of European politics, few would have thought so before the

event. Later, in 1879, Bismarck forged a long-term alliance with Austria-

Hungary. Given the structure of European politics, one would have pre-

344 Kenneth N. Waltz

dieted that an alliance made by two great powers would cause a coun-

teralliance to form . Indeed, the Franco-Russian Alliance may have become

inevitable with Prussia's victory over France in 1871, as Friedrich Engels

thought ( 1890:48-49). The measure of Bismarck's diplomatic virtuosity

is that the Alliance was not made until 1894, four years after his political

demise. With skill and determination structural constraints can some-

times be countered. Virtuosos transcend the limits of their instruments

and break the constraints of systems that bind lesser performers.

Thinking in terms of systems dynamics does not replace unit -level

analysis nor end the search for sequences of cause and effect. Thinking

in terms of systems dynamics does change the conduc t of the search and

add a dimension to it. Structural thought conceives of actions simulta-

neously taking place within a matrix. C hange the matrix-the structure

of the system-and expected actions and outcomes are altered. The past

causes the present , but the causes are mediated by the present system's

structure. The examination of structure tells us how a system does what

it does. A structure sets the range of ·expectations. Like any theory, a

structural theory leaves some things aside in order to concentrate on

others. Like any theory, a structural theory is limited to making predic-

tions and promoting the understanding of events at a level of generality

appropriate to the theory. A structural theory of international politics

can fix ranges of outcomes and identify general tendencies , which may

be persistent and strong ones but will not be reflected in all particular

outcomes. We cannot hope to predict spec ific outcomes, but if o ur theory

is good, we will see the kind of behavior and record the range of

outcomes the theory leads us to expect. From the dynamics of the system

we can infer general properties of behavior and outcomes within a system

and expected changes in those properties as systems change.

NOTES

I. This essay has benefited from critical readings by my wife, from correspon- dence and conversation over th e years with Barry Buza n, Robert Keohane, and Glenn Snyder, and from searching and sugges tive comments by Vinod Aggarwal.

2. States gain from trade beca use of their dissimilarities . Durkheim dwells on the r esemblance of the units of a mechanical society. This may be confusing. One needs to remember that like units are not identical. They may differ considerably

A RESPONSE TO MY CRITICS 345

in resource endowment and economic development. Functionally, however, they are

like units. 3. Cf. Park: "People live together on the whole, not because they are alike but

because they are useful to one another" (I952:80). See also March and Simon: "the greater the specialization by subprosrams (process specialization), the greater the in-

terdependence a mons orsanizacional subunits" ( 1958 : 159). 4. Excellent examples of test by confirmation, mainly from Newton, Lavoisier,

and Harvey, are found on pp. 161-178 of Harrris ( 1970). 5. In a book about foreign policy I concerned myself at length with the e ffect s

of internal structural differences on the ex ternal polic ies and behavior of the United

States and Britain ( 196 7).