37
goal in this paper is to stimulate further efforts to develop methods and bodies of evidence
appropriate to the specification and evaluation of possibilistic arguments.
The debates about the causes of war and peace that Jervis addresses also help illustrate
the difference between probabilistic and possibilistic arguments. Many critics of realism pointed
to the peaceful end of the Cold War as disproving a number of realist theories. Defenders of
realism have responded in one of two ways: some have tried to develop convincing realist
explanations for the end of the Cold War, while others have suggested this is just one data point
and such a single case cannot falsify a theory that has been accurate in many other cases. Both
responses could be valid rejoinders if one is treating realism as a body of probabilistic causal
hypotheses. When cause-effect relations are questions of probability rather than iron laws, one
case that is inconsistent with the hypothesis does not disprove it, since the theory implies there is
some probability that any given case will not turn out as the theory predicts. Aggregate data
provides a more compelling test than any single case. Similarly, one could defend the theory’s
usefulness as a source of probabilistic explanations if one could point to some configuration of
causal variables that are part of the theory’s repertoire and can explain the case in question. If
critics meant to suggest the end of the Cold War disconfirmed realism as a causal theory, this
inference can be challenged by the responses just described.
If one is focusing on possibilistic propositions, however, the realist rejoinders just
outlined do not rescue the realist case. In the realm of possibility, realism implied that two great
power rivals for hegemony would not end their rivalry peacefully, especially at a point when
their power capabilities had become relatively equal. The peaceful end of the Cold War is again
proof by existence that it is possible for bitter rivals to transform their relationship. The dramatic
changes in Argentine-Brazilian relations in the 1980s show the same transformation is possible
in the context of a rivalry for regional hegemony.
Such transformations could potentially be due, at least in part, to realist causes. If so,
however, that would only reinforce the point I am making here. It would show that causes that
have probabilistic effects do not necessarily determine the limits of the possible. Hypotheses
about what is and is not possible have to be evaluated in their own right, because inferences
derived from probabilistic explanatory theories will not always be correct. If even forces
emphasized by realism can sometimes lead serious adversaries to put their security competition
Convention