As the essence of deterrence is credibility, which is a subjective concept, the stability of deterrence depends on all kinds of delicate perceptual and psychological processes which might lead to gross miscalculation; if the threat to retaliate is underestimated, deterrence fails and an unintentional nuclear war would be the consequence. The system of deterrence is generally drifting towards becoming unstable, as indicated by the concern for scenarios like "Called Bluff". The assumptions of rationality underlying deterrence strategies are not justified in every situation. The North Atlantic Treaty Organization concept of deterring conventional aggression by the threat of a selective nuclear strike is constantly affected by doubts about its credibility and reliability. This concept also seems questionable because a limited nuclear war in Europe might destroy just what it is supposed to protect; these doubts are supported by the former Soviet Union's refusal to accept the concept of limited nuclear war. The North Atlantic Treaty Organization concept of extended deterrence (United States nuclear guarantees for European allies) also suffers from problems of credibility, thus aggravating the risk of miscalculation.
The two credibility problems (credibility of nuclear deterrence against conventional attack and credibility of USA guarantees for Europe) add up to one large credibility problem which serves to reinforce each separate problem. This creates regional crisis instability; and may create a temptation to engage in low-key 'probes' so as to assess the seriousness of commitments made within the framework of extended deterrence. The question whether there can be such a thing as a 'limited nuclear war', not necessarily escalating into an all-out nuclear war, is extremely controversial. There is at least some likelihood that the 'tactical' use of nuclear weapons would inadvertently trigger a full-scale nuclear exchange. Very serious risks of miscalculations emerge in third world regions, where the two major powers have made a multitude of poorly defined commitments in the sense of extended (mainly conventional) deterrence.
The risk of unstable deterrence which constitutes the very essence of strategic or crisis instability is currently being aggravated and multiplied by a series of strategic developments at both the global and regional levels. Doubts regarding the credibility of deterrence postures are increasing for a multitude of reasons, and this is conducive to miscalculation or inability to prevent miscalculation by the opponent. Nevertheless, credibility, although not absolute, is sufficient to stave off any failure of deterrence. Yet this situation of relative stability is not safe for an indefinite future. It also has a propensity to generate low-key crises of all kinds which in turn imply a risk of escalation.
For the foreseeable future, no nuclear power can rely with certainty on escaping any retaliations if it tries to disarm its opponent by a first strike, unless it has a completely insane risk-taking behaviour overriding any rational considerations which, in principle, will prevent gross misjudgement. The same can be said of the specific strategic situation prevailing in Europe, yet the prospects for continuing credibility of deterrence may be less far-reaching in time. Credibility is not absolute but sufficient. The uncertainties about possible disastrous consequences of a 'probe' of the deterrence postures in Europe are mutual: the Soviet Union is not sure about its chances of escaping USA retaliation. The USA is not sure about an immediate escalation into an all-out nuclear war by the Soviet Union, once it employs nuclear weapons. This does not exclude, however, low-key 'probes'.
Perhaps – as a growing school of thought in international relations theory argues – the precarious stability of a standoff is the best most regions can manage. But the mounting tensions between India and Pakistan should give pause. Stable deterrence is easy in theory, extraordinarily demanding in practice.