It is over 50 years since John Herz and Herbert Butterfield first drew scholarly attention to the 'absolute predicament', the 'irreducible dilemma': the security dilemma. For Butterfield, the security dilemma constituted 'the basic pattern for all narratives of human conflict, whatever other patterns may be superimposed on it later'. 1 While many other writers, including Herz himself, came to contest Butterfield's claim that the security underlies all human conflict, what is certainly not contested is the centrality of the concept in purporting to explain both international and domestic (ethnic) wars.
The utility of the security dilemma is in elucidating how actors, seeking more security, can nonetheless end up with less and provoke conflict in the process, be this competition and/or outright war. The 'tragedy' of this situation stems from the misperception that the other harbours malign intentions. For Butterfield, and other 'Classical' Realists, misperception is essentially the product of human fallibility, which prevents the individual from being able to enter into the other's 'counter fear'. For neo-Realists, the individual level of analysis is not necessarily inconsequential, but what is most crucial is structure: worst-case assumptions necessitated by the anarchical nature of the international system and its self-help corollary.
The emphasis on structure - rather than agency - seeks to provide further explanation as to why the tragedy of the security dilemma occurs. In doing so, however, the question of anarchy and self-help, together with other, additional elements introduced into the security dilemma, such as offence- defence considerations, has inevitably resulted in ever-wider interpretations of the concept. To be sure, I am not against what Jack Snyder and Robert Jervis acknowledge as 'concept stretching'. 2 Indeed, like Charles Glaser, I also recognise the utility of expanding some of the security dilemma's core assumptions.
The starting point for expanding the security dilemma is Jervis' distinction between the spiral model and the deterrence model. The difference between the two models essentially reflects actors' divergent intentions;that is, either