Some time ago, in the pages of the New Left Review, a claim was made on behalf of ‘rational-choice Marxism’ as ‘a fully fledged paradigm, which deserves to take its place beside the two other constellations of theory currently discernible within the broad spectrum ofprogressive social thought—namely, post-structuralism and critical theory’.footnote footnote1 More than that: ‘it is now only within the rational-choice context that some of the leading items on the classical agenda of Marxist theory—historical explanation and the delineation of social form, the collective dynamics of class struggle, the evolution and evaluation of capitalism—can be fruitfully discussed.’ These are very large claims, and if this new ‘paradigm’ can even partially live up to them, it deserves the vogue it is now enjoying in the Anglo-American academy. A theoretical advance in any one of the ‘leading items on the classical agenda of Marxist theory’ would be a worthy accomplishment; but it would indeed be a remarkable achievement if, even without driving any likely competitors from the field, this body of thought could be shown to merit the status of a ‘fully fledged paradigm’, a comprehensive theoretical ‘constellation’ with the explanatory range of classical Marxism.

There is a difficulty at the outset in evaluating the claims of rcm. What are to be our criteria of inclusion? Alan Carling tells us that ‘if there is one distinctive presupposition of the intended body of work, it must be the view that societies are composed of human individuals who, being endowed with resources of various kinds, attempt to choose rationally between various courses of action.’footnote2 On the face of it, this ‘distinction’ places rcm in the company of a dauntingly large and heterogeneous collection of writers—only part of which is exhausted by Carling’s own admission that this same ‘distinctive presupposition’ is ‘a commonplace of that broad sweep of conventional economics and philosophy conducted in the liberal tradition.’ The further specification that the ‘dramatic’ difference lies in rcm’s ‘joining of this presupposition with the classical agenda of Marxist theory’ only complicates matters. To guarantee the comprehensiveness of this paradigm, its full coverage of the classical Marxist agenda, Carling is obliged to collect under its rubric a wide and disparate range of writers, at least some of whom would contest their own inclusion (he specifically acknowledges Norman Geras’s disavowal of the rcm label, though without apparently accepting this self-exclusion). It is not enough to include figures like John Roemer or Jon Elster, those who would most readily accept a ‘rational-choice’ designation which carries the associations of game theory and methodological individualism. To make good his claims for rcm in the areas of historical explanation in general and the evolution of capitalism in particular—themes which on any conventional reckoning are central to the Marxist agenda—Carling must recruit into the rcm school writers who evince a very different attitude to methodological individualism—notably Robert Brenner and G.A. Cohen.

In other words, the ‘distinctive presupposition’ must remain very broadly defined indeed (so broadly, in fact, that it is difficult to see why that most ‘classical’ of Marxists, Friedrich Engels—who, after all, had something to say about the formation of social patterns from the unintended consequences of individual human actions—could not be taken on board). To say that this paradigm is also distinctively characterized by an ‘analytic’ mode of presentation (the term ‘analytic Marxism’ is sometimes taken to be synonymous with rational-choice Marxism) hardly advances matters, since the ‘analytic’ style of argumentation is in principle compatible with any set of substantive propositions. Alternatively, if the ‘distinctive presupposition’ is further specified, if rcm is distinguished by its game-theoretic methodological individualism, then the most distinctive characteristic of this paradigm—the formal abstraction and the static, ahistorical individualism of the rational choice model—is the one which is least congenial to those central items of the Marxist agenda: social change and historical process, and in particular, the ‘laws of motion’ specific to various modes of production, their characteristic crises, the specific principles of motion in the transition from one mode to another.

Carling himself seems aware of the difficulty when he acknowledges that Roemer’s theory of exploitation provides no substitute of its own for the ‘classic’ Marxist theory of history but remains ‘parasitic’ on that general theory (as interpreted by G.A. Cohen).footnote3 Indeed, Roemer makes no pretence that his distinctive rational-choice paradigm, which he offers as a substitute for the classical theory of exploitation based on the labour theory of value, can generate its own theory of historical process. What he offers us instead is a hybrid. As Carling puts it in his recent review of Roemer’s latest book Free to Lose (‘the textbook of contemporary Marxist theory’footnote4), Roemer’s ‘standard contemporary Marxism’ is ‘Cohen on history plus Roemer on class and exploitation’.

A sensible way to proceed, then, might be to begin with a specific definition of rcm, which recognizes the distinctiveness of its gametheoretic models and their methodological individualism, but without prejudging the capacity of this paradigm to enter into a fruitful alliance with a theory of social change and historical process from outside its distinctive methodological boundaries. This leaves us, in the first instance, with three obvious major candidates for inclusion: John Roemer, Jon Elster, and (lately) Adam Przeworski. The latter is included with some hesitation, because, although he sometimes carries the rational-choice model to extremes not envisaged even by Roemer, he is the least consistent of the three in his adherence to the paradigm, having done important political analysis which does not rely on the game-theoretic paradigm and which is, if anything, undermined by the model’s theoretical demands. In what follows, the focus will be mainly on Roemer, who has provided the clearest and most comprehensive account of the rcm paradigm, with excursions into Elster and Przeworski when they help to clarify some important point or explicitly depart from Roemer in some significant way. Erik Olin Wright will be excluded from consideration simply on the grounds that his theoretical formation appears to be in a critical stage of transition: his recent book Classes, which constructed a theory of class on the foundations of Roemer’s theory of exploitation, was almost immediately followed by an article co-authored with Andrew Levine and Elliott Sober which deflates the explanatory pretensions of methodological individualism practically to the vanishing point.footnote5 The connection of Brenner and Cohen to this core group remains to be established and will be explored when we come to the question of rcm’s treatment of history.

Roemer has written that ‘methodological individualism’, specifically in its ‘game-theoretic’ mode, is essential to the explication of historical materialism and its ‘key propositions’: ‘the key questions of historical materialism,’ he maintains, ‘require reference to the specific forms of class struggle, and . . . an understanding of such struggles is elucidated by game theory; . . . class analysis requires microfoundations at the level of the individual to explain why and when classes are the relevant unit of analysis.’footnote6 Let us say, then, that the differentia specifica of rcm is a theory of exploitation and class in which various modes of exploitation and classes are logically generated according to the principles of game-theoretic analysis. This by no means reduces rcm to a purely methodological strategy. On the contrary, as we shall see, the form of the rcm theory is to a large extent its substance, and in its gametheoretic assumptions are secreted vital substantive theses about the social world.

The elusiveness of rcm’s distinctive identity may have something to do with a certain lack of critical self-awareness among its practitioners. They are remarkably insensitive to the history and context of ideas in general, and their own ideas in particular; and they are generally inclined to remain within a very narrow universe of debate. rcmists, to judge by the extent of their mutually self-confirming footnoting and the restricted range of controversy encompassed by their writings, seem to talk largely to each other.