Analytical Marxism began with an act of subtraction. A small group of philosophers, economists, sociologists and political scientists, most of them formed inside the Marxist tradition and most of them still committed to its central questions about exploitation, class and historical change, concluded that one celebrated component of the inheritance had to go. The component was dialectics. Not this or that dialectical text, not a particular codification associated with a party or a state, but dialectics as such: dialectics as a logic distinct from ordinary logic, dialectics as a method that licensed conclusions unavailable to plainer forms of argument, and dialectics as a set of laws said to govern nature, history and thought alike. The group’s members suspected that whatever was true in dialectical writing could be said, and said better, in the vocabulary of causal explanation, structural constraint and strategic interaction, and that whatever could not survive that restatement was confusion dressed in impressive terminology. That suspicion, pursued over roughly two decades of books, articles and annual arguments, produced one of the most instructive episodes in the intellectual history of Marxism, because it forced a question that looser traditions had been able to postpone indefinitely. What exactly does dialectical language add to an explanation, beyond the feeling of depth?

The question matters beyond the boundaries of one school because dialectics occupies a peculiar place in Marxist culture. For many readers it functions as a badge of fidelity, the element whose presence certifies that an analysis is genuinely Marxist rather than a borrowing from liberal sociology or neoclassical economics. For others it functions as an embarrassment, the element blamed when Marxist writing descends into incantation. Between those reactions sits a genuine intellectual problem that neither loyalty nor embarrassment can dissolve. Some claims expressed in dialectical vocabulary appear to name real features of social life: the way opposed positions define one another, the way quantitative drift issues in qualitative change, the way a social form generates pressures that undermine its own stability. Other claims expressed in the same vocabulary appear to name nothing at all, and to survive only because their vagueness protects them from refutation. A tradition that cannot sort the first kind from the second will keep both, and will pay for the second with the credibility of the first. The analytical Marxists proposed a sorting instrument of disarming simplicity. Take any dialectical claim, restate it without the dialectical vocabulary, and see what remains. If everything remains, the vocabulary was decoration. If nothing remains, the claim was empty. If something specific is lost, name it precisely, and the tradition will have learned what dialectics actually contributes. That instrument is called the translation test in this article, and it organises everything that follows.

Scope needs stating at the outset, because analytical Marxism was a large enterprise and this article owns one question inside it. The school rebuilt the theory of exploitation on new foundations, reworked the theory of history, intervened in debates about justice and equality, and quarrelled internally about functional explanation, rational choice and the status of its own founding texts. Those substantive positions have their own treatments elsewhere in this series, and they enter here only where they bear on the dialectics question. What this article traces is the founding refusal itself: why a group of Marxists, at the end of the nineteen seventies, decided that dialectics should be abandoned; what standards of explanation they put in its place; what their arguments won; what their critics, speaking for the dialectical tradition at its strongest, said was being destroyed; and how the dispute looks when both cases have been given in full. The verdict offered at the end is deliberately double. On the narrow question of dialectics as a logic, the analytical case is very strong and has never been answered on its own terms. On the wider question of what an adequate social theory must be able to represent, the defenders identified a real limitation in the analytical toolkit, and the school itself, in its later self-criticism, conceded more of that limitation than its founding manifestos would have predicted.

Analytical Marxism and the founding refusal

Every school has a founding gesture, and the founding gesture of analytical Marxism was a refusal of obscurity. The gesture can be dated with unusual precision. In 1978 G. A. Cohen published a defence of Marx’s theory of history written entirely in the idiom of Oxford analytical philosophy: definitions stated before use, distinctions drawn between claims that careless readers merge, arguments reconstructed premise by premise, objections entertained in their strongest available form. The book contained no dialectical machinery whatsoever, and its absence was not an oversight. Cohen had set out to show that the core of historical materialism, the claim that the development of productive power explains the rise and fall of economic structures, could be stated as a set of clear propositions and defended against clear objections, with none of the apparatus of contradiction, negation and sublation that textbook Marxism treated as indispensable. The demonstration was itself the argument. If the theory of history could be rebuilt without dialectics and emerge stronger, then dialectics was not the load-bearing element its defenders claimed, at least for that part of the inheritance.

Why do analytical Marxists reject dialectics?

Analytical Marxists reject dialectics because they hold that dialectical claims either restate ordinary causal or structural points in obscure language or else are confused. They demand explanations grounded in the actions and beliefs of individuals, and argue that no distinctive dialectical explanation has ever been produced that resists plain restatement.

The refusal widened quickly from one book into a collective programme. Jon Elster, a Norwegian philosopher and social scientist working across Paris and Oslo, arrived at a parallel conclusion from the direction of social science methodology, and his Making Sense of Marx, published in 1985, subjected the whole corpus to the same treatment Cohen had applied to the theory of history. Elster’s chapter on dialectics became the school’s most quoted polemic: a patient march through the standard dialectical theses, asking of each one whether it asserted something testable, something true but banal, or something that dissolved under inspection. John Roemer, an economist at Yale, approached from another flank entirely, rebuilding the concepts of exploitation and class inside formal models whose assumptions were printed on the page and whose conclusions followed by proof rather than by evocation. Around these three gathered a shifting company that included the sociologist Erik Olin Wright, the political theorist Philippe Van Parijs, the historian Robert Brenner, the economist Pranab Bardhan and others, meeting each September to argue. What united them was not a doctrine but a discipline: a shared insistence that Marxist claims should meet the standards of clarity and rigour that the best work in their home disciplines demanded, and a shared judgement that dialectical writing, as practised, systematically evaded those standards.

Precision about the target matters, because the school has often been misdescribed as rejecting Marx. The members rejected a way of writing and reasoning that had become attached to Marx, and they rejected it in the name of Marx’s questions. Cohen described himself as defending Marx’s theory of history, not burying it. Elster’s book, for all its severity, was an attempt to find what in Marx remained defensible, conducted by a scholar who had spent years inside the texts. Roemer’s models were offered as a reconstruction of Marxian economics that preserved its central intuition, that exploitation rests on unequal ownership of productive assets, while discarding machinery he judged defective. The quarrel was internal, and its bitterness had the character of internal quarrels: each side accused the other of betraying the inheritance, the analysts by amputating its distinctive method, the traditionalists by protecting indefensible claims from the scrutiny every other theory had to face. Understanding the episode requires holding both accusations in view, since each contains a portion of the truth.

A further clarification separates three things that the single word dialectics habitually fuses, because the analytical objection lands differently on each. Dialectics can mean a logic: a claim that reality itself, or thought about reality, obeys laws that override or supplement the ordinary prohibition on contradiction. It can mean a method: a style of inquiry that proceeds through opposed determinations, attends to process and totality, and treats categories as historically formed. And it can mean an ontology: a picture of the social world as constituted by internal relations, in which the terms of a relation are what they are partly because of the relation itself. The analytical Marxists attacked the first meaning root and branch, treated the second as a set of heuristic reminders that needed no special name, and largely ignored the third, which is why the defenders’ strongest reply concentrates on the third. Much confusion in the secondary literature comes from scoring the debate as if a refutation of dialectics as logic settled the fate of dialectics as ontology. It does not, and the sections below keep the three meanings separate throughout.

A school built around a September meeting

Schools of thought usually announce themselves through journals, manifestos or institutions. Analytical Marxism announced itself through a diary commitment. Beginning at the end of the nineteen seventies, a group of scholars began meeting once a year, in September, to read and criticise one another’s work in progress, and the September Group, as it came to be known, supplied the school with its social skeleton. The format explains a good deal about the school’s character. Papers circulated in advance, discussion was prolonged and unceremonious, and reputations earned in print offered no protection in the room. A claim that could not survive a weekend of hostile questioning from colleagues who shared its author’s broad commitments was unlikely to survive in the work that followed. The group thus institutionalised the very standard it preached: explanation as something that must withstand explicit, premise by premise challenge from informed and unsympathetic readers.

Membership overlapped with, but never exactly matched, the roster of any single discipline, and the mixture was the point. Cohen brought the techniques of analytical philosophy, honed on questions about freedom, equality and the interpretation of texts. Elster brought a social science formed by game theory, decision theory and a historian’s appetite for mechanisms. Roemer brought mathematical economics and a conviction that Marxian claims should be expressed as theorems or abandoned. Wright brought empirical sociology, survey instruments and a willingness to operationalise concepts that philosophers preferred to leave abstract. Van Parijs brought political theory and an interest in what followed for policy once the analytics were done. Around this core moved a wider circle: Brenner with his historically grounded political Marxism, Bardhan with development economics, Hillel Steiner with the theory of rights, Adam Przeworski with the comparative study of labour movements and political transitions, Samuel Bowles with the economics of contested exchange. No membership list was ever closed, and the group’s own later reflections treat its boundaries as porous. What held the company together was less a set of conclusions than an agreement about how conclusions ought to be earned.

The group gave itself a nickname, and the nickname has become the most repeated fact about it, because it compresses the founding attitude into four words. Members referred to their enterprise, with deliberate self-mockery, as Non-Bullshit Marxism. The phrase was not coined by critics and was never a marketing label; it circulated inside the group as a joke with a serious edge, and Cohen later repeated it in print when recounting the group’s origins. Its meaning deserves unpacking, because it is easily misread as mere arrogance. The target of the joke was a recognised style of Marxist discourse in which portentous vocabulary performed the work of argument: passages in which the invocation of totality, mediation or the negation of the negation allowed a writer to pass over the exact points at which evidence, definition or inference was required. Against that style the nickname asserted a house rule. Say what the claim is. Say what would make it false. Say how the parts connect. A member who could not comply had not been profound; the member had been evasive, and the group reserved its rudest word for evasion. Readers who find the phrase offensive usually discover, on inspection, that the offence lies in its accuracy about a certain kind of writing they have themselves struggled through.

Formation at the end of the nineteen seventies was not an accident of the academic calendar. The decade had delivered a sequence of shocks to the Western Marxist traditions within which the group’s members had been educated. The revelations and invasions associated with actually existing socialism had discredited the official dialectical materialism of the Communist parties, whose textbooks presented dialectics as a completed science of everything and wielded it as an instrument of discipline. The collapse of the postwar boom and the onset of prolonged stagflation had revived interest in Marxian economics while exposing the technical weaknesses of its inherited form. In France, the structuralist Marxism associated with Louis Althusser, which had promised scientific rigour of a different kind, was losing its hold, partly through political disillusion and partly through the sheer difficulty of its texts. A generation of scholars trained in the expanding postwar universities, fluent in the methods of analytic philosophy and modern social science, found itself in possession of powerful tools and an inheritance that seemed unable to use them. Analytical Marxism was one answer to that situation: keep the questions that made Marxism distinctive, submit every inherited answer to the new tools, and let the casualties fall where they fell. Dialectics fell early because it was, by the group’s standards, the least defensible item in the inventory.

It is worth pausing on the word analytical, since it misleads readers who associate it with a narrow philosophical school. In the group’s usage it signalled a commitment to the methods of modern analytical disciplines generally: conceptual clarity from philosophy, formal modelling from economics, mechanism based reasoning from the social sciences, and ordinary standards of evidence throughout. Some members were analytical philosophers by trade; others had no connection to that tradition and some suspicion of it. Roemer’s mathematics owed nothing to Oxford. Wright’s class analysis owed more to survey research than to linguistic analysis. The umbrella term covered a family of practices unified by a negative: opposition to modes of theorising, dialectical writing prominent among them, in which the appearance of depth substituted for the labour of specification. The full portrait of the school as a school, its phases, its internal divisions and its eventual dispersal, belongs to the companion treatment of analytical Marxism as a complete guide, and is deliberately not duplicated here. This article needs the group only as the collective author of one argument, the argument against dialectics, and it is to the intellectual sources of that argument that attention now turns.

One more feature of the group’s practice shaped the anti-dialectical case in ways readers should notice. The September meetings rewarded a particular argumentative genre: the reconstruction. A member would take a canonical Marxist claim, state it as a set of explicit propositions, supply the strongest available defence, and then test the result against counterexamples, rival formalisations or recalcitrant evidence. Cohen’s treatment of the theory of history, Roemer’s treatment of exploitation and Elster’s treatment of Marx’s method all follow this pattern. Reconstruction is a respectful genre, far removed from dismissal, and it explains why the school’s polemics, even at their sharpest, are filled with careful exposition of the positions they reject. It also explains the school’s characteristic blind spot, which critics were quick to identify. A reconstruction can only test what can be stated as propositions with identifiable truth conditions. Whatever in a tradition resists that form, whether because it is nonsense or because it is a different kind of intellectual good, such as an orientation, a sensibility or a way of framing questions, will register in the reconstruction as nothing at all. The analysts concluded that dialectics was nothing at all. The defenders replied that the instrument, not the object, was at fault. Both claims deserve their full hearing.

Cohen and the defence of history without dialectics

Gerald Allan Cohen occupies the first place in any account of the school, not because he founded an organisation, for nobody did, but because his 1978 book demonstrated the possibility on which everything else depended. Karl Marx’s Theory of History: A Defence arrived with a quiet title and an immodest project: to extract from Marx’s scattered statements, principally the famous Preface of 1859, a coherent theory of historical development, and to defend that theory as a serious candidate for truth. Cohen proceeded as an analytical philosopher proceeds. He defined his terms, distinguishing the material from the social, the forces from the relations of production, with a patience that readers formed on dialectical commentary found either exhilarating or pedantic. He then stated the theory’s two central theses. The development thesis held that productive forces tend to develop through history, driven by human rationality confronting scarcity. The primacy thesis held that the character of a society’s production relations is explained by the level of development its productive forces have reached. Around those theses he built a defence that conceded difficulties, answered objections and distinguished the theory from cruder versions of economic determinism with which it was habitually confused.

The significance for the dialectics question lies in what the defence did not need. Marx’s Preface is a compact and metaphorical text, speaking of correspondence and conflict between forces and relations, of epochs of revolution, of forms of social consciousness. Generations of commentators had processed those metaphors through dialectical machinery, treating the conflict of forces and relations as a contradiction in the technical sense and the transition between modes of production as an instance of quantity passing into quality. Cohen simply declined the machinery. Conflict, in his reconstruction, meant incompatibility with consequences: relations that once fostered the growth of productive power come to fetter it, and the resulting dysfunction explains, functionally, why they are replaced. Correspondence meant suitability, specifiable in advance of any particular case. The famous architectural metaphor of base and superstructure was unpacked into claims about what explains what, and the unpacking exposed exactly which claims were strong, which were weak, and which had never been asserted by Marx at all. Whatever a reader’s final judgement of the theory, the experiment had been performed. A central Marxist doctrine had been stated, in full, in plain argumentative prose, and dialectical vocabulary had contributed nothing to the statement. The translation test had its first and most consequential application before it had a name.

Cohen’s own attitude to dialectics was cooler than Elster’s and is worth distinguishing. He rarely attacked dialectical writing head on, and his polemical energies were spent elsewhere: on the defence of functional explanation, on Nozick and the libertarian theory of rights, on the egalitarian critique of the difference principle. His anti-dialectics was practised rather than preached, a matter of writing books in which the thing simply never appeared, while its absence went unremarked by admiring readers who might have balked at an explicit repudiation. There is a lesson in that difference of temperament. The school contained both a demonstrative wing, which showed by example that Marxist theory could proceed without dialectics, and a prosecutorial wing, which argued that dialectical claims were actively confused. Cohen led the first; Elster led the second; Roemer, in his own domain, made the first wing’s case with mathematics. Later critics sometimes wrote as though the school spoke with Elster’s voice alone, and thereby missed how much of its persuasiveness came from Cohen’s quieter procedure. A doctrine is in deeper trouble when its opponents can ignore it successfully than when they denounce it, because denunciation at least confirms that something is there to be denounced.

The theory of history Cohen defended was, in his own later judgement, partly mistaken, and his willingness to say so belongs in this account. In essays written decades after the book, he reconsidered the confidence of the development thesis, questioned whether productive forces tend to develop across all recorded history, and moved his own centre of gravity from historical theory to normative political philosophy, to questions about justice, equality and the moral foundations of socialism. The trajectory matters for two reasons. It shows that the analytical standards were applied inward as well as outward: the school’s most distinguished figure subjected his own founding text to the same unsparing treatment he had applied to the inheritance, and revised it. And it complicates any simple story in which analytical Marxism figures as a machine for demolishing Marxism from within. Cohen’s destination was not liberalism but a more demanding socialism, argued with a moral seriousness that his critics in the dialectical tradition rarely matched. The rejection of dialectics, in his case, accompanied a deepening rather than an abandonment of commitment, a fact the school’s detractors have never found easy to absorb. The detailed anatomy of his most contested device, the functional explanation of production relations, belongs to the separate treatment of Cohen’s functional explanation and enters the next section only because Elster made it a principal exhibit in the case about method.

The quarrel over functional explanation

No internal dispute did more to clarify the school’s standards, or to expose their costs, than the quarrel over functional explanation. Cohen’s primacy thesis claimed that production relations are as they are because of their effects: a set of relations persists, roughly, because it suits the development of the productive forces at a given level. Explanations of that shape, which account for a phenomenon by its beneficial consequences for something else, are common in biology, where the heart’s structure is explained by its function of pumping blood, and Cohen argued that they could be legitimate in social theory too, provided the pattern of consequence was real. Relations rise when they would promote development and fall when they would obstruct it; their doing so is not an accident but the explanandum’s very form. He was aware that such explanations carry a danger. In biology, natural selection supplies the mechanism linking consequence to existence. In history, Cohen sometimes wrote as though the link could be left as an exercise, gestured at through revolutions, class struggles and the conscious designs of rulers, without being specified case by case. The gesture was precisely the sort of move the September Group existed to catch, and Elster caught it.

What is a functional explanation, and why did it become the school’s internal test case?

A functional explanation accounts for a social pattern by the effects that pattern produces, as when production relations are said to persist because they develop the productive forces. Elster objected that such explanations borrow the authority of biology without supplying a mechanism, and demanded the feedback loop connecting effect back to cause.

Elster’s objection was methodological and, within the group’s shared premises, devastating in its simplicity. A consequence explanation earns its keep only when a mechanism exists through which the consequence maintains its cause: a feedback loop, in the biologist’s case supplied by differential reproduction. Where no loop is specified, the explanation floats. It invites the theorist to infer, from the fact that an institution has convenient effects, that the effects explain the institution, and that inference, repeated across a theory, manufactures a picture of history as a self adjusting order in which whatever survives was somehow required. Elster saw in this pattern a survival of exactly what he disliked in dialectical and Hegelian habits of thought: a confidence that history possesses an inner logic that guarantees outcomes, dressed now in the respectable vocabulary of functions rather than in the older vocabulary of negation and sublation. His demand followed directly. For every functional claim, produce the microfoundations: the actors, their beliefs and desires, the constraints they face, and the process by which their interactions generate and sustain the pattern. Where the demand was met, the explanation stood, usually transformed in the process. Where it could not be met, the claim had to be withdrawn or demoted to a suggestive redescription.

Cohen’s reply, developed across several essays, conceded the desirability of mechanisms while denying that their absence voids an explanation. Science, he argued, often knows that a pattern of consequence explanation holds before it knows how: Darwin himself established that traits are explained by their contribution to fitness in advance of genetics, and the explanation was not empty in the interim. A consequence law, stating that relations of a given type tend to arise when they would be functional for the forces, could be confirmed by the pattern of cases even while its elaborations remained outstanding. The exchange never reached a verdict both parties accepted, and it is not necessary to award one here. Its importance for the dialectics question lies in what it revealed about the school’s centre of gravity. The deepest commitment of analytical Marxism was not to any particular Marxist thesis, since its own most cherished thesis could be put on trial, but to a standard of explanatory adequacy imported from the philosophy of science and the methodology of social research. Dialectics failed that standard in the eyes of the school not because it was Marxist but because it could not, or would not, state its claims in a form the standard could assess. Functional explanation survived, chastened and hedged, because it could at least be stated clearly enough to be argued about. The contrast between the fates of the two devices is the cleanest illustration available of what the group meant by rigour.

Historians of the school have often noted the irony that this dispute, conducted in the language of Anglo Saxon philosophy of science, reproduced within analytical Marxism the very tension its founders had hoped to escape. Cohen’s willingness to let a well confirmed pattern stand in advance of its mechanism has an affinity, which he would have resisted, with the dialectical confidence that the whole can be grasped before the parts are traced. Elster’s insistence that nothing counts until the individual level story is told has an affinity, which he embraced, with a methodological individualism that many Marxists regarded as the dissolution of social theory altogether. The school did not resolve the tension; it institutionalised it as an ongoing argument, and its best work emerged from the friction. Readers meeting the episode for the first time sometimes ask which side the school took. The honest answer is that the school was the argument, and that the argument’s terms, mechanism against pattern, micro against macro, set the agenda for the anti-dialectical case in its mature form.

Elster’s prosecution of dialectical thinking

If Cohen demonstrated, Elster prosecuted. Jon Elster’s engagement with Marx began earlier than the September Group and ran deeper than a single book: a study of Leibniz and Marx, a book on the logic of dialectics and the social sciences, and then, in 1985, Making Sense of Marx, the volume that fixed the school’s public image. The book’s method was exhaustiveness. Chapter by chapter it passed the Marxian corpus through a series of sieves, retaining what could be restated as defensible social science and discarding the rest with reasons given. Its chapter on dialectics performed the same operation on the tradition’s sacred vocabulary, and it remains the text any defender of dialectics must answer, because it refuses the usual polemical shortcuts. Elster did not claim that dialectical writers were stupid, nor that their concerns were unreal. He claimed that their central theses, examined one by one, decomposed into a small set of recurring types: claims that were true but trivial, claims that were interesting but false, and claims whose apparent content evaporated when an attempt was made to state what observation would count against them.

The law of the transformation of quantity into quality received the treatment in miniature. As a metaphysical law, Elster argued, it is either empty or false: no threshold is specified, no mechanism named, and the examples standardly offered, water boiling, a heap losing grains, a crowd becoming a mob, describe ordinary discontinuities that physics, chemistry and social psychology handle in their own vocabularies without residue. As a reminder that gradual change can have abrupt effects, it is true and worth knowing, but it is then a piece of methodological advice available to any tradition, not a discovery about the structure of reality. The unity and struggle of opposites fared similarly. Taken literally, it conflated several distinct relations: logical contradiction between propositions, causal opposition between forces, conflict between groups with opposed interests, and the functional interdependence of complementary roles. Each of those relations is real and important. The dialectical formula’s achievement was to run them together under one heading, thereby borrowing the urgency of conflict for interdependence and the certainty of logic for causation. The negation of the negation, the third of the canonical laws inherited from Engels, Elster treated as the weakest of all: a pattern discernible in any three stage sequence whatever, imposed retrospectively by the choice of where to start and stop counting, and incapable of predicting or explaining any particular transition.

Behind the itemised critique stood a diagnosis of why dialectical writing persisted despite its defects, and the diagnosis is the part of Elster’s case that defenders have found hardest to forgive. Dialectical formulas, he suggested, serve psychological and institutional functions independent of their truth. They confer an impression of depth on their user and of initiation on their audience. They license the theorist to have things both ways, since a position and its opposite can each be affirmed at different moments under the protection of the doctrine that opposites interpenetrate. They protect cherished conclusions from falsification, because any recalcitrant fact can be absorbed as a further turn of the spiral. And they flatter a movement’s intellectuals with the possession of a logic unavailable to outsiders, a private instrument that guarantees in advance that the tradition’s conclusions cannot be reached, or criticised, by ordinary means. A mode of thought with those functions would survive its own refutation indefinitely, which, Elster concluded, is exactly what had happened. The proper response was not a better dialectics but an exit: social theory conducted with the concepts of intention, belief, opportunity and interaction, whose credentials do not depend on initiation.

Fairness requires registering the limits of the prosecution as clearly as its force. Elster aimed principally at the dialectics of the textbooks: the codified dialectical materialism of party manuals and the Hegelian Marxism that traded in the canonical laws. More sophisticated dialectical thinkers, from Lukács at his most careful to the critical realists discussed below, disowned several of the theses Elster dismantled, and could fairly complain that the indictment convicted a vulgarised version of their position. Elster also wrote as a convert against his own past enthusiasms, and the convert’s zeal occasionally outran the evidence, as when suggestive remarks in Marx’s notebooks were treated as doctrines or when the whole Hegelian inheritance was priced at the value of its worst expositors. His subsequent career adds a further complication. He moved steadily away from Marxism itself, towards a general theory of rationality, emotion and social mechanisms, and in retrospect described his Marxist period with a detachment that embarrassed former allies. The school’s most aggressive anti-dialectician thus became, in time, an ambiguous witness: his arguments stand or fall independently of his allegiance, but the trajectory feeds the suspicion, examined later in this article, that the analytical method carried a solvent within it that would eventually dissolve the Marxism it was meant to purify. His wider position, and the question of what remained of it, is treated in the companion portrait of Elster’s Marxism, which this article deliberately does not duplicate.

The contemporary reception of Making Sense of Marx divided along lines that previewed the whole subsequent debate. Admirers praised its lucidity and its courage, and treated the dialectics chapter as a liberation. Critics, including reviewers sympathetic to analytical standards, objected that the book’s sieve had been calibrated to pass only what resembled late twentieth century social science, so that the conclusion, that Marx is defensible exactly where he anticipates that social science, was built into the instrument. The objection has force against Elster’s valuations of Marx. It has less force against the dialectics chapter taken narrowly, because there the procedure was not sifting by resemblance but the simpler demand that claims state their content. A defender who believes the sieve misclassified dialectical claims has a clear remedy available: exhibit one, state it fully, and show what the plain restatement misses. That remedy, endlessly available, structures the rest of this article. The doctrine whose claims are to be exhibited is set out in the series treatment of dialectical materialism as a complete guide, and the category at the centre of the dispute, contradiction, has its own examination in the treatment of contradiction in Marxist method; what follows concentrates on the analytical challenge and the strongest replies to it.

Microfoundations and methodological individualism

Beneath the quarrel about dialectics lay a commitment that did more than any slogan to define the school: methodological individualism, together with its practical corollary, the demand for microfoundations. The doctrine is chronically misunderstood, and the school’s critics and defenders alike have profited from the misunderstanding, so it pays to state it with care. Methodological individualism is a claim about explanation, not about politics, psychology or the nature of human beings. It holds that explanations of large scale social phenomena are complete only when they can be grounded, at least in principle, in the actions, beliefs, desires and opportunities of individual persons. Institutions, classes and structures are real in the sense that they constrain and enable what people do; the doctrine denies only that they act, intend or explain except through the people who compose and confront them. A statement such as the bourgeoisie imposed a factory code is, on this view, a promissory note. It must be redeemable, when challenged, into an account of legislators, inspectors, magistrates, owners and workers whose situated actions produced the code and secured compliance with it. Notes that cannot be redeemed may still be useful shorthand. They are not explanations.

What does the demand for microfoundations actually require of a Marxist claim?

The demand requires that any claim about classes, structures or historical tendencies be redeemable, at least in outline, into an account of situated individuals: their beliefs, desires, opportunities and interactions, and the unintended patterns those interactions generate. Shorthand above that level is permitted; explanation may not stop there.

The demand reshaped Marxist explanation at several points. Consider the concept of class interest, a staple of the tradition and a frequent site of dialectical vagueness, since interests were often attributed to classes as though classes were persons with a unified point of view. Under the demand for microfoundations, an interest claim becomes a conjunction of testable sub claims: that members of a class occupy positions with characteristic opportunities; that certain outcomes would benefit occupants of those positions; that beliefs about those benefits are formed under identifiable informational conditions; and that organisation, where it exists, aggregates individual motivations through mechanisms, selective incentives, solidarity norms, coercion, that themselves require specification. Nothing in the reformulation prevents a theorist from speaking of class interests. It prevents the theorist from using the phrase as a self certifying explanation of behaviour, which is how the phrase had too often functioned. Elster pressed the point into a general scepticism about collective actors, arguing that much Marxist prose derived its apparent explanatory power from grammatical subjects, the proletariat, capital, history, that named no agent capable of bearing the predicates assigned to them. Wright, from within the school, resisted the most austere versions of the doctrine and defended a place for structural concepts with causal powers of their own, and the resulting internal argument about how much individualism the method required was never fully settled. The unsettled state of the argument is itself informative: the school shared a demand for mechanisms while disagreeing about the ontology that mechanisms presuppose.

Two misconceptions should be cleared away, because they recur in almost every popular account. Methodological individualism does not entail political individualism. A theorist can hold that explanations must bottom out in individuals while advocating thoroughgoing collective ownership, as several members of the school did, because the first claim concerns how social processes work and the second concerns how they ought to be arranged. Nor does the doctrine claim that individuals are self sufficient atoms whose properties are fixed independently of their relations. Elster and Roemer both allowed that beliefs, preferences and capacities are socially formed; the demand is that their formation, too, be explained through processes involving people in situations, rather than attributed to the emanations of a structure. Whether that allowance is enough to capture what the dialectical tradition means by the social constitution of the individual is precisely the point at issue in the defence from internal relations, and readers should hold the question open until that defence has been heard. What can be said at this stage is that the analytical position was a considered one, not a failure to notice that people are social. Its proponents noticed, and proposed to explain the sociality rather than to presuppose it under a honorific name.

The practical yield of the doctrine appeared in the school’s characteristic explanatory genre, the mechanism. Where dialectical writing tended to move from a totality to its moments, analytical writing moved from a situation to the processes it set running: how a change in asset distribution alters bargaining power, how bargaining outcomes aggregate into investment patterns, how those patterns feed back into the distribution of assets. Mechanisms are portable in a way totalities are not. A mechanism identified in one setting, say the process by which employers use unemployment to discipline wage claims, can be looked for in others, refined, and combined with neighbouring mechanisms into larger accounts without any claim to have grasped the whole. The genre has obvious affinities with the mature social sciences and equally obvious limits, which the defenders exploited: a chain of mechanisms explains how a system reproduces or changes, given its terms, and has less to say about how the terms themselves, the very categories of the analysis, came to be and might cease to be. That limitation, the problem of historical form, returns in the ledger below. First, however, the school’s most technically ambitious achievement needs its hearing, because Roemer’s models are the strongest evidence that something of substance, and not merely a standard of complaint, emerged from the rejection of dialectics.

Roemer’s models: exploitation rebuilt on new foundations

John Roemer’s contribution gave the school its proof of concept in economics, and no part of the analytical achievement better illustrates what the rejection of dialectics made possible. Marx’s theory of exploitation, as inherited, rested on the labour theory of value: the doctrine that commodities exchange in proportion to the labour embodied in them, from which it followed that profit arises because workers produce more value than their labour power costs. By the nineteen seventies that foundation was in serious technical trouble. The transformation problem, the difficulty of deriving prices from values consistently, had consumed a century of scholarship, and the most rigorous Marxian economists, following the Sraffian critique, had shown that prices and profits could be computed directly from physical production conditions and the wage, without values entering at any stage. A tradition committed to dialectical reassurance could respond that the value theory operated at a deeper level of essence beneath the appearances of price. Roemer treated that response as the paradigm of what had to go. In A General Theory of Exploitation and Class, published in 1982, he asked what exploitation could mean if the labour theory of value were simply set aside, and answered with models.

The models deserve description rather than fake mathematics, and their logic can be followed by any reader willing to track a thought experiment. Imagine an economy of agents who differ in one respect: the productive assets, land, tools, capital, they start with. Agents can work for themselves, hire others, or be hired, and they choose the arrangement that leaves them best off. Roemer defined exploitation through a counterfactual withdrawal test modelled on cooperative game theory. A group counts as exploited if it would do better by withdrawing from the economy with its per capita share of society’s productive assets, while the complementary group would do worse. The definition sounds technical, and it is, but its moral and analytical content is transparent. Exploitation, on this account, is not fundamentally about labour being extracted in production at all. It is about the distribution of property. People are exploited because they lack assets and must therefore work on terms set by those who have them, and the transfers observed inside production are the symptom of that prior inequality rather than its essence. Roemer then proved a result that scandalised traditionalists and delighted analysts: under specified conditions, exploitation in his sense emerges in equilibrium and corresponds to class position, with asset rich agents hiring labour and asset poor agents hiring themselves out, so that the class structure Marx described falls out of the model as a consequence of the asset distribution rather than being assumed at the start.

How can a model define exploitation without the labour theory of value?

Roemer’s withdrawal test defines a group as exploited when it would fare better by leaving the economy with an equal per capita share of productive assets. Exploitation is thereby grounded in the distribution of property rather than in embodied labour, and class positions emerge from that distribution inside the model.

The results that followed cut in several directions at once, which is why the book is taken seriously even by many of its critics. Roemer showed that exploitation could persist in an economy with a credit market but no labour market, where the asset poor borrow capital and surrender interest instead of wages, suggesting that the labour market, the institution at the centre of Marx’s account, was one vehicle of exploitation among others rather than its unique seat. He showed conversely that labour transfers could occur without exploitation in his sense where asset distributions were equal, detaching the moral charge of exploitation from the mere fact that some people’s work benefits others. And he constructed cases in which a person could be an exploiter by the labour transfer criterion while being poor, or exploited while comparatively rich, anomalies that the inherited theory had no resources to discuss and that any adequate successor theory had to face. Each result was obtained by the method the school preached: assumptions printed on the page, definitions fixed in advance, conclusions following by argument that a hostile reader could check line by line. Nobody needed to invoke the cunning of reason, the negation of the negation, or the inner logic of capital to follow the reasoning, and nobody could evade its conclusions by redescribing them at a higher level of abstraction. Whatever the models’ ultimate standing, they demonstrated that the central moral intuition of Marxian economics could be given a precise, vulnerable, improvable form. That demonstration is the school’s strongest exhibit in the case that the translation programme produced gains and not only losses.

Precision also requires stating what the models assumed away, since the defenders of the older tradition located their reply exactly there. Roemer’s agents arrive with assets and preferences already formed, confront prices already given, and choose within rules whose own origins lie outside the model. History enters only as the initial distribution, a snapshot whose provenance the theory brackets. Change, where it occurs, is movement to a new equilibrium rather than the transformation of the framework itself, and the categories, owner, worker, asset, market, are treated as stable across the analysis. A dialectical reader will say that the model thereby assumes precisely what a theory of capitalism should explain: how those categories were historically constituted, how the agents’ preferences are shaped by their positions, and how the system generates the crises through which its own forms are periodically remade. The analytical reply is that no single model must do everything, that bracketing is the condition of rigour, and that historical constitution can be studied in further models and narratives once the static relations are clear. Both statements are true, and the disagreement is about the division of intellectual labour: whether bracketed questions stay on the agenda once the bracketing technique becomes a habit and a career. The school’s later history, in which formal modelling flourished while historical work thinned, gave that question an empirical edge its founders had not anticipated.

Roemer’s own intellectual itinerary illustrates the school’s trajectory in one career. The economist who rebuilt exploitation theory in the nineteen eighties went on to write on equality of opportunity, on the political economy of socialism after the Soviet collapse, and on intergenerational justice, moving steadily from the reconstruction of Marx towards free standing contributions to economics and political philosophy that a reader could value without any Marxist commitment. He remained, by his own account, on the egalitarian left, and the models retained their place in the Marxian literature. Yet the pattern repeated across the group: the tools, once sharpened, cut loose from the tradition that had provoked their sharpening, and the school’s members increasingly addressed audiences for whom Marx was an occasion rather than an allegiance. Whether that outcome represents the method’s triumph or its self dissolution is the question reserved for the dispersal section below. What belongs here is the narrower point. The Roemer episode shows the anti-dialectical programme at its most constructive, producing theorems where the older idiom produced commentary, and it therefore sets the standard any defence of dialectics must meet. The defence must show not that models are limited, which everyone concedes, but that some specific content of the Marxist tradition is available in dialectical form and unavailable in any form the analytical toolkit recognises.

Game theory as the replacement machinery

Alongside the models stood a humbler instrument that the school used so constantly that it became a signature: game theory, the formal study of strategic interaction among agents who know that their outcomes depend on one another’s choices. Its role in the anti-dialectical programme deserves separate treatment, because game theory supplied what dialectics had often promised and rarely delivered, a precise language for conflict, interdependence and unintended outcomes. The tradition had always insisted that social life is constituted by opposed interests whose interplay generates results nobody intends: competition forcing capitalists to accumulate against their individual inclinations, workers’ combinations arising from and reacting back upon employers’ strategies, states balancing between classes. Dialectical prose evoked those structures with words like contradiction and mediation. Game theory specified them. A situation is rendered as players, strategies, information and payoffs; an equilibrium is identified; comparative questions, what happens when the payoffs shift, when information improves, when a coalition forms, can then be asked and answered rather than gestured at.

The famous dilemmas did genuine work in Marxist argument once they were imported. The tendency of competition to compel accumulation, which Marx described in the language of coercive laws imposing themselves on individual capitals, becomes in game theoretic dress a structure in which each firm’s best response to rivals’ investment is investment of its own, so that the system level pattern, accumulation as a condition of survival, is explained by individual level incentives rather than attributed to a systemic will. The fragility of workers’ solidarity, which the tradition explained by false consciousness or betrayal, becomes partly a problem of collective action with known structure: benefits that accrue to all whether or not each contributes, temptations to free ride, and solutions, unions, norms, sanctions, whose design can be studied rather than lamented. Class compromise, the postwar settlements between labour and capital, becomes modelable as an equilibrium sustained under conditions, full employment, growth, credible commitments, whose erosion then explains the settlement’s decay without recourse to a rising tide of class consciousness or its mysterious ebb. In each case the translation into strategic form did not merely restate the older claim. It refined it, exposed the assumptions on which it depended, and generated new questions, about information, enforcement and time horizons, that the older vocabulary had not contained. Translation, practised at this level, was a creative act.

Critics replied, reasonably, that the instrument carries its own metaphysics in disguise. Game theoretic models typically take preferences as given and stable, treat players as antecedently individuated, and represent their relations as external: each agent confronts others as facts of the environment, however strategically entangled their choices become. A tradition formed by Hegel and Marx suspects exactly those starting points. It holds that the agents of class society are not merely entangled but internally related, constituted as the agents they are by the positions they occupy, so that a capitalist’s preferences, horizons and identity are products of the same structure that the model treats as the game’s rules. The suspicion cannot be answered by technique, and the school’s honest members did not pretend otherwise. Elster treated preference formation as a further topic for mechanism based study, pointing to adaptive preferences and sour grapes reasoning he had analysed elsewhere. Roemer’s framework could accommodate endogenous assets more readily than endogenous selves. Wright worried the problem from the sociological side. What none could claim is that game theory, as standardly practised, represents constitutive relations with the same naturalness as it represents strategic ones. The instrument excels at interaction among the already formed. Formation itself, of agents, categories and rules, is where its grip loosens, and that looseness marks the exact spot at which the dialectical defence plants its flag.

A second caution, entered by practitioners rather than opponents, concerns the seduction of solvability. A model that reaches equilibrium is intellectually satisfying in a way that history rarely is, and the satisfaction can mislead. Analytical Marxists were aware that their reconstructions selected, from Marx’s sprawling corpus, the claims most amenable to formal treatment, and that a reader who met Marxism only through the reconstructions would receive a distorted impression of both the man and the tradition: all statics and incentives, none of the feverish historical narrative, the literary violence, the sense of a world being made and unmade that gives the original texts their hold on readers. Cohen, whose prose retained a literary quality his discipline rarely rewards, occasionally acknowledged the loss in asides. The acknowledgement matters for the fairness of the ledger. The school’s internal standard could register a translation as successful whenever the propositional content survived, but propositional content is not the whole of what a text does, and a method that cannot register rhetorical, orienting and imaginative functions will systematically undervalue the traditions it examines. That concession, small in the school’s own accounting, grows larger when the defence has been heard in full.

What dialectics claimed to be

A fair hearing for the defence requires the defended doctrine to be stated in its strongest form rather than in the textbook form Elster prosecuted. Dialectics descends to Marxism through Hegel, for whom it named the movement of thought and reality through determinations that undermine themselves: a category, pressed to its limit, reveals its dependence on what it excluded, and the resulting richer unity preserves the opposed terms in transformed guise. Marx’s relation to that inheritance was never the simple appropriation of party legend. In the afterword to the second German edition of Capital he famously distinguished his dialectic from Hegel’s, protesting that in Hegel it stood on its head and required inversion to disclose the rational kernel within the mystical shell. Whatever the metaphor means precisely, and scholars have contested it for a century, it announces a debt and a difference together: a method retained, a metaphysics rejected, a debt to the very philosopher whose system the mature Marx was supposedly leaving behind. The analytical Marxists read the evidence biographically and textually and concluded that the debt was smaller, and the mystical shell thicker, than the tradition admitted. Cohen found virtually no dialectical reasoning in the texts of Marx he valued. Elster found residues he classified as youthful Hegelianism imperfectly outgrown. Neither denied that Marx used the vocabulary on occasion. They denied that the vocabulary did work the plainer parts of his argument could not do without it.

Engels complicates the story, because the dialectics the twentieth century inherited came mediated through his popularisations. In Anti-Dühring and in the manuscripts published posthumously as Dialectics of Nature, Engels extracted from Hegel three laws, the interpenetration of opposites, the transformation of quantity into quality, and the negation of the negation, and presented them as laws of nature, history and thought alike, illustrated from physics, biology and mathematics. The codification proved fateful. It converted a difficult and contested method into a portable catechism, gave party schools something teachable, and supplied the later Soviet diamat with its skeleton. When Elster attacked the three laws, he was attacking Engels’s extraction and its textbook descendants at least as much as anything in Marx, and defenders of dialectics have often replied by disowning the codification and retreating to stronger ground: to Marx’s practice in Capital, where categories unfold through their internal tensions; to Lukács, for whom dialectics named the standpoint of totality and the identity of subject and object in the proletariat; or to the Frankfurt School’s negative dialectics, which refused every positive system. The retreat is legitimate, but it has a cost the analysts were quick to note. A doctrine whose defenders relocate it whenever it is criticised, from laws to method, from method to sensibility, from sensibility to the ineffable practice of a single book, becomes difficult to distinguish from the evasive phenomenon the Non-Bullshit nickname targeted. The defence must therefore be pinned to specific claims. Two such claims, the strongest available, occupy the next sections: the philosophy of internal relations, and the critical realist doctrine of emergence.

Before those defences are heard, the analytical side is entitled to one summary exhibit, the sheer range of meanings the word dialectics has carried within the tradition itself. It has named a logic of contradiction and a logic of change; a theory of nature and a refusal to extend beyond society; the method of Capital and the catechism of its epigones; a standpoint epistemology in which the oppressed see truly, and a research heuristic in which opposites are sought as a matter of routine. Sophisticated practitioners distinguish these senses carefully. The tradition’s ordinary discourse does not, and the slippage between senses performs argumentative work: criticisms aimed at the catechism are parried by appeal to the method, claims made for the method are banked as vindications of the logic, and the whole acquires a resilience that no single one of its senses could earn. The analysts’ demand for disambiguation was thus not pedantry. It was the precondition of the debate having an outcome at all, and even readers who finally side with the defence generally concede that the debate, after the September Group, could no longer be conducted in the old fused vocabulary. That concession is a permanent gain of the episode, whatever else is entered on the ledger’s other side.

The defence from internal relations

The philosophy of internal relations, given its fullest Marxist statement by Bertell Ollman, holds that the basic error of analytical thought is its treatment of relations as external connections between independently constituted things. On the internal relations view, some relations enter into what their terms are. Capital and wage labour supply the canonical case. A capitalist is not first a person with certain properties who subsequently enters a relation with workers; the position is unintelligible except as one pole of a relation whose other pole is wage labour, and the same holds in reverse. Ollman argued that Marx’s method in Capital presupposes this ontology throughout: abstractions are drawn at shifting levels of generality, the same phenomenon is viewed from the standpoints of production, circulation and the whole, and the categories are expected to change their meaning as the inquiry widens, because each category’s content includes its place in the developing totality. Dialectics, on this account, is not a set of laws pasted onto the material. It is the discipline of thinking internal relations without prematurely freezing them into external ones, and its apparent obscurities are the scars left by an ontology that ordinary language, built for billiard balls and substances, handles badly.

Can a relation constitute the things it relates, rather than merely connect them?

Defenders of internal relations argue that capital and wage labour are constituted by their relation: neither position exists, or can be defined, without the other. Analytical critics reply that constitutive dependence can be stated as a claim about definitions and constraints, without a special logic, and the dispute turns on that reply.

The analytical response came in two waves, one dismissive and one serious, and only the second deserves space. The dismissive response observed that mutual definitional dependence is commonplace and harmless: buyer and seller, landlord and tenant, husband and wife are internally related in exactly the sense that neither term is definable without the other, yet no special logic has been found necessary for the sociology of markets, property or marriage. The serious response, developed by critics such as Andrew Levine and Elliott Sober in their methodological writings and echoed in Elster’s later work, granted the ontology its strongest case and asked what followed for explanation. Suppose positions are internally related. It remains true, the reply runs, that any particular process occurring within those positions, a wage bargain, an investment strike, a technological displacement, unfolds through the actions of situated people, and can be modelled as such once the positions and their constitutive rules are specified as constraints. Internal relations then function as a claim about the deep grammar of the social world: true or false as metaphysics, but without consequences that compete with mechanism based explanation, because the two operate at different levels. The defender’s counter reply is that the levels cannot be kept apart so cleanly. If the terms are constituted by the relation, then the individuals who populate a model, with their preferences and self understandings, are themselves products of the process under study, and the analyst’s starting point, agents with given properties choosing under constraints, is not a neutral foundation but a snapshot that mistakes a moment of a process for its ground. The argument at this depth becomes a genuine philosophical standoff, about whether explanation must always begin from the individuated or may legitimately begin from the whole in process. Readers should notice that the standoff is not between sense and nonsense. It is between two research programmes with different native strengths, and the honest conclusion of this section is that the analysts won the argument about logic while the internal relations defence survived, in reduced and clarified form, as an argument about ontology and the choice of explanatory starting points.

Ollman’s defence also carried an implication for how Marx should be read that outlasted the immediate quarrel. If categories shift with levels of abstraction, then the hunt for Marx’s final definitions, the procedure Cohen’s reconstructions exemplified, misconceives its object: a definition true at one level of generality is a partial view whose correction lies not in more precision at the same level but in the movement to a wider one. Analytical readers found this doctrine exasperating, since it appeared to license the author to escape any refutation by ascending a level. Yet the doctrine had a defensible core that even unsympathetic scholars have acknowledged in their practice. Social categories are historical: what counts as money, as a firm, as unemployment, changes with the institutional whole, and a theory that fixes its terms in advance risks studying a world that has moved on. The mature position, toward which both camps edged without ever signing a treaty, treats definitional stability as a local and revisable achievement, indispensable for particular explanations and dangerous when mistaken for the structure of reality. That position owes something to both sides, and it is recorded in the ledger below as one of the items on which the loss has been partly made good, from an unexpected direction.

Critical realism and the argument from emergence

The second strong defence arrived from philosophy of science rather than from within the older Marxist schools. Critical realism, associated principally with Roy Bhaskar, whose A Realist Theory of Science appeared in 1975 and whose Dialectic: The Pulse of Freedom attempted a full dialectical system in 1993, offered a stratified ontology in which the world consists of structures endowed with causal powers, operating in open systems where outcomes are jointly determined and rarely exhibit the constant conjunctions of the laboratory. On this picture, social structures are concept dependent and activity dependent yet real: they pre exist each generation of agents, constrain and enable what agents do, and are reproduced or transformed through that activity, in the account Bhaskar called the transformational model of social activity. Andrew Collier and other critical realists drew the moral for the Marxism debates. A social science modelled on that ontology could honour what dialectics had been groping towards, interconnection, internal relations, emergence, process, while stating its claims in the disciplined vocabulary of contemporary philosophy, and could therefore meet the analysts halfway: no laws of negation, no logic of contradiction, but a principled refusal of methodological individualism, grounded in the claim that structures possess emergent powers irreducible to the properties of their current occupants.

Emergence is the load bearing notion, and it deserves a precise statement because both camps used it loosely when it suited them. A property is emergent when a structured whole possesses it in virtue of the organisation of its parts, and the whole’s behaviour cannot be deduced, even in principle by a finite reasoner, from the properties the parts would have in isolation. Liquidity is the textbook physical example; a market’s liquidity serves better here. No individual trader is liquid; the market is, when its organisation allows assets to change hands without moving prices, and that systemic property constrains every participant while belonging to none. The critical realist argues that class structures exhibit emergence in this sense, and that an explanatory practice which insists on redeeming every structural claim into individual level mechanisms will systematically lose the systemic level at which the most important determinations operate. The analytical reply, by now familiar, distinguishes the ontological claim from the explanatory demand. Emergent properties may be real; explanations invoking them still earn their keep only when the processes sustaining them can be indicated, and in the market example they can: order flow, market makers, rules of the exchange. Where a structural claim resists every attempt at process indication, the suspicion returns that the claim is performing rhetorical rather than explanatory work. The critical realist rejoinder, that open systems make complete mechanism specification an unreasonable demand and that the analysts confuse an ideal of explanation with its reality, has genuine force in the philosophy of science, where the same argument has been conducted about biology and neuroscience with no victory for reduction.

The historical significance of critical realism for this story is that it changed the defence’s terrain. Before Bhaskar, the defence of dialectics tended to rest on textual authority, on what Marx or Lukács had said, or on the political virtues of a holistic standpoint. After Bhaskar, a defender could argue from the structure of scientific explanation itself, with allies in mainstream philosophy of science who had no Marxist commitments to protect. The analysts noticed. Wright’s later work engaged critical realism seriously, and the younger generation of Marxist sociologists often combined mechanism based explanation with a realist ontology, treating the September Group’s methodological individualism as one tool among others rather than as a constitution. That synthesis, untidy and unratified, is probably the closest thing the debate produced to a settlement: dialectics as a logic was dead beyond resurrection, dialectics as a reminder that wholes organise their parts survived by changing its name, and the word itself retreated to the traditions that had never doubted it. Whether the settlement satisfies either original party is doubtful. Settlements in intellectual history rarely do. What can be said is that the terms on which Marxist method is discussed, in the journals and in the better textbooks, were permanently altered by the collision, and that readers who wish to see the defence practised rather than merely stated should consult the series treatment of dialectical method in research, where the claims of totality and internal relations are put to work on concrete material.

Contradiction: one word, several relations

The single most contested category in the dispute deserves its own dissection, because the pattern of the whole debate repeats inside it. In ordinary Marxist usage, contradiction names the relation between capital and labour, between forces and relations of production, between socialised production and private appropriation, and dozens of further pairs. The analytical critics asked a disarmingly simple question: which relation, in each case, does the word denote? Their catalogue of candidates remains the standard one. A contradiction might be a logical inconsistency between propositions, which cannot be jointly true and cannot obtain in the world at all. It might be a causal opposition, in which two processes push a variable in opposite directions, as thrift and consumption push demand. It might be a conflict of interest between agents, in which each party’s gain is the other’s loss. It might be a functional incompatibility, in which two requirements of a system’s reproduction cannot be simultaneously met for long. Or it might be a constitutive interdependence misdescribed in the vocabulary of conflict, as when the poles of an internal relation are said to contradict one another merely because they are distinct and mutually defining. Each candidate relation is real, important and open to study. The critics’ charge was that the dialectical term gains its aura by sliding among them: a passage begins with incompatibility in the functional sense, borrows the inexorability of logic in mid argument, and concludes with the drama of conflict, having nowhere established that its subject exhibits all three at once.

The strongest dialectical reply concedes the catalogue and disputes the bookkeeping. On the reply, the point of the term is precisely that in class society these relations arrive fused. The incompatibility between developing forces and confining relations is not merely like a conflict of interest; it is lived as one, because the relations are embodied in classes whose interests collide, and the collision is not merely like a logical tension; it issues, through crisis, in the destruction of one term of the pair. To insist on separating the senses, the defender argues, is to perform the characteristic analytical operation of taking apart what reality joins, and then to express surprise that the reassembled account lacks the original’s power to explain why change in such societies is convulsive rather than smooth. Elster’s counter was characteristic and, on its own ground, effective: fusion in reality, if real, can be exhibited case by case by showing the distinct relations and their coupling, and a writer who has done that work no longer needs the fused term, while a writer who has not done it is using the term to conceal the omission. The exchange reached the same resting point as the internal relations dispute. Nothing in the analytical critique prevented a Marxist from writing about fused, mutually reinforcing antagonisms. What it prevented was citing the word contradiction as though the citation were itself the analysis. Most working Marxist scholarship, in the decades since, has quietly complied with the prevention while retaining the word, which suggests that the critique succeeded as a discipline even where it failed as a prohibition.

One genuine difficulty for the analytical side should be recorded here, because the school’s own historians have acknowledged it. The demand that every fused usage be decomposed carries an implicit promise that decomposition loses nothing, and the promise is empirical rather than logical. Languages preserve fused terms where fusion tracks something: the history of science retains concepts that bundle structure and process, from gene to institution, and attempts to replace them with decomposed vocabularies have repeatedly collapsed under their own clumsiness. The analysts could reply that clumsiness is a small price for clarity, and Cohen, whose prose was anything but clumsy, showed that clear restatement need not read badly. Still, the episode suggests a more modest formulation of the translation test than its founders sometimes used. Translation is a challenge and an audit, an instrument for locating what a claim commits its holder to, and not a metaphysical guarantee that whatever resists translation is void. The mature use of the test, defended in the penultimate section of this article, keeps the audit and drops the guarantee.

What dropping dialectics gained

The gains side of the account can be stated without triumphalism, because most of them are visible in the subsequent literature of all camps. The first gain was disambiguation. After the September Group, a Marxist writer who used the central categories was expected, by referees, editors and informed readers, to be able to say what relation contradiction denoted, what an appeal to totality committed the analysis to, and what would count as evidence against a claim of mediation or negation. Writers who could say these things wrote better dialectics for the discipline; writers who could not lost the protective fog in which assertion had passed for argument. The second gain was a new body of positive theory. Roemer’s reconstruction of exploitation, Cohen’s defence and subsequent self revision of the theory of history, Wright’s empirically operationalised class analysis, and the school’s collective work on equality and justice constitute a shelf of Marxist scholarship that is read, assigned and argued with by scholars who owe the tradition nothing. None of it required dialectical machinery, and much of it could not have been produced inside the older idiom, because the idiom lacked the devices, equilibrium analysis, mechanism specification, explicit counterfactuals, on which the results depend. The third gain was conversational. Analytical Marxism reconnected Marxist theory to the mainstreams of philosophy, economics and sociology from which decades of dialectical self sufficiency had insulated it, and the reconnection ran both ways: mainstream debates about exploitation, distributive justice and the explanation of historical change acquired interlocutors who knew the Marxist corpus from inside.

A fourth gain is subtler and concerns the tradition’s relation to its own founder. Dialectical deference had functioned, in practice, as a barrier to criticising Marx: a defective argument could be protected by the reflection that its defect was an appearance dispelled at a higher level of the dialectic. The analytical procedure removed the protection. Marx’s claims were treated as claims, some vindicated, some refuted, some found unintelligible as stated and reconstructed in altered form, and the resulting portrait, of a powerful and fallible theorist rather than a prophet guarded by a special logic, is arguably more respectful than the deference it replaced, since it extends to Marx the treatment every other major thinker receives as a matter of course. Elster’s verdicts were often harsh, and Roemer’s reconstruction discarded the labour theory of value outright, yet both men engaged Marx’s arguments at a level of detail that devotional commentary rarely attempted. Readers who value the texts have reason to prefer critics who read them closely to defenders who insulate them from reading.

Set against these gains is a question of attribution that fairness requires. Some part of what the school achieved would have arrived anyway, carried by the general professionalisation of the social sciences and the decline of the party orthodoxies that had policed dialectical orthodoxy. The September Group accelerated and concentrated changes larger than itself, and its members, Cohen especially, were too honest to claim otherwise. The defensible claim is narrower and still substantial: that this identifiable group of scholars, through identifiable texts and an identifiable standard of argument, changed what counted as a respectable Marxist claim in the Anglophone academy within a single generation, and that the rejection of dialectics was the cutting edge of the change. No comparable group working within the dialectical idiom achieved a similar transformation in the same period, and the contrast is part of the evidence the ledger must weigh.

What dropping dialectics cost

The losses side of the account begins with the item the defenders named from the start: totality. Marxist analysis in its classical form aspired to grasp a social formation as a structured whole, in which the meaning and weight of any element depends on its place among the others, and in which explanation moves between part and whole in both directions. Analytical practice, proceeding by decomposition into mechanisms and models, has no native place for that aspiration. A mechanism explains a fragment well; the sum of fragments is a heap, not a totality, and the school’s works characteristically decline to say how their fragments compose a society. Something real is surrendered here, and it is not merely aesthetic. Phenomena that exist only at the level of the whole, the general rate of profit as a discipline on every firm, the world market as a constraint on every national policy, the ideological atmosphere that renders some questions unaskable in a period, appear in analytical treatments as background conditions rather than as objects of theory. The tradition that thought in totalities had developed instruments, imperfect but genuine, for keeping such phenomena in view. The analytical toolkit’s silence about them is a limitation of vision and not only of vocabulary.

Historical form is the second loss, anticipated in the discussion of Roemer’s models. The dialectical tradition, whatever its confusions, never forgot that its categories were historical products: that capital, wage labour, the firm and the market had been made, were being remade, and could be unmade. Analytical models, built for tractability, treat their categories as fixtures and study behaviour within them. The school’s historical writing, with Brenner’s honourable exception, thinned accordingly, and its theory of history, in Cohen’s austere version, purchased generality by abstracting from the very processes of category formation that a historical materialism might be expected to centre. A reader who comes to the school from Marx’s historical works, the Brumaire above all, will feel the temperature drop: the crowded, contradictory, darkly comic world of that text, in which classes fracture, traditions weigh like nightmares and political forms take on lives of their own, has no counterpart in the models, and the school’s defenders rarely pretended otherwise. Whether the loss is a necessary price of rigour or a self inflicted narrowing is the deepest question the ledger poses. The verdict section returns to it.

A third loss concerns the theory of ideology and of the theorist’s own position. Dialectical Marxism, in its Lukácsian line, held that social knowledge is itself socially situated, and that the standpoint from which a totality is viewed affects what can be seen of it. Analytical Marxism proceeded as though that problem had been solved by the ordinary canons of inquiry, and its works are largely innocent of reflection on the conditions of their own production: who funds the models, which questions the profession rewards, why exploitation theory flourished in the seminar rooms of wealthy universities during decades when organised labour was retreating. Innocence of that kind is not refutation, and the analysts were right that a standpoint confers no automatic truth. Lukács’s stronger claims for proletarian epistemology have worn badly. Yet a social theory that brackets its own situation leaves a dimension of its subject unexamined, and the dialectical tradition’s insistence on that dimension, however entangled with untenable epistemology, named a genuine task that the school mostly declined. Critical theory and the sociology of knowledge have since pursued the task with other tools, which suggests the loss was of a function rather than of an irreplaceable method. The ledger records it as a loss partly made good from outside the school.

The ledger itself can now be drawn. The gains and losses ledger, set out below, lists one row per consequence of dropping dialectics, with the strongest defence of each trade and a judgement on whether the loss has since been made good. It compresses arguments made in prose across this article, and readers should treat it as an index to those arguments rather than as a substitute for them.

Consequence of dropping dialectics	What analytical Marxism gained	What it gave up	Strongest defence of the trade	Loss made good since?
Dialectical logic abandoned	Claims that state their content and can be tested, disputed and revised	A fused vocabulary for fused antagonisms	Decomposition can exhibit coupling case by case without a special logic	Partly, through careful usage that retains the word under discipline
Microfoundations demanded	Mechanisms linking structures to situated action	Pattern level explanation that outruns its mechanisms	Patterns without loops invite teleology and false confidence	Partly, through critical realist accounts of structure and activity
Exploitation remodelled	A precise, checkable theory grounded in asset distribution	The labour process as the lived seat of exploitation	The moral core, unequal property, survives translation intact	Largely, within formal theory, less so in workplace studies
Totality set aside	Models of fragments that cumulate and travel	The structured whole as an object of theory	Totality functioned too often as a promissory note	Weakly, through synthesis attempts that remain unratified
Categories treated as fixed	Tractability, comparison across cases	Historical formation and dissolution of categories	Bracketing is the condition of rigour in any single study	Partly, through historical work at the school’s edges
Standpoint reflection dropped	Ordinary canons of inquiry applied without exemption	Theory of the theorist’s own social situation	A standpoint confers no automatic truth	Partly, by other traditions rather than by the school

Reading the ledger vertically teaches more than reading any row alone. The gains cluster around accountability: claims made answerable to evidence, definitions, and hostile readers. The losses cluster around scope: the whole, the historical, the reflexive. A tradition choosing the analytical trade was choosing accountability over scope, and the choice looks different depending on the state of the tradition at the moment of choosing. In the late nineteen seventies, Marxist discourse in the West was long on scope and short on accountability, rich in totalising gesture and poor in claims that could be taught, tested and built upon by unsympathetic readers. The trade was, on that diagnosis, well judged. The school’s error, where it erred, was to treat a corrective as a constitution: to elevate the standards appropriate to a rescue operation into a permanent definition of social theory, and to dismiss as nostalgia every attempt to recover scope under the new discipline. The recovery attempts, critical realism prominent among them, were themselves made possible by the correction, since nobody after the September Group could responsibly write totality talk in the old unexamined way. Loss and gain thus interpenetrate in the historical record, which is the closest this article comes to a dialectical observation of its own.

A break with Marx, or a quarrel inside his house?

The question of fidelity hangs over the whole episode, and it is usually posed too crudely to answer. Whether rejecting dialectics breaks with Marx depends on what in Marx the dialectics is, and that question divides into three, with different answers. Textually, dialectical vocabulary in Marx is real but unevenly distributed: dense in the early writings and in the drafts known as the Grundrisse, intermittent in the journalism and political works, and present in Capital chiefly in the architecture of the exposition and in a handful of celebrated formulations, the negation of the negation in the chapter on the historical tendency of accumulation among them. A reader can therefore construct, by selection, either a Marx whose method is dialectical through and through or a Marx whose substantive arguments proceed by ordinary causal and historical reasoning beneath a Hegelian presentation. Cohen and Elster constructed the second Marx with great learning. Their opponents constructed the first with equal learning. Textual evidence alone does not adjudicate between the constructions, because the corpus genuinely supports both emphases, a fact that partisans of each side find inconvenient.

The second question is systematic rather than textual: does the content of Marx’s mature theory require dialectical logic for its statement? Here the analytical case is at its strongest, and the preceding sections have assembled its evidence. The theory of history was restated without loss by Cohen. The theory of exploitation was restated, and in Roemer’s judgement improved, without the labour theory of value, let alone without dialectics. Class analysis was operationalised by Wright in forms a survey could carry. Against that record, the defender must exhibit a substantive Marxian theorem, not a mood or a reading experience, that resists restatement, and the candidates produced in a century of trying remain the ontological claims about internal relations and historical form discussed above: important, contested, and concerned with the framework of theorising rather than with any theorem inside it. On the systematic question, therefore, the verdict leans analytical. Nothing Marx is standardly credited with discovering appears to depend for its truth or its evidence on a special logic.

The third question is about identity and allegiance, and it is the one the participants themselves cared about most. The September Group’s members, with individual exceptions that widened over time, understood themselves as Marxists conducting an internal reform, and their self understanding was neither cynical nor merely tactical. Cohen’s socialism deepened across his career. Wright spent decades attempting to give the tradition an empirically respectable class analysis and ended with a programme of real utopias intended as a successor project, openly continuous with the Marxist critique of capitalism. Van Parijs argued for basic income with tools learned in the group and aims taken from the tradition. Elster and Roemer travelled furthest from the fold, and even their departures were argued, step by step, in terms the tradition supplied. The fairest description of the episode is therefore neither apostasy nor orthodoxy. It is a demonstration that a tradition’s identity can migrate from its method to its questions: analytical Marxism kept the questions, about exploitation, class, historical change and the justification of equality, and surrendered the method that had been treated as the tradition’s soul. Whether a tradition so redefined is still itself is a question traditions answer by living, and the living continues in scholars who cite Cohen and Ollman in the same bibliography without feeling a contradiction, in any sense of that overworked word.

A subordinate fidelity question deserves notice because it shaped the school’s reception in the wider movement. Activist and party Marxisms experienced the analytical intervention as an attack from the academy at a moment of political defeat, and responded by questioning the school’s right to the name. The response was understandable and mostly unproductive. It spared the movement’s intellectuals the labour of answering arguments they could instead excommunicate, and it confirmed the analysts in their suspicion that dialectical loyalty functioned as institutional protection. The healthiest engagements came from Marxists who accepted the analytical challenge as a challenge: Alex Callinicos, whose Making History defended a structured account of agency against both Elster and the structuralists, and the critical realists, who rebuilt the ontological case on new foundations. Their work shows what the tradition looked like when it treated the September Group as an occasion for its own clarification rather than as an enemy to be denounced, and the contrast with the excommunicating response is instructive for any reader weighing the school’s net effect.

Dispersal and self-criticism

Schools founded on a standard rather than a doctrine have a characteristic fate: the standard succeeds, diffuses, and the school dissolves into the disciplines it reformed. Analytical Marxism followed that trajectory with unusual fidelity. Through the nineteen eighties the September Group’s programme expanded and ramified, producing the multi volume Foundations of Analytical Marxism and a generation of students trained in its methods. By the early nineteen nineties the conditions that had called it into being were dissolving. The collapse of the Soviet Union removed the political backdrop against which Marxist theory had defined itself, and academic Marxism contracted sharply. Members’ interests diverged along the lines their tools suggested. Cohen completed his migration to normative philosophy. Elster pursued a general science of mechanisms and the emotions, treating Marx as one historical exhibit among others. Roemer addressed equality of opportunity and intergenerational transfers. Wright turned to institutional design and democratic experiments. Van Parijs to basic income and linguistic justice. The group continued to meet for a period, but the collective project had been replaced by distinguished individual ones, and later gatherings had the character of reunions as much as of working sessions.

The dispersal was accompanied by a self criticism more candid than most schools manage, and it belongs in the record because it bears directly on the dialectics verdict. In a widely discussed reassessment, Levine, Sober and Wright asked what had gone wrong with the Marxist project the school had set out to rescue, and conceded that the rescue had altered the patient: rational choice modelling had imposed Procrustean simplifications, the tradition’s concern with historical dynamics had atrophied, and the school’s Marxism had thinned into a set of detachable theses whose connection to any emancipatory practice had become tenuous. Wright’s subsequent work can be read as an attempt to answer that self criticism from inside the analytical inheritance, restoring history, institution building and political strategy to the centre while keeping the standards of clarity the school had taught him. Cohen’s late work performed a parallel restoration on the normative side, insisting that socialism required not only feasible institutions but an ethos, a communal principle governing everyday conduct, which no mechanism in the school’s toolkit could model and which he defended in deliberately unfashionable terms. The school’s two most loyal members thus ended by revaluing precisely the dimensions, historical process and lived ethos, that the founding purge had slighted. Neither returned to dialectics. Both returned, chastened, to scope.

Critics from the dialectical side read the dispersal as a verdict on the method: the solvent that dissolved dialectics proceeded, predictably, to dissolve Marxism, leaving its practitioners as distinguished contributors to liberal egalitarianism, mechanism based social science and formal economics, fields that had never needed the Marxist detour. The reading has evidence behind it and should not be waved away. It is nonetheless incomplete, for two reasons the record supports. The liberal and social scientific fields in question were themselves altered by the passage of the Marxists through them: exploitation is discussed in mainstream moral philosophy in terms Roemer fixed, the theory of history is taught through Cohen’s reconstruction even by its opponents, and class analysis in sociology bears Wright’s imprint on scholars with no Marxist allegiance. A tradition can survive by colonisation as well as by continuity. And the counterfactual remains unflattering to the critics. The dialectical Marxisms that declined the analytical challenge did not, in the same decades, preserve a flourishing alternative; most contracted into textual scholarship or activist subculture, keeping the method and losing the audience. The school’s fate therefore cuts both ways. It demonstrates that method without allegiance drifts, and it demonstrates that allegiance without method stagnates, and a reader drawing a single moral from the episode will have to choose which failure weighs more heavily. This article records both and chooses neither, since the choice depends on what one thinks a theory is for, which is a question the episode clarifies without settling.

The dispersal also had a generational mechanism worth naming, because it explains why the school could not reproduce itself even where its standards triumphed. A school reproduces through students who take its problems as their own and its canon as their starting point. The September Group’s students, trained across philosophy, economics, sociology and political science departments, took the standards and scattered with them into their home disciplines, where the rewards, journals and appointments were. Few departments anywhere could offer a young scholar a career as an analytical Marxist, as opposed to a career as a philosopher, economist or sociologist who had passed through the school’s arguments. The tradition’s older forms had reproduced through parties, movements and dedicated institutes that paid in loyalty what they lacked in prestige; the analytical school, having defined itself against those institutions, had no successor body of its own, and its dissolution was therefore written into its sociology from the start. A method that lives in disciplines does not need a school to survive it. A movement does, and the distinction between a method and a movement is, once again, the whole story.

The translation test as a working criterion

It is time to state the article’s organising instrument with the formality it has so far only promised. The translation test runs as follows. Take any claim advanced in dialectical vocabulary. Produce the most charitable restatement of the claim in the vocabularies of causal explanation, structural constraint and strategic interaction, drawing on mechanisms where the original gestures at processes. Compare the two versions. Whatever content the restatement preserves was never distinctively dialectical, however illuminating it may be. Whatever content the restatement loses must then be specified by the defender, in terms precise enough for a third party to judge whether the lost item is a genuine feature of the world or an artefact of the original phrasing. Where the lost item can be so specified, the defence scores, and the item enters the small class of claims on which the debate genuinely turns. Where it cannot, the presumption stands that the dialectical version owed its impressiveness to its obscurity. The test is not a machine for refuting dialectics wholesale. It is a procedure for converting a quarrel of allegiances into a series of answerable questions, and its history in the episode justifies it: nearly everything productive in the debate, on both sides, happened when a participant accepted the procedure and produced a case.

What would a dialectical claim have to do to pass the translation test and survive?

A claim survives when its defender can name a specific content, some relation, process or determination, that the plain restatement demonstrably omits, and can show that the omission matters for explanation or prediction. Candidates that meet the standard cluster around internally constitutive relations and historically formed categories.

Three features of the test deserve emphasis, because careless versions of it circulate. First, charity is built in: the restatement must be the strongest available, produced if possible by the claim’s defenders themselves, so that the test cannot be passed fraudulently by translating badly and then mocking the translation. Cohen’s reconstruction of the theory of history met that condition; it was the work of a sympathiser, which is why its outcome carried weight. Second, the test is symmetric in principle. A claim in analytical vocabulary can be subjected to the reverse procedure, asked what it presupposes about formation, totality and standpoint that its own statement omits, and the losses section of this article applied exactly that reverse audit to Roemer’s models. A criterion that only cuts one way is a weapon; a criterion that cuts both ways is a method, and the school’s best practitioners, whatever their rhetoric, used it as a method. Third, the test’s outcomes are revisable. A claim that fails against the restatement resources of one decade may succeed against a later panel of judges as vocabularies develop, and conversely, claims once thought untranslatable have repeatedly fallen to better modelling, as the fate of the labour theory of value showed. The test therefore delivers verdicts with dates on them, which is the most that honest method ever delivers.

The best candidates for survival, on the evidence assembled in this article, share a recognisable shape. They concern relations that constitute their terms, where the restatement as external interaction between antecedently given agents omits the constitution itself; they concern categories whose historical formation is part of the explanandum, where a model that takes the categories as fixtures omits the making and unmaking that the theory exists to explain; and they concern wholes whose organisation generates properties that organise the parts in return, where a sum of mechanisms omits the downward constraint of the organised whole. Each candidate has been met by partial analytical counters: constitutive rules can be stated, formation can be narrated and sometimes modelled, emergence can be given a realist interpretation that mechanism based inquiry can then investigate. The counters narrow the surviving residue without eliminating it, and the residue, small beside the vast territory the old dialectics claimed, is nonetheless real. A fair summing up is that the test destroyed dialectics as a logic, disciplined dialectics as a method, and located, inside dialectics as an ontology, a handful of problems that social theory neglects at its cost. That summing up is the article’s verdict in miniature, and it sets up the three test cases through which the instrument can be watched at work.

Three test cases: crisis, transition and ideology

Abstract criteria persuade less than cases, so consider three dialectical claims of descending fame, each passed through the test. The first concerns crisis. In the classical idiom, capitalist crisis expresses the contradiction between socialised production and private appropriation, maturing until the shell of property relations bursts asunder. Translated, the claim becomes a family of mechanisms: investment driven by competition outruns the consumption that wages can sustain; credit stretches across the gap and then snaps; profitability falls under pressures from rising composition of capital or from wage resistance; expectations, formed on extrapolated prosperity, reverse together. The translation is a research programme, parts of which have been executed with distinction, and it loses nothing that the original could cash out in evidence. What it loses is the original’s cadence of necessity, the suggestion that the outcome was written into the form itself. Defenders call that cadence content: the claim that crisis is not an accident befalling capitalism but a periodic product of its normal functioning. Analysts reply that the translated family says exactly that, in the only form in which it can be checked, by exhibiting the mechanisms whose normal operation generates breakdown. Verdict: translation succeeds, and the case counts for the school.

The second case concerns transition. The classical idiom describes feudalism negated by the bourgeoisie it nourished, and capitalism pregnant with its successor, each order generating the agent of its own supersession. Translated, the claim becomes comparative history: which classes, with what resources and alliances, under what state structures and international pressures, carried particular transitions, and with what unintended results. The translation loses the guarantee. In the original, the direction of travel is underwritten by the logic of negation; in the translation, transitions can fail, stall, regress or issue in forms nobody intended, as Brenner’s own comparative work on agrarian transitions demonstrated against simpler Marxist expectations. Defenders of the original idiom must then decide whether the guarantee was content or consolation. If content, it is false content, since the record contains too many stillborn and reversed transitions for any underwriting to survive audit. If consolation, the translation has done its work by exposing it. Verdict: translation succeeds again, and more instructively, because the case shows the test operating as an instrument of intellectual honesty against the tradition’s own hopes. A method that can disappoint its users is worth more than one that ratifies them.

The third case concerns ideology, and here the verdict reverses. The classical claim, in its sophisticated Lukácsian form, holds that the categories of thought available in a society are themselves shaped by the society’s structure, so that the analyst’s very instruments are internal to the object analysed; bourgeois economics, on this view, does not merely err about capitalism but thinks in capitalism’s own categories, mistaking historically specific forms for eternal ones. Translated into mechanism talk, the claim becomes a set of propositions about funding, professional incentives, training and the selection of tractable problems, all true and all studied by the sociology of knowledge. Something, however, escapes. The strong claim was not only that thought is influenced by position, which no analyst denied, but that the forms of appearance of a social order are generated by that order itself: that the wage form presents exploitation as fair exchange, that the commodity form presents social relations as relations between things, in the analysis Marx called fetishism. A restatement in terms of incentives and selection describes why people hold the appearances; it does not reproduce the claim that the appearances are the object’s own mode of presentation, true surface of a contradictory depth. Defenders can specify the lost item precisely, as the translation test requires: it is the doctrine that certain illusions are structurally generated and objectively grounded, rather than errors imposed by interest or ignorance. Analysts have replies available, treating fetishism as a thesis about systematic misperception that cognitive and institutional mechanisms can jointly explain, and the replies narrow the residue once more. They do not, in the judgement of this article, eliminate it, because the generation of appearance by structure is a claim about the object, not about the psychology of its observers, and mechanism talk is native to the second topic and a guest in the first. Verdict: a genuine remainder, small, specifiable, and exactly the kind of result the test was built to find.

What would settle the dispute

Sophisticated readers eventually ask of any long philosophical quarrel what could, even in principle, settle it, and the question usefully disciplines this one. Three kinds of result would move the verdict materially. The first would be a worked body of theory, not a single example, in which internally constitutive relations do explanatory work that the best available mechanism based rival demonstrably cannot reproduce: a domain where beginning from the whole yields predictions or explanations that beginning from situated individuals keeps missing, with the margin of victory specified in advance. Defenders believe the theory of social form already supplies such a body; the analytical side replies that belief has not yet been converted into the comparative demonstration. The challenge is well formed on both sides, and nothing prevents its being met in a future decade. The second kind of result would run the other way: a mature computational practice in which agents with endogenously formed preferences, identities constituted by their positions, and categories that dissolve and reform under the dynamics, are modelled with the same rigour the September Group demanded of statics. Agent based and complexity approaches gesture in that direction, and if the gesture matures, the analytical objection to dialectical ontology loses its practical force, since the allegedly untranslatable contents will have been translated by construction, and the defence will have won by absorption rather than by argument. The third kind of result is historical rather than technical. Traditions are also judged by their fruits across generations, and a future Marxist scholarship that is simultaneously mechanism respecting, totality grasping and historically deep would settle the quarrel in the only way intellectual history finally settles anything, by existing. Wright’s late work and the critical realist syntheses are partial drafts of that scholarship. Whether anyone completes it depends on scholars not yet trained, which is a properly undialectical note of contingency on which to leave the dispute: the outcome is not written into the structure of the question, and will be decided, if it is decided, by work.

How the debate is worked with now

The episode’s afterlife is most visible where Marxist theory is produced rather than commemorated. In the journals, the analytical standards operate as a floor: a submission trading in dialectical vocabulary will be asked, by referees formed in the aftermath, what relation its contradictions denote and what mechanisms its mediations name, and the asking is no longer experienced as hostility from a rival camp but as ordinary editorial hygiene. In research practice, the school’s genres have been absorbed and combined: a study of a labour process may move between mechanism based explanation, realist claims about structural powers, and historical narration about the formation of the categories involved, acknowledging debts to Cohen, to Bhaskar and to the traditions both men criticised, without feeling obliged to declare a winner. Keeping the working papers, model notes and successive drafts of such a project in order is itself part of the discipline the episode teaches, and readers assembling their own materials on this debate can use the structured notes at VaultBook’s Marxism study notes to keep claims, translations and counter translations side by side, which is the practical form the translation test takes at a researcher’s desk.

Teaching absorbed the episode differently, and the difference instructs. Undergraduate syllabuses in social and political theory standardly include the analytical Marxists as the terminal station of the Marxism unit, the point at which the tradition turns its tools on itself, and examination questions on dialectics are now almost unanswerable without the translation challenge, since the challenge defines what a critical answer must engage. School level teaching, by contrast, rarely reaches the school at all, and its treatment of dialectics remains at the level of definition and example, a gap the teacher layer below addresses briefly. The humanities received the episode more coolly than the social sciences, and literary and cultural studies continued to develop dialectical idioms, Adornian and Jamesonian lines prominent among them, largely untouched by Elster’s prosecution. That divergence is itself evidence in the case. Where the aim is explanation answerable to evidence, the analytical standards prevailed. Where the aim is interpretation, orientation or the criticism of cultural forms, the older idiom retained functions the translation test, by its own propositional design, cannot assess. Readers deciding what the episode means for their own work should therefore begin by deciding what their work is for, since the verdict follows the purpose more often than either camp admitted.

One further afterlife deserves record, because it concerns the word the group joked about. The Non-Bullshit standard escaped its origins and circulates now as a general test in intellectual life, applied to management theory, to policy science, to academic fashions far from Marxism: the demand that impressive vocabulary present its credentials in plain claims, mechanisms and evidence. The school’s members did not invent that demand, which belongs to the Enlightenment inheritance they shared, but they gave it a memorable local habitation inside the tradition least expected to welcome it. If analytical Marxism is remembered when its particular theorems have been absorbed or superseded, it will likely be for that act of importation: proof, performed on a hard case, that a tradition can survive the application of ordinary standards to its most protected possessions, and can even be strengthened by the ordeal, provided its members care more for its questions than for its armour.

Cohen’s theses under audit

Because Cohen’s defence of the theory of history served as the school’s founding demonstration, the fate of its two theses under subsequent audit is the best single measure of what the analytical method did for, and to, its inheritance. The development thesis claimed that productive forces tend to develop throughout history, grounded in three premises: human beings are rational, their situation is one of scarcity, and they possess intelligence capable of seizing chances to expand productive power when those chances appear. The audit began inside the school. Critics observed that rationality under scarcity can counsel many things besides development: rest, ritual expenditure, warfare, the refinement of consumption, and that whole civilisations have prized stability over growth for centuries without obvious irrationality. Cohen replied by weakening the claim’s scope, treating the tendency as real but defeasible, operative where no countervailing structure suppresses it, and he pointed to the long run record of expanding productive capacity as evidence that the tendency, however often interrupted, reasserts itself across millennia. The exchange left the thesis standing as a broad generalisation about the species rather than as a law of history, a demotion its author accepted with characteristic candour. A dialectical reader notes that the demoted thesis, a tendency working itself out against counter tendencies over long periods, resembles nothing so much as the older idiom’s talk of laws operating through their opposites, and may fairly ask whether the translation has here produced humility rather than clarity. The analyst answers that humility with specified exceptions is exactly what a tendency claim should be, and that the older idiom’s version of the same thought arrived with a guarantee the record never issued. Both descriptions fit the facts of the exchange, and readers should register the case as one where translation clarified the claim’s epistemic status more than its content.

The primacy thesis faced a harder audit. It claimed that the nature of a society’s production relations is explained by the level of development of its productive forces, the relations rising when they suit development and being swept away when they fetter it. The borderline cases arrived immediately. Knowledge, the most expansive modern productive force, sits awkwardly in the category of material forces: is a production technique a thing, a recipe, or a social relation in disguise? Cohen drew careful distinctions, counting instruments and labour power as forces and treating knowledge as embodied in them or in the labourer’s skills, but the boundary kept leaking, and with it the clean separation between the explaining level and the explained. A second difficulty concerned the direction of explanation in particular episodes. Property law, state violence and military competition have repeatedly imposed relations that no prior development of forces selected, and Cohen’s framework had to classify such cases as either temporary deviations awaiting correction or as evidence that the superstructure enjoys more autonomy than the primacy thesis allows. His handling was honest and ingenious, distinguishing the explanatory primacy of the base from any claim that nothing else matters, and introducing a gradation of cases from selection by the forces to mere compatibility with them. By the end of the audit the thesis had become a claim about the deep constraints within which history moves rather than a mechanism generating its turns. That is no small thing; constraint claims of that depth are rare and valuable in social theory. But it is a different thing from the engine of history that the Preface, on the traditional reading, had advertised, and the difference was produced not by enemies of Marxism but by the most sympathetic reconstruction the theory has ever received. The episode generalises into the article’s deepest finding about the school. Analytical rigour did not merely prune Marxism’s excrescences. It repriced its central assets, converting several of the tradition’s laws into tendencies, its tendencies into constraints, and its constraints into research programmes. Dialectics was the first and loudest casualty of the repricing, but the process that consumed it consumed a good deal else besides, and the school’s members, to their credit, mostly published the new prices rather than disguising them.

One further audit item belongs here because it shows the repricing at its most constructive. Cohen’s functional defence of primacy treated production relations as selected for their developmental effects, and the functional form, whatever its vulnerabilities, suggested a way of reading the superstructure that proved unexpectedly fertile. Legal and political forms, on the extended account, persist partly because they stabilise the relations the forces require, and that suggestion generated precise questions, about which legal innovations stabilised which property forms in which periods, that legal historians could actually investigate. A generation of scholarship on the co evolution of property law and productive organisation grew in that space, much of it indifferent or hostile to Marxism as a whole, which is precisely the school’s signature outcome: contributions detached from their origin and absorbed into the common stock. The dialectical tradition had discussed base and superstructure for a century without generating a comparable research yield, largely because its idiom discouraged the decomposition of the metaphor into investigable claims. Whatever else the ledger records, that contrast, between a metaphor contemplated and a metaphor cashed out, is the analytical achievement in its purest form, and any defence of the older method must explain why a century of dialectical commentary on the Preface produced so little cumulative inquiry beside the two decades of analytical work that followed Cohen’s book.

Mechanisms as the school’s second life

The anti-dialectical programme had an afterlife its founders only partly foresaw, and tracing it completes the gains side of the ledger. Elster’s demand for microfoundations matured, in his own later work, into a general methodology of mechanisms: explanations assembled from small, portable, frequently observed causal patterns, each specified at the level of individual action and interaction, which combine in concrete cases to generate the large outcomes that grand theory prefers to invoke wholesale. His image for the contrast was deliberately unglamorous. Grand theories offer nuts and bolts rhetoric about entire systems; mechanism based explanation offers cogs and wheels, an inventory of parts whose workings can be inspected, tested in isolation, and reassembled. The programme spread well beyond Marxism. In sociology, Peter Hedström and his collaborators built an explicitly analytical school on foundations owing more to Elster than to any Marxist text, treating the September Group’s standards as a constitution for social science generally. In political science, mechanism language became a standard demand in the evaluation of causal claims. In philosophy, the new mechanical philosophy gave the pattern a metaphysics. The school that began by refusing a vocabulary ended by exporting one, and the exported vocabulary had the property the imported one lacked: its users could say, of any given mechanism, exactly what it claimed and what would refute it.

The export had a cost that returns the article to its theme. Mechanisms travel by being detached from the contexts that produced them, and the mechanism inventory of the mature programme is pointedly unhistorical: patterns such as sour grapes adaptation, the bandwagon, the self fulfilling prophecy and the tragedy of collective action are presented as permanent features of human interaction, available in every period and regime. A Marxist reader is entitled to mixed feelings. On one side, the permanence is the point and the power: these are the closest things social science possesses to laws, and their generality is what the dialectical laws falsely promised. On the other side, a theory of society assembled from transhistorical parts has quietly abandoned the founding Marxist insistence that the forms of interaction themselves, the very shape of preference, belief and strategy, are historically made, and that a science of capitalism must therefore differ from a science of interaction in general. Elster, who had ceased to describe himself as a Marxist, could accept that consequence with equanimity. Those who remain inside the tradition cannot, and their attempts to historicise the mechanism programme, by asking which mechanisms a given mode of production selects, amplifies and suppresses, constitute one of the more promising unfinished projects the episode left behind. It is, in structure, the translation test running in its reverse direction: the analytical inheritance being audited by the historical question it bracketed, with the outcome still genuinely open.

A final observation about standards closes the substantive argument. The September Group is sometimes remembered as a episode of Anglo Saxon common sense defeating continental depth, and the memory flatters everyone involved at the cost of the truth. Its leading figures were a Welsh philosopher formed at Oxford, a Norwegian polymath formed in Paris and Oslo, an American economist, and a company drawn from several countries and disciplines, arguing in a second language for some and about texts written in German by a thinker formed on Greek philosophy and French politics. National styles explain little here. What the episode actually displays is a recurrent possibility inside any mature tradition: that a generation will arise which loves the tradition’s questions enough to risk its vocabulary, submits the vocabulary to audit, and discovers, claim by claim, which parts were load bearing and which were decoration. Dialectics emerged from that audit reduced, disciplined and, in its surviving ontological core, more precisely located than a century of devotion had left it. Marxism emerged repriced. Both outcomes are the audit’s work, and the audit, rather than either camp’s banner, is what this article has tried to hand to the reader.

For students and exam candidates

Students meet this material most often as the evaluation section of a dialectics question, and the winning structure is the one this article has modelled. State the analytical charge precisely: dialectical claims either restate plain causal or structural points obscurely or dissolve under inspection, supported by the demand for microfoundations and the translation challenge. Then state the strongest defence, internal relations or emergence, with a named case such as capital and wage labour. Then judge by cases rather than wholesale: the charge succeeds against dialectics as a logic, partially succeeds against fused uses of contradiction, and leaves a residue around constitutive relations and fetishism that a good answer names explicitly. Answers that award total victory to either side read as unexamined. Answers that stage the translation test on one worked example read as command of the topic. Cohen, Elster and Roemer should appear with their distinct contributions separated, since conflating them forfeits the precision marks the topic exists to award.

For teachers and seminar leaders

This topic suits undergraduate teaching better than school teaching, because its substance is methodological and its protagonists reward slow reading. A productive seminar format mirrors the September Group itself: circulate a short dialectical passage in advance, require each participant to bring a written translation of its central claim into causal or strategic language, and spend the session comparing translations and hunting for residue. The exercise teaches the material and the method simultaneously, and it reliably surfaces the difference between stylistic difficulty, which translation removes without loss, and genuine ontological commitment, which translation exposes for debate. Cohen’s opening chapters and a section of Elster’s Making Sense of Marx make a well matched pair of set texts, with Ollman or Bhaskar supplying the reply. Teachers pressed for time can compress the unit into a single session on the translation test, using the crisis and fetishism cases from this article as the worked examples, and students will retain the instrument long after the dramatis personae fade.

For researchers, historians, and journalists

Researchers can treat the translation test as a working criterion rather than a historical curiosity. Applied prospectively, it disciplines the writing of theory: every claim about totality, contradiction or emergence in a draft should be able to survive the question of what its mechanism level or constitutive content is, and the discipline costs little where the claim is sound. Applied retrospectively, it supplies a historiography of the school itself, since the dispersal of analytical Marxism into neighbouring disciplines is the best documented case of a theoretical programme dissolving through its own success. Historians of social science will find the September Group an unusually legible object, small, self documenting and argumentative in writing. Journalists covering Marxist ideas should note the episode’s central correction to the standard caricature: the most damaging critique of dialectical Marxism came from Marxists, was motivated by fidelity to the tradition’s questions, and cannot be reported as an external assault without falsifying the record.

The verdict, kept double

The instruction this article set itself was to give both cases at full strength and then judge without awarding a coronation, and the judgement can now be collected. Against dialectics as a logic, as a set of laws overriding ordinary inference and guaranteeing history’s direction, the analytical prosecution succeeded completely and permanently; no defender of standing has restored the laws, and the tradition’s own best work no longer asks them to bear weight. Against dialectics as an unexamined idiom, a fused vocabulary in which contradiction, totality and negation circulated as credentials, the prosecution succeeded as a discipline, visible in the cleaned standards of everything written since, including most writing that calls itself dialectical. For dialectics as an ontology, a claim about internal relations, historical form and structurally generated appearance, the prosecution located real problems and then, in its austere moods, mistook the location of a problem for its dissolution; the residue identified by Ollman and Bhaskar survives, narrowed and clarified, as a standing agenda for social theory. And about the school itself the record supports the double judgement entered at the start: a rescue that transformed its patient, a standard that outlived its school, and a group of scholars whose argument with their own inheritance remains the most instructive quarrel modern Marxism has produced. The reader who closes this article able to run the translation test on a fresh claim, and to say what would count as its surviving residue, possesses the episode’s essential yield, which is not a side but an instrument.

One caution should accompany the instrument out of the door. The translation test audits claims; it does not generate them, and a tradition that only audits will eventually have nothing left to examine. The September Group could afford its severity because it inherited a century of rich, reckless theorising to work on, and its own positive constructions, Roemer’s models, Cohen’s defence, Wright’s class maps, came from members who theorised first and audited second. The proportion matters for anyone tempted to make a career of the audit alone. Dialectics at its worst was generation without audit, all system and no check. Analysis at its worst would be audit without generation, all check and no system, and the symmetry of those failure modes is the final lesson the episode teaches. The school’s founders, whatever their polemics, lived on the generative side of the line often enough to leave a shelf behind them. Their critics in the dialectical tradition, at their best, have done the same under the discipline the school imposed. Between those two facts the quarrel dissolves into a division of labour that neither party planned and both parties, on the evidence of their later work, have quietly accepted.

The exposition question: dialectics inside Capital

One text resisted the school’s treatment more stubbornly than any other, and it was the tradition’s central one. Capital is not organised as a treatise of propositions. Its first chapter begins with the commodity, unfolds the value form through a sequence of expanding shapes, exhibits money as the necessary form of appearance of value, and only then approaches capital, in an architecture that generations of readers have compared, with Marx’s encouragement, to Hegel’s Logic. Defenders of dialectics regard that architecture as the method in action: each category is shown to be inadequate in isolation, to pass into its complement, and to be preserved in a richer whole, so that the order of exposition is itself an argument about the object, whose forms really are internally connected in the way the presentation displays. An analyst who extracts propositions from such a text, the defender continues, is like a reader who summarises a proof in geometry as a list of true statements about triangles: the list may be accurate while missing the thing the proof was for.

The analytical response developed in stages, and its mature form is more interesting than a flat denial. Cohen’s reconstructions deliberately bracketed the order of exposition and asked after truth: whatever the architecture’s pedagogical or rhetorical virtues, the claims Capital advances about value, surplus and accumulation either hold or fail, and holding or failing is independent of the sequence in which they are encountered. Elster added that the architecture had demonstrably misled readers, including Marx himself at his least careful, into treating transitions between categories as though they were transitions in history or in logic, when the genesis of money from the value form is neither a historical event nor a deduction but a piece of conceptual scene setting whose fruits must be judged downstream. Yet a concession grew within the analytical camp as its members aged. The value form analysis, the part of Capital most saturated with dialectical presentation, turned out to contain a claim, about the social form of wealth and the fetishism it generates, that resisted the school’s standard extractions, as the ideology test case above recorded. Later scholarship on the value form, some of it by writers with no dialectical allegiance to defend, rebuilt that material with analytical care and found content the first reconstructions had passed over. The episode within the episode carries the by now familiar moral. Translation audited Capital and repossessed most of it for plain argument; the residue it could not repossess was concentrated exactly where the defence had always said it would be, in the theory of form.

The road not taken: structuralism and the French alternative

The September Group was not the only attempt of its period to rescue Marxism by reforming its method, and comparison with its great rival sharpens what was distinctive in the analytical choice. In France, Louis Althusser and his students had proposed, a decade earlier, a different purification: Marxism as a science of structures, purified of humanism and historicism, in which history is a process without a subject, social formations are structured in dominance, and events are overdetermined by the interaction of relatively autonomous instances. The structuralist programme shared the analysts’ impatience with Hegelian expressivism, their suspicion of the philosophy of consciousness, and their insistence that Marxism submit to standards beyond devotion. It differed in everything else. Where the analysts grounded explanation in individuals and mechanisms, the structuralists grounded it in structures and their effects, and treated methodological individualism as the ideology of bourgeois social science. Where the analysts prized the plain sentence, the structuralists cultivated a technical vocabulary of their own, whose opacity Elster’s generation found no more excusable for being French.

The rivalry was never a formal debate, and the two programmes mostly developed in mutual ignorance punctuated by contempt, yet history staged a comparison anyway. Structural Marxism produced distinguished historical work and a generation of concepts, overdetermination among them, that migrated into cultural theory, but its central theoretical claims proved difficult to state in forms that successors could test, refine or refute, and its school dissolved amid political disillusion without leaving a cumulative research programme in the social sciences. Analytical Marxism, whatever its losses, left exactly such a programme, together with the theorems, models and datasets by which outsiders could judge it. Defenders of the dialectical tradition sometimes enlist this comparison on their own side, observing that structuralism, for all its anti-Hegelianism, preserved a notion of the structured whole that the analysts surrendered, and that its concepts survived in the humanities where questions of form and meaning dominate. The observation is just, and it refines rather than rescues the comparison. Each reform kept what its home disciplines valued: Paris kept the whole and lost the audit; Oxford and its allies kept the audit and lost the whole. The ledger of this article is, in that sense, the record of a choice that the entire generation faced and that no faction escaped paying for somewhere.

Rational choice and the politics of class compromise

The school’s method faced its most revealing political test when Adam Przeworski turned it on the strategic history of European labour. The classical Marxist expectation, rarely stated as a theorem but pervasive as an assumption, held that workers, as they became a majority and recognised their interests, would vote and organise their way towards socialism, so that history and arithmetic were allies. Przeworski modelled the situation instead as a sequence of choices under constraints: workers choosing between electoral participation and revolutionary abstention, parties choosing between class purity and the broader appeals that winning elections requires, unions weighing present wages against investment and employment. The results, developed with John Sprague in Paper Stones, published in 1986, were sobering in a way dialectical consolation had obscured. Electoral participation itself transformed the participants: parties seeking majorities diluted their class appeal, workers’ identities competed with national, religious and occupational ones for the same voters, and the transition to socialism by ballot confronted dilemmas, about investment strikes, capital flight and the sequencing of reforms, that no accumulation of votes dissolved. History had not failed to ripen. The strategic structure had never contained the expected outcome.

Traditionalist critics objected that the modelling assumed the very individualism under dispute, treating class identity as one preference among others rather than as a constituted standpoint. The objection connects with the internal relations defence and shares its partial force. What the episode demonstrated, however, was the analytical method’s capacity to disappoint its own side’s hopes with evidence, a capacity whose absence in dialectical practice had been Elster’s central complaint. A framework in which the expected outcome fails to appear is embarrassed, and must choose between auxiliary hypotheses and revision; Przeworski’s work forced revision, and the tradition’s understanding of social democracy was permanently altered by it. When the school is accused of having been a solvent, this case is the strongest counter exhibit to set beside Roemer’s models: here the method did not dissolve a Marxist claim into liberal common sense but produced a harder, sadder Marxist knowledge, about the structural traps awaiting labour movements, that the older idiom’s confidence had prevented the tradition from acquiring. Knowledge that hurts is the surest sign that a method is a method and not a liturgy.

The same appetite for strategic structure illuminated smaller questions across the school’s work: why revolutions occur in the wrong countries by the theory’s timetable, which becomes a question about state breakdown, elite division and peasant organisation rather than about ripeness; why workers accept compromises their grandparents struck as betrayals, which becomes a question about credible commitment and the shadow of future interactions; why some welfare states decommodify labour deeply and others barely, which becomes a question about coalition geometry among classes and fractions. Each question survived translation, gained precision, and lost the eschatological frame in which the tradition had first posed it. Whether the loss of the frame counts as a loss of knowledge or of illusion is, once more, the whole dispute in miniature, and by this point in the article the reader has the materials to judge it case by case, which is all the translation test ever promised to provide.

Wright between two camps

Erik Olin Wright’s position inside the school illuminates its internal range better than any other biography, because Wright was the member most determined to keep what the others were willing to bracket. His class analysis, developed from the mid nineteen seventies onwards, began from a problem the dialectical tradition had handled by gesture: the location of managers, professionals and supervisors who neither own capital nor resemble classical workers. His answer, the doctrine of contradictory class locations, treated such positions as simultaneously inside opposed class relations, sharing in the domination of capital while remaining subordinate to its owners. The vocabulary of contradiction in that doctrine was not Elster’s target; it denoted a specifiable duality of interests and experiences, and Wright proceeded to do with it what the school’s standards demanded, operationalising the locations in survey instruments, testing their associations with income, attitudes and organisational behaviour across several countries, and revising the scheme when the data objected. Contradiction, in his hands, survived translation so completely that its dialectical ancestry became a biographical footnote. Critics who cite Wright against the school therefore misdescribe him. He was the school’s demonstration that its standards could be turned on the tradition’s favourite word without destroying the underlying insight, provided the insight was real.

Wright also became the school’s most persistent internal dissenter on the question this article has tracked. Trained as a sociologist, he never fully accepted methodological individualism in Elster’s austere form, and his mature methodological writings defend a layered picture in which structures, institutions and individual action each have their explanatory place, with mechanisms as the connective tissue rather than as the foundation. His engagement with critical realism in his later career was the natural expression of that position: an attempt to hold the September standards of clarity and evidence together with an ontology roomy enough for emergence, historical process and transformative projects. His final programme, the study of real utopias, institutional designs that embody egalitarian values within existing societies, can be read as the school’s own translation test applied to its politics. The dialectical tradition had promised a successor society gestating within the old; Wright replaced the promise with a portfolio of partial, inspectable designs, credit unions, participatory budgets, basic income schemes, whose merits could be argued one by one. Something was unmistakably lost in the replacement, the grandeur of a total alternative, and something unmistakably gained, proposals that could be tried, costed and criticised without awaiting the negation of the negation. Readers weighing the whole episode could do worse than take Wright’s career as its emblem: the tradition’s questions retained, its method surrendered, its hopes rebuilt at smaller scale and higher resolution.

Obscurity reconsidered: what difficult writing is for

One premise of the analytical case deserves separate scrutiny, because it is the premise most often smuggled past the reader: that clarity is an unqualified virtue and obscurity an unqualified vice. The school’s rhetoric treated the matter as settled by the nickname, and much of its persuasiveness with younger scholars rested on the sheer relief of reading Cohen after the textbooks. Yet the philosophy of language and the history of science both complicate the picture. Difficult writing has functions beyond evasion. It can mark the strain of saying something for which the available vocabulary is badly shaped, as Ollman argued dialectical prose does for internal relations. It can slow a reader into noticing assumptions that fluent prose smuggles past, a function the Frankfurt School prized and exploited deliberately. And it can protect a young research programme from premature demands for a precision it cannot yet supply, a protection several mature sciences enjoyed in their infancy. A critic armed with the translation test should therefore apply it with a sense of timing. Demanding instant translatability from a formed doctrine with a century of commentary behind it, as the analysts demanded of dialectics, is fair. Demanding it of every difficult text on first encounter is a recipe for mistaking one’s own incomprehension for the author’s emptiness, an error Elster’s fiercer pages occasionally commit against Hegel himself.

The concession has limits, and they matter as much as the concession. None of the legitimate functions of difficulty licenses permanent immunity. A programme that invokes the strain of new content must, within a reasonable interval, produce the content in forms others can use, as the value form literature eventually did for parts of Capital and as critical realism did for parts of the dialectical ontology. A style that slows readers must repay the slowness with perceptions unavailable at speed, and readers are entitled to audit the repayment. The analysts’ lasting contribution to this question was to shift the burden of proof. Before the September Group, obscurity in Marxist writing enjoyed a presumption of depth, and the reader who failed to penetrate a passage blamed his own formation. After the group, the presumption runs the other way: the writer who cannot, on demand, descend from the dialectical passage to its plain content is presumed to be hiding a vacancy, until shown otherwise. Presumptions can be rebutted, and the defence sections of this article rebutted this one in specified places. But a presumption is not a verdict, and the school’s occasional habit of treating the shifted burden as the end of the argument, rather than its beginning, was the point at which its rigour shaded into the philistinism its critics alleged. The mature position keeps the burden where the school put it and the argument open where the school sometimes closed it.

Reading the protagonists

Readers who wish to go to the sources will find the episode unusually well documented by its participants, and a short orientation may save wasted effort. Cohen’s Karl Marx’s Theory of History remains the school’s most achieved book and the right place to start: its opening chapters perform the analytical method on the tradition’s central text with a clarity that even hostile reviewers conceded, and its later expanded edition adds retrospective essays in which the author audits his younger self. Elster’s Making Sense of Marx is the essential polemic, best read with a defender open beside it, since its sieve metaphor invites the reader to notice what the mesh was designed to catch. Roemer’s A General Theory of Exploitation and Class demands more technical patience and repays it with the episode’s most constructive achievement, while the anthology he edited under the school’s name offers the programme in miniature across several authors. For the defence, Ollman’s Dialectical Investigations states the internal relations case at full length, and Bhaskar’s philosophical works supply the realist ontology in its systematic form, at the price of a difficulty that will test the reader’s sympathy for the arguments of the preceding section. Wright’s methodological essays and his late work on real utopias close the circle, showing where the school’s most loyal sociologist finally stood. Read in that order, the shelf tells the story this article has told in summary, and readers can run the translation test for themselves on the original materials, which is the only examination of the episode that finally counts.

Frequently Asked Questions

Q: Why do analytical Marxists reject dialectics?

They reject dialectics because, on examination, its claims divide into kinds they cannot use. Some restate ordinary causal or structural points in needlessly obscure language, as when conflict between classes is redescribed as the unity of opposites. Some assert a special logic of contradiction that no defender has stated in a form others can test. Some, such as the negation of the negation, fit any sequence whatever and therefore explain none. Cohen showed that the theory of history could be defended without the apparatus, Elster prosecuted the apparatus directly, and Roemer rebuilt exploitation theory on foundations that owed it nothing. A tool that no job requires, and that obstructs several jobs, is a tool the school laid down.

Q: What was the September Group?

The September Group was the informal circle of scholars whose annual meetings, held each September from the end of the nineteen seventies, gave analytical Marxism its social form. Its core included the philosopher G. A. Cohen, the social scientist Jon Elster, the economist John Roemer, the sociologist Erik Olin Wright and the political theorist Philippe Van Parijs, with historians, development economists and political scientists moving in its wider orbit. Members circulated work in progress and subjected it to prolonged, unsparing criticism from colleagues who shared their Marxist commitments but not their conclusions. The group had no manifesto, no membership roll and no doctrine beyond a standard: claims had to be stated clearly, defined before use and defended against the strongest objections available. Its influence came from the books its meetings disciplined.

Q: What did Elster say about dialectics?

In Making Sense of Marx and in his earlier methodological writings, Jon Elster argued that dialectical theses decompose under inspection. The transformation of quantity into quality is either a banal reminder that gradual change can have abrupt effects or an empty law that specifies no threshold and no mechanism. The unity of opposites conflates distinct relations, logical inconsistency, causal opposition, conflict of interest and functional interdependence, borrowing the force of each for the others. The negation of the negation fits any three stage sequence because the counter chooses where counting starts and stops. Behind the itemised critique stood a diagnosis: dialectical formulas survive because they confer depth on their user, license having things both ways, and protect conclusions from falsification. His remedy was explanation through intentions, beliefs, opportunities and mechanisms.

Q: Can you be a Marxist without dialectics?

The analytical Marxists built their school on the answer yes, and the school’s existence is the strongest evidence for it. Cohen defended the theory of history, Roemer rebuilt the theory of exploitation, and Wright operationalised class analysis, each while dispensing with dialectical logic entirely, and each retained the tradition’s defining questions about exploitation, class, historical change and equality. Defenders of dialectics reply that something essential is surrendered in the process: the ontology of internal relations, the grasp of the social whole, and the theory of historically formed categories. The dispute is therefore not about whether the word Marxist may be used but about what the tradition minimally contains. If its core is a set of questions and explanatory claims, dialectics is dispensable. If its core includes a method and an ontology, dispensing with them changes the tradition’s identity even where its questions survive.

Q: What is the demand for microfoundations?

The demand for microfoundations holds that explanations of large scale social phenomena are complete only when they can be grounded, at least in outline, in the actions, beliefs, desires and opportunities of individual people. A claim that a class imposed a law, or that a structure reproduced itself, is treated as a promissory note, redeemable on demand into an account of situated actors whose interactions generated the outcome. Elster pressed the demand hardest, arguing that consequence explanations without a feedback mechanism invite teleology, the illusion that history arranges outcomes because they are needed. Cohen resisted the strictest version, defending pattern explanations in advance of their mechanisms. The demand does not deny that structures constrain people. It denies that structures explain anything except through the people who compose, confront and reproduce them.

Q: What did analytical Marxism gain by dropping dialectics?

The gains cluster around accountability. Central categories had to state their content: contradiction had to name a relation, totality had to name a structure, and mediation had to name a process, so the fog in which assertion passed for argument dispersed. Positive theory followed, most visibly Roemer’s reconstruction of exploitation from the distribution of productive assets, which produced theorems a hostile reader could check, and Wright’s class analysis, which survey research could carry. Marxist theory reconnected with mainstream philosophy, economics and sociology after decades of insulation, and the reconnection ran in both directions. Finally, Marx himself became criticisable: stripped of dialectical protection, his claims could be vindicated, refuted or rebuilt one by one, which the school regarded as a more serious respect than deference. The gains were purchased, as the losses show, rather than found.

Q: What did analytical Marxism lose by dropping dialectics?

The losses cluster around scope. Totality, the grasp of a social formation as a structured whole whose parts take their character from their place within it, has no native home in a practice built from mechanisms and models, so wholes appear as background conditions rather than as objects of theory. Historical form suffered similarly: models take their categories, owner, worker, market, as fixtures, where the dialectical tradition never forgot that such categories are made, remade and unmade in history. Reflexivity thinned as well, since the school largely declined the older tradition’s question about the theorist’s own social standpoint. Defenders add the case of fetishism, where the claim that a social order generates its own appearances resists restatement in terms of individual belief and incentive. The school’s own later self criticism conceded several of these losses in substance.

Q: How do defenders of dialectics reply?

The strongest replies concede the textbook laws and relocate the doctrine. From the philosophy of internal relations, associated with Bertell Ollman, defenders argue that positions such as capital and wage labour are constituted by their relation, so a method that begins from independently given individuals has built its conclusion into its starting point. From critical realism, associated with Roy Bhaskar, they argue that structures possess emergent causal powers irreducible to their occupants, so methodological individualism mistakes a part of explanation for its foundation. From the reading of Capital, they argue that the value form analysis shows categories unfolding through their own inadequacies, a movement no list of extracted propositions reproduces. These replies do not restore dialectics as a logic. They identify a residue, constitutive relations, emergence and historical form, that the analytical toolkit handles awkwardly, and the debate continues on that narrowed ground.

Q: What does non-bullshit Marxism mean?

Non-Bullshit Marxism was the September Group’s joking name for its own enterprise, coined by the members rather than by critics and repeated by Cohen when recounting the group’s origins. The phrase targeted a recognised style of Marxist writing in which portentous vocabulary performed the work of argument: passages where invoking totality, mediation or the negation of the negation allowed a writer to skip the points at which definition, evidence or inference was required. Against that style the nickname asserted a house rule. State the claim. Say what would make it false. Show how the parts connect. A member who could not comply had not been profound but evasive, and the group reserved its rudest word for evasion. The joke’s serious edge explains its survival: it named, in four words, the standard by which the school asked to be judged and by which it judged the tradition it inherited.

Q: Is rejecting dialectics a break with Marx?

The question divides into three, with different answers. Textually, dialectical vocabulary in Marx is real but uneven, dense in the early writings and the Grundrisse, intermittent elsewhere, architectural in Capital, so selection can construct either a dialectical Marx or a substantially non dialectical one. Systematically, the analytical case is strong: the theory of history, the theory of exploitation and class analysis have each been restated without dialectical logic, and no substantive Marxian theorem has been exhibited that requires it. In terms of allegiance, the school’s members mostly understood themselves as conducting an internal reform, keeping the tradition’s questions while surrendering its protected method, and several deepened their socialist commitments as they did so. Rejecting dialectics is therefore a break with one influential image of Marx and with the orthodoxies built on it, and a continuation, by other means, of much that Marx argued.

Q: What did the September Group require of an explanation before accepting it?

The group required an explanation to state its claims as propositions whose content a hostile reader could identify, to define its central terms before trading on them, and to specify, at least in outline, the process connecting cause to effect. Where a claim explained a pattern by its beneficial consequences, members asked for the feedback loop through which the consequence sustains its cause, or for the claim to be demoted to a redescription. Where a claim invoked a collective actor, they asked how the actor’s conduct reduces to the situated actions of its members, or what organisational machinery does the reducing. Counterexamples were treated as data rather than as impertinence, and a position that survived a September meeting had usually been rewritten in the surviving. The requirements were never codified in a rulebook. They were enforced socially, by colleagues who shared the author’s commitments and were therefore impossible to dismiss as unsympathetic.

Q: How should a teacher present Roemer’s game theoretic account of exploitation?

A teacher should begin with the puzzle the account solves: the inherited theory grounded exploitation in the labour theory of value, whose technical difficulties had become severe, so the class needs to know what exploitation could mean without it. Then stage the withdrawal test as a thought experiment students can run themselves, with a small economy of agents holding unequal assets, asking which groups would improve their position by withdrawing with an equal per capita share. The definition lands best when students discover that the exploited group in the exercise is identified by its propertylessness rather than by any transfer inside production. Follow with the anomalies, exploitation without a labour market, labour transfer without exploitation, since anomalies teach the concept’s shape. Close by asking what the model brackets: formed preferences, given rules, history as an initial snapshot, which returns the class to the wider argument about what formal reconstruction gains and loses.

Q: Why does analytical Marxism reject dialectics as a logic while keeping Marx’s questions?

Because its members judged the two inheritances separable, and the judgement was argued rather than assumed. The questions, why exploitation persists, how class positions are generated, what drives large scale historical change, and what equality requires, seemed to them as urgent as ever, and independent of any particular method for pursuing them. The logic, by contrast, failed every test they applied: its laws specified no mechanisms, its central term fused several distinct relations, and its proudest results, in the theory of history and of exploitation, proved restatable in plainer vocabularies without loss. Keeping the questions while dropping the logic was therefore not eclecticism but the outcome of an audit, claim by claim, in which some parts of the inheritance earned their retention and others did not. Critics dispute the audit’s fairness at specified points, around internal relations and historical form, and the debate survives precisely because the audit can be re-run on those points.

Q: What is lost when class is rebuilt from individual choices alone?

Rebuilding class from individual choices captures the strategic surface of class life with unmatched precision: who can wait, who must settle, which coalitions are stable, why solidarity frays. What it tends to lose is the constitution of the choosers. A model begins with agents whose preferences, horizons and identities are given, but class analysis in the older tradition insisted that those inner furnishings are themselves products of position, formed by the workplace, the market and the culture a position supplies. It also loses the relational character of the positions themselves, since capital and labour in the model confront one another as owners of different bundles rather than as poles of a single relation that makes each what it is. Wright’s career inside the school can be read as a long attempt to recover those losses without surrendering the gains, by layering structure, institution and action instead of founding everything on choice.

Q: How does methodological individualism change the meaning of a class explanation?

It changes the explanation’s grammar and its evidence. In the older grammar, classes themselves are the actors: the bourgeoisie imposes, the proletariat resists, capital expands, and the verbs carry the explanation. Under methodological individualism those verbs become shorthand that must be redeemable, on demand, into accounts of owners, officials, organisers and workers acting on their beliefs and opportunities, with class outcomes emerging, often unintended, from their interactions. The change disciplines two habits at once. It blocks the attribution of interests to classes as though classes were persons with a single point of view, requiring instead that interest formation, information and organisation be shown. And it blocks the treatment of structural outcomes as self explaining, requiring the mechanisms that reproduce a class structure to be named. What survives the change is the relational definition of class itself, which Roemer’s models show can be stated with full precision inside the individualist grammar, grounded in the distribution of productive assets.