The comparison between analytical Marxism and the classical tradition it broke from is usually introduced as a quarrel about jargon, as if one side had merely chosen clearer words for doctrines both sides shared. That introduction misleads from the first sentence. The disagreement concerns whether Marxism possesses a method of its own, a way of reasoning that belongs to this theory alone and that licenses conclusions unreachable by ordinary argument. Everything else follows from the answer given to that single question. The analytical current answers no: a claim that cannot be stated plainly, defended with evidence and exposed to counterexample has not been deepened by its obscurity, only sheltered by it. The classical current answers yes: the categories of the theory are historically specific, the object of study is a totality whose parts take their character from the whole, and a method that isolates elements for separate inspection distorts the object before inquiry begins. A reader who grasps that prior fork, and grasps that each side argues in good faith from it, holds the key to every later dispute between the two camps, from the fate of the labour theory of value to the standing of the working class as historical agent.

Two seminar tables with the eight differences grid and the founding texts of both currents, Insight Crunch

This article is organised as a comparison and a decision. It first states the prior disagreement in full, because the brief for the comparison insists that the reader meet it before any enumeration of differences. It then derives eight differences from that source, setting them out in the eight differences grid, a findable table in which each row gives the analytical position, the classical position, the root from which the disagreement grows, and a judgement about which side the evidence favours where evidence can bear at all. After the grid, the article gives the longest hearing to the strongest classical objection, an objection aimed at method rather than at any single result: the charge that analytical models build their conclusions into their assumptions. A balanced account then records what each side got right, at equal length and without scorekeeping disguised as fairness. The closing verdict carries its own defeater, a statement of the finding that would reverse it, because a judgement that nothing could overturn is a loyalty statement rather than an assessment. Throughout, the comparison runs between stated methods and stated claims. It does not run between the political commitments of adherents, and the slide from methods to motives, common in polemics against both camps, is refused wherever it appears.

Two cautions govern the reading that follows. The first concerns naming. Classical Marxism here means the tradition that received Marx through the Second and Third Internationals, through the dialectical materialism codified in party schools, and through the Western Marxist writers who preserved dialectical and totalising habits of thought inside universities. Analytical Marxism means the loose school gathered around G. A. Cohen, Jon Elster, John Roemer, Erik Olin Wright, Philippe Van Parijs and their interlocutors, a school whose own history is told in the companion guide to analytical Marxism as a school, to which this comparison links rather than retells. The second caution concerns fairness of a particular kind. Each position is attributed to named proponents with named works, because a comparison between anonymous tendencies invites caricature, and caricature is the besetting sin of the literature this article tries to improve on. Where a writer changed position across a career, the earlier and later statements are distinguished, since several figures in the story argued against their own earlier books with a candour their critics rarely matched.

The generative disagreement at the heart of analytical Marxism

What is the core disagreement between analytical and classical Marxism?

whether Marxism has a distinctive method. The analytical current holds that valid arguments must be statable in ordinary logical form and that no special dialectical logic exempts them. The classical current holds that the method is inseparable from the content. Every other difference follows.

The sentence that does the work in that answer is the last one. Differences between schools are usually presented as a bundle, a cluster of positions that happen to travel together because the same people hold them. The claim developed across this article, named here as the generative disagreement, is stronger: the bundle has a generator. Take a position on whether the theory owns a proprietary method, and positions on value, exploitation, class, explanation, ethics, history and agency follow with something close to logical pressure. A thinker who grants a distinctive dialectical method will find functional and holistic explanation natural, will treat the labour theory of value as a moment of a larger whole rather than a free-standing price theory, and will resist the demand for individual-level mechanisms as a category error imported from a rival science. A thinker who denies any proprietary method will ask, of each doctrine separately, whether it can be restated clearly, modelled exactly and tested against evidence, and will keep or discard doctrines one by one as they pass or fail. The two tempers produce different literatures even when they begin from the same texts of Marx, which both camps read closely and quoted against each other for decades.

Cohen’s opening move in Karl Marx’s Theory of History, published in 1978, illustrates the analytical temper at its most constructive. Cohen announced that he would state historical materialism with the precision of analytical philosophy, distinguishing claims that the tradition had run together and defending the survivors. The book’s famous rigour was not decoration. By separating the development thesis from the primacy thesis, Cohen made each claim assessable, and assessability was the point: a theory stated so that evidence could tell against it had been given a chance to be knowledge rather than catechism. Elster pressed the same temper harder in Making Sense of Marx in 1985, sifting the corpus claim by claim, keeping what survived translation into clear prose and discarding what dissolved under it. Elster’s verdicts were often negative, yet the method of the book embodied the school’s founding wager, that Marxism would be strengthened, not betrayed, by being treated as a set of arguments answerable to general standards.

The classical answer begins from a different intuition about the object. Georg Lukács, in History and Class Consciousness in 1923, argued that orthodoxy in Marxism refers exclusively to method, and that the method in question grasps society as a totality in which facts receive their meaning from their place in a historical process. On this view, isolating an element, a wage bargain, a price, a preference, and examining it apart from the whole does not yield a smaller truth but a false one, because the element in isolation is not the element as it functions. The dialectical tradition descending from Hegel through Marx holds that contradictions internal to a social form drive its transformation, and that a logic of static identity, which sorts things into fixed classes and asks what follows, systematically misses movement, mediation and reversal. When analytical writers demanded that dialectical claims be restated in ordinary logical form, classical writers heard a demand that the theory surrender its object as the price of admission to discussion.

Between these tempers there is no neutral referee, and the article does not pretend to supply one. Standards of clarity are themselves part of what is disputed, since the classical side regards the demand for isolated statement as a substantive commitment wearing procedural dress. Equally, the analytical side regards the appeal to totality as a substantive evasion wearing methodological dress. Recognising that the dispute is about the rules of dispute explains its longevity and its heat. Arguments conducted under shared rules can be settled by the evidence both sides accept. Arguments about which rules govern can be settled, if at all, only by showing what each rule set delivers across a run of cases, which is why this article proceeds by consequences: eight differences, traced back to the fork, and then weighed by what each branch made visible and what each branch hid.

Why does the method question outrank every doctrinal question?

Because doctrines inherit their fate from the standards used to judge them, so settling any single doctrine while the standards remain disputed settles nothing beyond that case.

A verdict on value theory reached by modelling will be rejected by a reader who distrusts models, and a verdict reached dialectically will be refused by a reader who distrusts dialectics, whereas the method dispute, once seen clearly, predicts both rejections and explains them.

The practical consequence for students of the dispute is a reordering of study. Polemical summaries usually begin with conclusions: one side abandoned the labour theory of value, the other kept it; one side embraced rational choice, the other denounced it. Beginning with conclusions makes the quarrel look like a shopping list of agreements and disagreements, and invites the lazy question of which list is longer. Beginning with method makes the quarrel intelligible as a single branching. The lists then explain themselves. A reader who sees why a denial of proprietary method leads, step by step, to game-theoretic models of class and to property-relations definitions of exploitation has learned something transferable to fresh cases, such as the reception of later analytical work on equality, where the same temper produced egalitarian conclusions by arguments any liberal philosopher could audit. A reader who sees why an affirmation of dialectical method leads to suspicion of those same models has likewise learned the grammar of a tradition, and can predict its response to proposals not yet made.

There is also a historical reason the method question outranks the doctrinal ones. The analytical school formed in reaction to a specific intellectual environment, the Marxism of the postwar decades in its Althusserian and party variants, in which, as its members complained, obscurity functioned as a credential and quotation as an argument. Cohen recalled seminars in which a claim’s resistance to paraphrase counted in its favour. The school’s founding documents are unintelligible except as protests against that environment. The classical defence likewise formed against an environment: the reduction of Marxism, in the official doctrine of the socialist states, to a catechism of laws, and the reduction of Marx, in hostile commentary, to a failed price theorist. Lukács wrote against the first reduction; later classical writers wrote against the second. Each camp’s method talk encodes a memory of what the other method, in power or in fashion, had done to the theory they shared. Polemic that ignores those memories mistakes scar tissue for mere stubbornness.

How did the two tempers arise from the same inheritance?

Both currents descend from Marx’s own writing, which contains systematic science and dialectical presentation side by side, often within a single chapter, and each current canonised one strand of the inheritance.

The analytical writers elevated the Marx who insisted that his conclusions rested on evidence any investigator could check, while the classical writers elevated the Marx who mocked political economy for treating historically specific forms as eternal categories.

Marx’s texts genuinely support both appropriations, which is why appeals to the master’s voice settle so little. The preface to Capital presents a scientific work, arranged by the logic of its object, with empirical material marshalled in its support. The same volume describes forms, value, money, capital, that exist only within a determinate mode of production and dissolve outside it, a historicity that sits awkwardly with timeless modelling. Engels, editing and popularising, tilted the inheritance toward system and law, most strongly in the dialectics of nature writings that later party doctrine canonised. The Western Marxist line, from Lukács through the Frankfurt School to the Italian tradition of Della Volpe and Colletti, split internally over exactly this tilt: Colletti, a classical dialectician by formation, attacked the dialectics of nature as a metaphysical accretion while defending contradiction in the social object, a combination that confuses any simple map of the camps. The analytical school inherited Colletti’s suspicion of Hegelian nature philosophy through a different route, arriving at a similar destination from the direction of Oxford and Oslo rather than Rome. Genealogy therefore cuts across allegiance, and the grid that follows should be read as a map of arguments, not of bloodlines.

The eight differences grid and what it orders

The grid below compresses the comparison into a findable form. Each row names a difference, states the analytical position in one phrase and the classical position in another, identifies the source from which the disagreement follows, and records a provisional judgement where evidence bears on the matter. Several rows resist any evidence-based judgement, because the disagreement is constitutive rather than empirical, and the grid says so instead of forcing a verdict. The prose sections after the table then take each row in turn, at the length its importance deserves, since a grid that replaced argument would commit the very sin of compression that the classical side warns against.

Difference Analytical position Classical position What the disagreement follows from Which side the evidence favours
Method Formal modelling and plain statement Dialectical analysis of a totality Whether a distinctive Marxist method exists Undecidable in general; clarity gains documented case by case
Labour theory of value Largely abandoned as a price theory Retained as the theory of the value form Whether doctrines are assessed one by one Analytical, on prices; classical, on what the theory was for
Exploitation Distributional, defined by asset inequality Relational, defined by domination in production Whether class and exploitation need production relations Divided; the distributional account travels further
Class Defined by assets or market positions Defined by relations of domination and struggle Whether categories are isolable without distortion Divided; each definition captures what the other misses
Explanation Mechanisms and individuals Structures and functions Whether functional claims need microfoundations Analytical, on the logic of explanation
Normative theory Embraced as explicit philosophy Rejected as ideology or moralism Whether critique requires stated values Analytical, on candour; classical, on suspicion of abstraction
Motor of history Development of the productive forces Class struggle Whether history has a directional mechanism Neither decisively; the record underdetermines both
Historical agent No privileged agent The working class as universal agent Whether emancipation needs a designated bearer Classical premises predicted the dispute; events favoured neither

Reading the grid vertically teaches the article’s central lesson. The fourth column repeats itself with variations because the rows share a root. Method generates the treatment of value: a school that assesses doctrines separately will test the labour theory of value as a theory of relative prices and report the result, while a school that treats doctrines as moments of a whole will ask what work the value form does inside the analysis of capital and refuse the separate test as a misunderstanding. Exploitation and class then follow from the value decision and from the method directly. Explanation follows from whether wholes may be invoked without decomposition. Normative theory follows from whether a science of society may also be a philosophy of justice. The motor of history and the identity of its agent follow from how much directional structure the method permits. A student who memorises the grid without the root column has a revision aid; a student who grasps the root column can regenerate the grid from a single premise, which is the difference between revision and understanding.

Difference one: method, modelling against dialectics

The first difference is the generator itself, stated as a row for completeness. Analytical method, as practised by the school’s exemplars, consists of definition before argument, distinction between claims that tradition fused, formal models where interaction matters, and a standing demand that any explanation identify the mechanism through which an effect is produced. Cohen’s defence of historical materialism proceeded by distinguishing theses and testing each. Roemer’s models proceeded by constructing small economies in which definitions could be checked for their consequences. Wright’s class analysis proceeded by building typologies whose cells could be filled with survey respondents. In each case the method promises auditability: a critic can locate the premise doing the work and contest exactly that premise. The classical method promises fidelity to a moving object: categories that develop, forms that invert, a whole that explains its parts while being reproduced by them. Its exemplars, from Lukács on essence and appearance to Adorno on negative dialectics, treat the demand for static definition as a falsification introduced at the start, since a category frozen for inspection has already ceased to be the category at work in history.

What does formal modelling add that dialectical exposition cannot supply?

Modelling forces a writer to specify who acts, with what resources, under which constraints, and what follows when those elements interact, so hidden assumptions surface as explicit premises that opponents can target.

Dialectical exposition can gesture at interaction indefinitely without committing to any particular account of it, whereas a model that fails fails informatively, showing exactly which premise the world refused.

The strongest analytical statement of the methodological creed came from Elster, whose demand for mechanisms, for nuts and bolts beneath functional patterns, reorganised how a generation of social scientists read Marx. Elster’s point was logical before it was political. To say that an institution exists because it benefits capital explains nothing until a feedback loop is supplied through which the benefit sustains the institution, and supplying that loop requires descending to actors, incentives and information. Cohen, who had defended functional explanation as legitimate in form, conceded much of the demand in practice by elaborating the mechanisms his functional claims presupposed. The exchange between them, conducted in print across the nineteen eighties, remains the cleanest exhibition of analytical method improving a Marxist argument from inside, since both parties accepted the same tribunal of clarity and differed only on what clarity required.

The classical rejoinder deserves equal precision. Its best form is not a defence of obscurity, which no serious classical writer offers, but a claim about the cost of decomposition. Some objects, the argument runs, have properties that exist only at the level of the system: liquidity in a market, confidence in a currency, the value form itself. Decomposing such objects into individual transactions does not reveal their secret; it dissolves the phenomenon and studies the residue. The classical writer points to Marx’s own procedure in Capital, where the analysis begins with the commodity as a cell form and unfolds its contradictions, as evidence that the theory’s founder chose dialectical presentation deliberately, as the only form adequate to an object that develops by internal opposition. Whether that defence succeeds is taken up in the section on the modelling objection, where it receives the longest hearing this article gives any argument. Here it is enough to register that the first row of the grid is not a preference for tidy prose against messy prose. It is a disagreement about what social reality is like and therefore about what knowing it requires.

Difference two: the labour theory of value, abandoned against retained

Few rows of the grid carry more symbolic weight. For the classical tradition, the labour theory of value anchors the analysis of exploitation: value is congealed labour time, surplus value is unpaid labour, and profit is a transformed appearance of a transfer that occurs in production. Abandoning the theory therefore looks, from that side, like abandoning the explanation of profit and retaining only a complaint about distribution. The analytical school examined the theory as a set of propositions and reported, with regret in Cohen’s case and without much in Elster’s, that the propositions failed. Roemer’s work delivered the sharpest internal blow: models in which exploitation, defined by unequal exchange of labour, emerges in economies where the labour theory’s price predictions do not hold, and in which agents can be exploited in the relevant sense through assets other than labour power. If exploitation can be defined and detected without the labour theory, the theory loses its claim to be the indispensable core, however elegant its architecture.

The classical retention of the theory rests on a different reading of what the theory asserts. Writers in the value-form tradition, descending from Isaak Rubin in the nineteen twenties and renewed by later scholars of Marx’s monetary theory, argue that the labour theory is not chiefly a device for predicting prices. It is an account of why social labour in a commodity society takes the form of value at all, why private labours are validated as social only through exchange, and why wealth appears as an accumulation of things. On that reading, the transformation problem that exercised generations of critics addresses a question the theory was not primarily asking, and the analytical verdict, however correct about prices, passes beside the theory’s real object. The analytical reply is characteristic: if the theory is about the form of value rather than about magnitudes, then its defenders should state that claim plainly, surrender the magnitude claims that generations of textbooks taught, and accept that the retained theory is a different and smaller theory than the one the tradition celebrated. Much of the heat in this row comes from both sides suspecting, often correctly, that the other is changing the subject while claiming to defend the same doctrine.

Why did the value dispute become a test of loyalty?

Because the labour theory stood at the entrance of the system, taught first to every recruit and printed in every primer, so surrendering it felt like surrendering the theory’s identity rather than revising one component.

Analytical writers treated that feeling as evidence of dogma, while classical writers treated the analytical readiness to surrender it as evidence that the school had never grasped what the theory held together.

For the comparison’s purposes, the row teaches a methodological lesson that recurs below. The two sides were frequently answering different questions while using identical words, and the analytical insistence on distinguishing questions, however pedantic it appeared, repeatedly exposed the equivocation. Price theory, form analysis and exploitation accounting are three enterprises, and a verdict on one does not settle the others. The classical side’s counter-lesson is equally durable: distinguishing questions can become a way of never facing the largest one, which is why the parts of Marx that impressed the classical tradition, the account of a society whose members are governed by abstractions they produce, survive disaggregation badly. The grid records the row as divided, analytical on prices, classical on purpose, and the division is not a fudge. It is the accurate shape of a dispute in which each side won the argument it had chosen to have.

Difference three: exploitation, distributional against relational

Roemer’s reconstruction defines the third row. In A Theory of Exploitation and Class, published in 1982, Roemer proposed that a group is exploited when it would do better by withdrawing from the economy with its per capita share of society’s assets. The definition is distributional: exploitation is diagnosed from the inequality of asset ownership, not from the extraction of surplus in production, and it deliberately generalises beyond labour. A person exploited through credit markets or through scarce skills falls inside the concept, and Roemer constructed the class exploitation correspondence principle to show when asset positions map onto class positions in production. The elegance of the construction is analytical method at its best: a contested moral notion rebuilt as a testable counterfactual, with corollaries that can be checked. Its accompanying provocation was equally deliberate. If exploitation rests on unjust asset distribution, then the moral work is done by a theory of justice in holdings, and Marxism needs the normative philosophy it had long disdained, a conclusion developed in the article on Roemer’s theory of exploitation and resisted fiercely by classical writers.

The classical definition is relational. Exploitation names something that happens between people inside production: one class, owning the means of production, compels another to labour beyond the time required to reproduce its own subsistence, and appropriates the surplus. Domination is not an implication of the concept but its content. Ellen Meiksins Wood and other classical critics objected that Roemer’s definition, by locating exploitation in distribution, dissolves the specifically capitalist relation into a general complaint about inequality, available to any liberal egalitarian, and loses the analysis of how the surplus is produced, contested and hidden by the wage form. A feudal lord and a shareholder become instances of one pattern, which the classical side counts as a loss of historical specificity, the very virtue the tradition claimed against abstract moralising. The analytical reply concedes the generality and claims it as a gain: a concept that travels can be tested across cases, and a Marxism that cannot speak of exploitation outside the factory gate has narrowed its critical reach to fit its favourite example.

What is lost when exploitation is defined by assets alone?

The scene of production disappears, and with it the daily experience of command, pace and obedience that made exploitation a lived relation rather than an accounting outcome.

Workers who own small assets through pension funds become awkward cases, managers with no assets become invisible, and the classical insistence that something happens at the point of production, irreducible to the distribution that precedes it, goes unanswered by construction.

Both definitions illuminate cases the other obscures, and the honest comparison says so. The distributional definition handles the skilled worker whose scarce credentials extract rents, the small shareholder who labours under a manager poorer in assets than herself, and the cooperative member exploited by a market the cooperative does not control. The relational definition handles the foreman’s authority, the lengthening of the working day, and the difference between being badly paid and being commanded, a difference that matters to those who live it and to any politics organised around it. The analytical school eventually absorbed part of the objection: Wright’s later work reintroduced domination and exploitation as distinct dimensions, and Roemer himself moved toward a property relations account supplemented by an explicit theory of justice. The classical side absorbed less, publicly, though value-form writers quietly conceded that asset distributions do explanatory work their categories had neglected. The row stands in the grid as divided, with the distributional account travelling further across cases and the relational account retaining the phenomenon that made the concept worth having.

Difference four: class, assets against relations

Class is where the methodological fork becomes sociological. Wright’s project, the most empirically ambitious the analytical school produced, defined class positions by assets: ownership of means of production, of organisational resources, and of skills or credentials. The typology of contradictory class locations, managers and supervisors and experts lodged between the great classes, was built to be operationalised, and Wright filled it with survey data gathered across several countries, testing hypotheses about consciousness and income that the classical tradition had asserted without measurement. The gains were real. Claims about the middle class that had circulated as rhetoric became estimates with margins of error, and the schema’s failures, cells that behaved unexpectedly, generated research rather than apologetics. A fuller account of the underlying dispute belongs to the guide to the Marxist theory of class, which this article links for readers who want the sociological detail.

The classical definition refuses the asset inventory as a starting point. Class, on that account, is not a slot in a distribution but a relation lived in struggle: classes constitute themselves against each other, in workplaces and communities and political organisations, and a snapshot of assets misses the process by which a collection of similarly placed people becomes, or fails to become, a class in the active sense. E. P. Thompson’s preface to The Making of the English Working Class gave the position its canonical form, insisting that class is a historical phenomenon, an happening in human relationships, not a structure to be photographed. From this side, Wright’s cells commit a category error with good data: they measure stratification and call it class, while the phenomenon Marxists cared about, collective capacity formed in conflict, evaporates between the survey waves. Analytical writers replied that a process invisible to measurement is conveniently immune to refutation, and that Thompson’s own narrative relies on occupational and material facts his theory cannot disown.

Can a survey measure a relation defined by struggle?

Only partially, and the partiality is instructive, since surveys register positions, resources and reported attitudes with precision while the relational element, solidarity under pressure, organisation in motion, shared memory of defeats, appears in the data as residue or not at all.

Wright’s instruments captured where people stand; Thompson’s histories captured what standing there did to them across time, and neither instrument substitutes for the other.

The mature positions converged more than the polemics admit. Wright’s later work, especially Envisioning Real Utopias, treats class compromise, organisation and institutional design as processes, not merely positions, and draws on case histories in a Thompsonian spirit. Classical historians, for their part, use occupational data constantly, whatever their prefaces say, because narrative without distribution floats free of the material it claims to explain. What did not converge is the starting definition, and the starting definition decides what counts as a surprising result. For the asset theorist, a quiescent working class with rising consumption is a puzzle about consciousness. For the relational theorist, quiescence is not an anomaly within class but evidence about the state of a relation, to be explained by defeats, institutions and memories. The grid marks the row divided for the same reason as the exploitation row: each definition was built to catch what the other lets through, and the choice between them is finally a choice about which blindness a researcher prefers, which returns the decision, once again, to the generative disagreement about what kind of knowledge social theory is.

Difference five: explanation, mechanisms against structures

The fifth row concerns how anything gets explained at all. Analytical writers, following Elster’s critique of functional explanation in Marxism, hold that a social pattern is explained when the mechanisms sustaining it are specified: the actors, their beliefs and opportunities, and the aggregation by which individual actions compose the pattern. The demand produced the microfoundations debate, treated at length in the companion article on Marxism and microfoundations, and it aligned the analytical school with methodological individualism, the doctrine that social explanations should terminate in facts about individuals, suitably understood. Cohen’s defence of functional explanation, that a trait can be explained by its beneficial effects for a system when an elaboration connects effect to persistence, was the school’s internal dissent, and the Cohen and Elster exchange taught both men to qualify: Cohen elaborated mechanisms, Elster admitted that some functional claims could be rehabilitated once their feedback loops were named.

Classical explanation moves in the opposite direction. Structures, modes of production, class relations, ideological forms, are held to possess causal powers not reducible to the individuals who instantiate them, and the characteristic classical explanation shows how a structure selects, constrains and forms the agents who then reproduce it. Althusser’s structural causality, whatever its obscurities, named a genuine problem for individualism: preferences and beliefs, which individualist explanations take as inputs, are themselves products of the structures under study, so an explanation that starts from given individuals starts one step too late. The analytical tradition eventually took the point seriously as a problem about endogenous preferences, modelling how institutions shape the agents who inhabit them, though the modelling preserved the individualist form by adding a prior stage. Whether that addition answers the objection or restates it in equations is one of the live seams of the comparison, and it connects directly to the modelling objection examined below, where the givenness of preferences becomes the central charge.

Why do functional explanations need mechanisms beneath them?

Because a benefit cannot cause the existence of the thing that confers it unless some process links the benefit back to the thing’s survival, and without that link the explanation mistakes a coincidence or a consequence for a cause.

Elster’s demand simply applies, to Marxist claims about ideology and the state, the standard that evolutionary biology meets through selection and that Marxist writers themselves applied to their opponents.

The verdict on this row is the grid’s clearest. On the logic of explanation, the analytical side won the argument inside the profession: textbooks in the philosophy of social science teach the mechanism demand as settled, and classical writers who continue to offer functional claims without elaboration are read, even by sympathisers, as trading on credit the tradition no longer holds. Winning the logical point, however, settled less than the victors hoped. A mechanism requirement disciplines explanation without supplying explanations, and the stock of confirmed mechanism-based accounts of large historical change remains thin on both sides. The classical tradition’s holistic narratives, for all their logical softness, continue to organise research into long-run transformations that mechanism talk has yet to reproduce at comparable scope. The grid therefore credits the analytical side on the logic while the article records, as a standing caution, that a correct theory of explanation is not itself an explanation of anything.

Difference six: normative theory, embraced against rejected

Marx’s own attitude toward moral philosophy was contemptuous in print and saturated with moral judgement in practice, a tension the tradition inherited unresolved. The classical resolution treated normative theory as ideology: justice, on the reading associated with Allen Wood, is a juridical notion internal to a mode of production, so capitalism is not unjust by its own standard, and condemning it in the name of justice mistakes the terrain of critique. Steven Lukes mapped the resulting paradox, a Marxism that denounced exploitation while disowning the moral vocabulary needed to say why exploitation is wrong. Classical writers added a strategic suspicion: moralising substitutes exhortation for analysis, and appeals to fairness invite the reply that fairness, properly understood, supports the market order, as Robert Nozick’s entitlement theory argued from the other side.

The analytical school broke the taboo deliberately. Cohen’s later work, from Self-Ownership, Freedom and Equality to his lectures on rescuing justice, argued that Marxists needed an explicit normative theory and that the tradition’s unacknowledged egalitarianism should be brought into the open and defended as philosophy. Norman Geras had earlier shown, against Wood, that Marx did condemn capitalism as unjust by standards Marx’s texts supply, whatever Marx’s official doctrine said. Once the taboo fell, an entire research programme opened: analytical Marxists wrote on equality, freedom, self-ownership and exploitation as moral notions, engaging Rawls and Nozick on shared ground. The classical objection, that this engagement dissolved Marxism into liberal moral philosophy with historical footnotes, is the seed of the retreat from class argument treated in a later section, and it is not trivial: a theory whose distinctive claims are historical and structural may indeed have least to say precisely where its adherents began to publish most.

Does condemning capitalism require a theory of justice?

Cohen’s mature answer was yes in practice, since every condemnation smuggles in a standard, and honesty requires stating the standard and defending it against rivals rather than pretending the facts condemn themselves.

The classical answer remains that the standard is supplied by the movement of history and the interests of a class, so importing a free-standing theory of justice adds a tribunal above history that the materialist conception was built to refuse.

On candour, the grid credits the analytical side. A critique that relies on values it refuses to state is vulnerable to exactly the exposure Geras performed, and the tradition’s evasions on this point had become, by the nineteen seventies, a scholarly embarrassment that sympathetic readers admitted in private. The classical suspicion nevertheless retains force as a warning about abstraction. Normative theory conducted at the level of principles and ideal distributions can drift free of the institutional and historical analysis that made Marxist critique distinctive, ending in a ranking of possible worlds with no account of the road between them. Cohen himself felt the pull of that warning, which is why his egalitarianism kept returning to questions of ethos, feasibility and the structure of incentives. The row is scored for the analytical side on candour and for the classical side on the danger it named, a pattern the reader will recognise from earlier rows: the school’s corrections were usually genuine, and the tradition’s warnings about their cost were usually genuine too, and the comparison fails its subject whenever it records only one of those truths.

Difference seven: the motor of history, forces against struggle

Historical materialism, in Cohen’s reconstruction, assigns the driving role to the productive forces: technology and labour power develop through history, production relations rise and fall according to whether they foster or fetter that development, and class struggle is the mechanism by which transitions occur rather than the engine itself. The reconstruction gave the theory a directional logic that could be stated, criticised and defended, and Cohen defended it with formidable care. Its analytical character lies in the separation of roles: an engine, a transmission and a criterion of progress, each examinable apart from the others. Critics inside the school, Elster prominently, doubted whether the development thesis had the evidence or the mechanism its role required, and the doubt illustrates the school’s virtue from within, since the objection was pressed hardest by a member against the founding text of another member.

The classical tradition reverses the priority. The opening of the Communist Manifesto declares the history of all hitherto existing society to be the history of class struggles, and classical writers from Lukács to Thompson read development itself as an outcome of struggles over its direction, pace and beneficiaries. On this account, forces do not develop in the abstract: enclosure, resistance, sabotage, legislation and war decide which techniques spread and which rot, so treating struggle as a transmission belt for an independent technical logic reverses cause and effect. Robert Brenner’s intervention in the transition debate gave the classical priority empirical teeth, arguing that similar commercial and demographic pressures produced divergent outcomes in different parts of Europe because class relations and their political outcomes differed. The Brenner debate became a rare arena in which the two tempers met over evidence rather than manifestos, and historians who cared for neither label found themselves using its terms.

What would settle the contest between forces and struggle?

A run of transitions in which technical development predicted outcomes regardless of the balance of class forces would vindicate the primacy thesis, while repeated cases in which similar techniques issued in opposed relations, traceable to struggle, would vindicate the classical priority.

The historical record supplies instances of both patterns, which is why the grid scores the row as underdetermined and why honest partisans on each side quietly borrow the other’s variable when writing narrative.

The deeper point for the comparison is that the row is generated, not independent. A method comfortable with separable mechanisms will distinguish engine from transmission and ask which does the work. A method committed to totality will resist the separation, insisting that forces exist only as wielded, resisted and organised by classes, so the question of priority is malformed. Readers can watch the generative disagreement operating here in miniature: the same archive, two questions, and no experiment that both sides accept as decisive, because the acceptance criteria are themselves downstream of the fork. That is not a counsel of despair about historical knowledge. It is a reason to read the best work on both sides, Cohen’s reconstructions and Brenner’s cases, as complementary instruments, each calibrated to detect what the other is built to miss.

Difference eight: the agent, none against the working class

The final row is the most consequential politically. Classical Marxism designates the working class as the universal agent of emancipation: the class whose particular interest, abolishing itself as a class, coincides with the general interest in ending class society. Lukács built an entire epistemology on the designation, arguing that the proletariat occupies the standpoint from which society becomes intelligible as a totality. The designation gave the tradition its political architecture, parties claiming to embody the class, a theory of consciousness and its lags, and a standard against which betrayals could be named. It also gave the tradition its characteristic vulnerability, since every decade in which actual workers declined the assigned role required either a theory of false consciousness or a revision of the assignment.

The analytical school, following its method, asked what could justify designating any agent in advance. If claims must be argued rather than inherited, then the universality of the working class is a hypothesis about interests and capacities, to be tested like others. The tests, as analytical writers read the evidence, went badly: affluence, fragmentation of occupations, cross-cutting identities and the demonstrated capacity of capitalist democracies to absorb discontent all told against the strong classical claim. Adam Przeworski modelled the electoral dilemmas of socialist parties with characteristic bleakness, showing how the pursuit of majorities dilutes the class appeal that mobilisation requires. One by one, analytical writers withdrew the designation, some to a broader coalition of the disadvantaged, some to no agent at all, leaving a normative theory in search of a bearer. The trajectory from designated agent to none is the empirical spine of the retreat from class argument, and classical critics were quick to note that the withdrawal was not an accidental drift but the working out of premises: a method that begins from individuals and their given interests has no place to put a class whose consciousness is supposed to be formed by its position in a totality.

Is a theory of emancipation incomplete without a designated agent?

Classical writers answer that a critique with no bearer is a sermon, since principles do not enact themselves and history supplies no other candidate with both the interest and the numbers to transform production relations.

Analytical writers answer that designating a bearer by theory, against the evidence of the bearer’s own choices, substitutes wish for analysis and has licensed parties to speak in the name of workers who never authorised them.

The grid scores this row with care. Events favoured neither side’s strong claim: the working class of the advanced economies did not make the classical revolution, and no alternative agent performed the analytical school’s hopes either, so the political record is a scoreless draw between two disappointed expectations. What the classical side can claim, and the article grants it, is foresight about the internal dynamic of its rival. The prediction that analytical premises would dissolve the agent was made early, by critics who understood the method better than some of its practitioners, and the prediction came true inside the school’s own publications. Foresight about a rival’s trajectory is not proof of one’s own doctrine, a distinction the verdict section will insist on, but it is evidence that the classical critics had hold of something real about how methods shape conclusions over time. With the eighth row, the derivation is complete: eight differences, one root, and a standing invitation to test the claimed derivation by finding a difference between the currents that the root cannot explain. None of the standard candidates survives that test for long, since disputes over the state, over imperialism and over the party all decompose, on inspection, into the same fork about method, evidence and the whole.

The modelling assumptions objection at full strength

Every comparison has a centre of gravity, and for this one it is the classical objection to analytical modelling. The objection is frequently misstated by both sides, by critics who inflate it into a rejection of rigour and by defenders who deflate it into a complaint about mathematics. Stated at full strength, it is neither. It grants, for the sake of argument, that the analytical school’s particular results may be internally valid, that its models may be consistent, its definitions clean, its inferences sound. It then denies that validity settles anything, because the conclusions were installed at the entrance, inside the assumptions a model must make before its machinery can turn. An objection of that form cannot be answered by improving the machinery. It can be answered only by showing that the assumptions are innocent, or dispensable, or shared by any possible inquiry, and the classical claim examined in this section is that no such answer was produced.

The assumptions in question are two, and they interlock. Analytical models of social interaction, in the game-theoretic and rational choice forms the school favoured, represent persons as agents endowed with resources and preferences, choosing among options under constraints. The first assumption is that the wage relation can be represented as a transaction such an agent enters, an exchange of labour power for a wage between parties each free to decline. The second is that the agent’s preferences are given prior to the interaction, formed elsewhere and imported into the model as data. Both assumptions are, for the modellers, conveniences of representation, the price of tractability. For the classical critic, both assumptions are the disputed theory itself, restated in the grammar of choice. Marx’s analysis contests precisely the claim that the wage bargain is a voluntary exchange between equals in the relevant sense, and contests precisely the claim that wants and identities precede the relations that form them. A framework that assumes the contested points does not refute the contesting theory. It declines to represent it, and then reports, accurately, that the theory’s conclusions do not arise inside the framework.

Ellen Meiksins Wood gave the objection one of its principal statements in her critique of rational choice Marxism, asking of the whole enterprise whether the game was worth the candle, a phrase this article borrows because it captures her accounting: the costs of the framework are paid in the currency of the theory’s distinctive claims, and the benefits arrive in a currency any liberal social science already holds. Wood’s argument, developed across her essays on the retreat from class, was that rational choice modelling does not merely borrow a technique. It imports an ontology, a picture of society as an aggregate of choosing individuals whose relations are exchanges, and that picture is the methodological individualism of the liberal tradition Marx set out to criticise. To adopt the picture as a neutral instrument is to decide, before the evidence is heard, that the Marxist account of social form is false, since that account holds that exchange relations are themselves the surface of production relations that choice talk cannot reach.

What exactly does the modelling objection accuse the school of doing?

It accuses the school of begging the question at the level of framework rather than argument: of choosing representations in which the wage relation appears as free exchange and preferences appear as prior facts, then presenting the non-appearance of Marxist conclusions inside those representations as a discovery about Marxism, when the representations guaranteed the result before any model was run or any evidence consulted.

The objection gains force from the school’s own trajectory, which is why classical writers return to it with such confidence. Begin with agents, endowments and choices, and a series of results follows with an air of inevitability. Exploitation becomes a fact about asset distribution, because distribution is what the framework can see. Class becomes a classification of asset holders, because positions in a choice structure are what the framework can sort. History becomes a sequence of equilibria and their disturbances, because change must enter as a shock to parameters the framework holds fixed. Agency dissolves, because a class in the classical sense, a collective subject formed by a totality, has no representation among choosing individuals. Each step was defended on its merits by able writers, and the classical critic does not need to deny any step’s local merits. The critic points at the sequence and asks whether a method whose every application thins the theory in the same direction is discovering the theory’s errors or expressing its own constitution. That question, unlike the local disputes, was never taken up by the school as a school, and the silence is the objection’s best evidence.

Consider the wage relation more closely, since it is the objection’s home ground. In a standard model, a worker with labour power and a capitalist with money meet, and each accepts the bargain because each prefers it to the alternative of no bargain. The model registers consent, and from consent a family of conclusions flows about the moral and explanatory character of the relation. The classical analysis asks what the model has cropped out of the frame. Behind the worker stands a history: the separation of producers from means of production, accomplished by enclosure, law and violence across generations, which left labour power as the one commodity the worker owns. Behind the capitalist stands a structure: competition among capitals that punishes any employer who pays above the going rate or shortens the working day unilaterally. Inside the workplace, after the bargain is struck, stands a relation of command that the contract does not describe and the model does not revisit. The classical claim is not that consent is absent from the scene. It is that consent, framed as a choice between the bargain and destitution, is the form the compulsion takes in a market society, and that a representation beginning after the separation, with endowments distributed and preferences formed, has assumed away the very processes the theory of capitalism exists to explain.

The preferences assumption cuts as deep. Models require agents who want things, and the wants must be specified before the model runs. Yet a central Marxist claim, from the analysis of commodity fetishism onward, is that capitalist relations form the wants of those inside them: that needs are produced alongside the commodities that satisfy them, that the desire for status goods, for security, for accumulation itself, are social products with histories. A framework that treats preferences as given cannot represent that claim, not because the claim is false but because the framework’s grammar lacks a place to put it. Later analytical work on endogenous preferences, in which institutions shape desires across periods, attempted to build the missing room, and the attempt is instructive: the models that resulted were admired as technique and quietly set aside as Marxism, since a theory in which class position shapes preference only through individual updating rules had reproduced, in more elaborate notation, the individualist premise under challenge. The classical critic’s point is not that the elaboration was worthless. It is that the elaboration conceded the diagnosis while refusing the remedy, which would have required starting from relations rather than from choosers.

The answers attempted, and why they fall short

Fairness requires stating the answers the analytical side did give, in their strongest forms, before judging them insufficient. The first answer is the innocence of instruments. A model, on this view, is a lens, not a metaphysics: economists model firms as profit maximisers without believing that managers dream in calculus, and Marxists may model agents as choosers without believing that society is nothing but choices. Roemer made a version of this answer when he presented his models as explorations of concepts, asking what exploitation must mean if it is to do moral and explanatory work, rather than as portraits of any actual economy. The second answer is the fertility of the framework. Whatever its assumptions, modelling generated results the tradition had not reached by other means: precise conditions for class formation, counterexamples to cherished claims, a reconstruction of exploitation that survived the collapse of the value theory in its classical form. A framework that discovers things its rivals missed has earned its assumptions a hearing. The third answer is tu quoque: classical analysis also assumes, also selects, also begins somewhere, and its beginnings, totality, contradiction, the standpoint of a class, are assumptions with far less scrutiny applied to them. Every inquiry begs some questions; the analytical school at least publishes its begged questions as premises where critics can shoot at them.

Which analytical contributions escape the objection entirely?

The contributions stated as pure conceptual or logical results escape it, since a proof that functional explanation requires a feedback mechanism, or that a definition of exploitation has a given corollary, depends on no representation of the wage relation and no assumption about where preferences come from.

Elster’s mechanism demand and Cohen’s distinctions survive the objection untouched, which is why the objection targets the modelling programme rather than the school’s whole output.

That answer, the strongest available, also marks the objection’s boundary, and the article marks it plainly. The school’s logical and conceptual work, its clarifications of historical materialism, its exposure of lazy functionalism, its insistence that normative commitments be owned, none of that depends on representing exchange between endowed agents, and the classical objection does not touch it. What depends on the contested assumptions is the school’s constructive social theory: the models of exploitation, class and transition in which agents, endowments and given preferences do the representing. There, the objection stands in its full form. Roemer’s exploitation framework assumes a distribution of assets as its starting point and treats the transactions that follow as exchanges among owners, which is the contested grammar in its purest state. Wright’s class typology begins from asset positions held by individuals, and its celebrated empirical programme inherits the starting point. The transition models begin from agents reacting to parameters. In each case the classical question recurs: show that the conclusion survives when the starting point is surrendered, or admit that the starting point is a premise of the theory rather than a finding of it.

No such showing was produced. Analytical writers answered objections to particular models with better models, which is the normal and honourable rhythm of a research programme, but the framework objection is not an objection to a particular model and cannot be met by one. Meeting it would have required a model of a capitalist economy in which the wage relation is not entered as a free bargain struck by endowed agents whose wants are taken as settled, and in which Marxist conclusions, exploitation as a relation of domination, class as a product of production relations, nevertheless emerge. The absence of that model is the fact on which the verdict of this article rests, and the defeater attached to the verdict, stated formally in the closing section, simply describes the missing object: produce the model, and the judgement that the objection was never answered falls. Until then, the classical side holds the strongest single argument in the comparison, an argument about method that its opponents’ method prevented them from hearing as anything but an excuse for vagueness. The tragedy of the exchange, if the word is allowed, is that each side’s deepest insight was the other’s blind spot: the analytical school saw that unclarity protects error, the classical tradition saw that clarity is purchased with assumptions, and a synthesis that held both insights at once was sketched by individuals on both sides without ever becoming the property of either camp.

What the analytical current got right

The ledger of analytical achievements begins with a rescue. Before the school, large parts of the Marxist tradition were defended by quotation and policed by excommunication, and a student who asked what exactly the primacy of the productive forces asserted, or how a superstructure could be both determined and effective, received a bibliography instead of an answer. The analytical writers supplied answers, and in supplying them they recovered Marx as a thinker whose claims could be wrong, which is the only form in which a claim can also be right. Cohen’s reconstruction of historical materialism remains the standard against which alternatives are measured, not because its theses all survived, but because it stated them so that survival and failure could be told apart. That service to the tradition is independent of the school’s later trajectory, and classical writers who use Cohen’s distinctions, as many do silently, testify to it whenever they write.

Second, the school was right about functional explanation. The tradition had explained institutions by their benefits to capital for a century, in a grammar that confused consequence with cause, and Elster’s demand for mechanisms, however abrasively delivered, was a permanent correction. After the correction, a Marxist who explains an ideology by its service to the ruling class owes the feedback loop, the process by which that service secures the ideology’s reproduction, and the debt has improved even the work of writers who resent the creditor. Third, the school was right that formalisation makes claims testable in a way that protects the honest and exposes the evasive. Roemer’s counterexamples to the labour theory of value were not acts of vandalism; they were the kind of result a theory should want from its friends, delivered early and in public, rather than from its enemies at a moment of crisis. Fourth, the school was right that the tradition needed a normative theory and was damaged by pretending otherwise. The exposure of Marxism’s unowned egalitarianism, and the construction of explicit arguments about equality and freedom, gave critics of capitalism a vocabulary that survives contact with trained philosophers, a gain no amount of classical suspicion about abstraction should be allowed to erase.

Did clarity ever cost the analytical school a truth it valued?

Yes, and the school’s best members recorded the cost themselves, which distinguishes honest loss from careless loss.

Cohen’s later writing on equality circles a good, the value of community and of relations unmediated by market calculation, that his own analytical instruments could state only awkwardly, and his acknowledgement of the strain is among the most instructive passages in the school’s literature.

Fifth, the analytical current was right about the scholarly virtues as political virtues. A movement’s theory, they argued, is not strengthened by immunity from criticism, and the habits of the seminar, distinction, counterexample, concession where concession is due, are not bourgeois ornaments but the machinery by which any tradition corrects itself. The point sounds procedural and is in fact substantive, because the alternative machinery available to Marxist traditions in the twentieth century was the party line, whose corrections arrived as purges of persons rather than revisions of claims. When the school insisted that Marx be read as an author among authors, answerable to evidence and argument, it was defending the only relationship to a founder compatible with intellectual adulthood. Classical writers, many of whom suffered under the alternative machinery or watched it disgrace causes they shared, rarely defend that machinery outright; their objection is to the tribunal’s jurisdiction, not to the existence of error. The ledger therefore credits the analytical side with the tradition’s self-correction, a contribution that outlasts every particular thesis the school abandoned, since a tradition that can correct itself can recover from the loss of any thesis, including the ones the school lost on its behalf.

What the classical current got right

The classical ledger opens with the objection already examined: the modelling assumptions are not neutral, and a tradition that noticed this early, and said it in season and out, performed the service of a permanent auditor. Every generation of formal social science rediscovers that its instruments encode a worldview, and the classical Marxists were saying it, with a worked example, while the analytical school was still treating method as plumbing. The point generalises beyond Marxism, which is why the article returns to it when considering adjacent disciplines: a framework’s innocence is a claim requiring defence, not a default requiring none. Classical writers were right, secondly, about class. Removing production relations from the definition, whatever the gains in measurability, costs the theory its explanatory core, the account of why the distribution of assets reproduces itself and why those inside the relation experience it as command rather than as bad luck in a lottery. The relational definition keeps hold of the phenomenon that made class a revolutionary category rather than a statistical one, and the later analytical concessions toward domination and process amount to a partial repayment of the point.

Thirdly, the classical side was right about trajectory. The forecast that analytical premises would lead, by internal pressure, to the abandonment of the working class as agent and of Marxism’s distinctive claims one by one was made early and made precisely, by critics who are rarely credited with prophecy because their other predictions failed. Foreseeability matters to the comparison for a specific reason: it shows the classical critics understood the method they opposed, understood it as a system of pressures rather than a bag of tools, at a time when its practitioners described it as mere clarity. A critic who can predict a research programme’s path from its premises has demonstrated the very holism the programme denied, applied to the programme itself. Fourthly, the classical tradition was right that formalisation can distort its object, not always and not fatally, but in identifiable cases where the phenomenon is constituted by relations, meanings and histories that a model’s primitives cannot carry. The value form, class consciousness and ideology remain the standing examples, domains where the analytical literature is thinnest relative to the classical, and where even sympathetic surveys concede that something in the object resists the treatment.

Does formalisation distort every object it touches?

No, and the classical tradition’s serious writers never claimed it did, since their claim concerns specific objects whose properties are relational and historical rather than aggregative.

Prices, quantities and distributions formalise well, and classical economists of Marxist formation built models of reproduction schemes with enthusiasm; the distortion claim begins where the object is a social form whose existence depends on the meanings its participants attach to it.

Fifthly, the classical side was right about the inseparability of method and content in one limited but important sense: the choice of method is itself a theoretical act with consequences that no amount of subsequent care can fully quarantine. That is a lesson the analytical school’s history teaches better than any classical text, since the school set out to clarify a doctrine and ended by revising it beyond recognition, not through carelessness but through fidelity to its instruments. The two ledgers, analytical and classical, are deliberately balanced in length and in weight. A reader who finds one ledger more persuasive has learned something about the reader’s own prior answer to the generative question, since the weighing of the ledgers is itself method-dependent: count the analytical gains if clarity is your standard, count the classical gains if fidelity to the object is. The article’s verdict, when it arrives, therefore rests on the narrowest and firmest ground available, the single unanswered objection, rather than on a summation of ledgers that no neutral arithmetic exists to perform.

The retreat from class and the question of combination

The phrase retreat from class, which Wood used as the title of her study of the school’s political trajectory, names the pattern the classical ledger’s third entry predicts. Its stages can be traced through the school’s own bibliography. A theory of exploitation detached from production became a theory of justice. A theory of justice required principles that any reasonable person could accept, and principles of that kind do not designate a revolutionary class. A class analysis rebuilt on assets discovered a landscape of fragmented positions with no majority class to designate. A history driven by productive forces required no agent’s consciousness to turn its wheels. At each stage the move was argued, and at each stage the destination resembled, more closely, the liberal egalitarianism of the surrounding academy. Classical critics read the sequence as a reductio: premises that begin by dissolving the theory’s distinctive object end by dissolving its politics. Analytical defenders read the same sequence as integrity: conclusions followed arguments, and a school that had promised to abandon whatever failed examination could hardly stop at the conclusions its founders liked.

Was the school’s trajectory a refutation of its method?

Not by itself, since a method is not refuted by the unwelcome destinations of its honest users, and the trajectory is compatible with two readings: that the method exposed a doctrine’s weaknesses, or that the method was loaded against the doctrine from the start.

The trajectory becomes evidence only in combination with the modelling objection, which supplies the mechanism, the loaded assumptions, that the first reading lacks and the second requires.

Combination is the remaining constructive question, and the article treats it as a genuine possibility rather than a diplomatic formula. A combined practice would take from the analytical side its disciplines: state claims singly, name mechanisms, own normative commitments, accept counterexample as a gift. It would take from the classical side its audits: interrogate every framework for the conclusions smuggled into its primitives, keep production relations inside the definitions of class and exploitation, and remember that preferences and agents are formed by the structures under study. Individuals have practised the combination without naming it. Brenner’s historical work weds classical categories to evidence handled with analytical care. Wright’s late work rebuilds class analysis around processes and institutions while keeping the empirical conscience of his earlier surveys. Scholars of the value form have learned to state their claims about Marx’s monetary theory in forms a sceptic can test. None of these syntheses commands a school, and perhaps none can, since a school needs a banner and the combination’s banner would have to read: be clear, and suspect your clarity. As advice to individual readers, however, the combination is the most useful sentence in this article, and the persona sections below translate it into instructions for students, teachers and researchers who must handle the dispute in practice.

Cohen and the defence that defined a school

Any comparison of the two currents must spend time with G. A. Cohen, because Cohen is the figure both sides claim and neither can surrender. Trained in analytical philosophy at Oxford, raised in a communist household in Montreal, Cohen brought to Marxism the unusual combination of filial loyalty and professional ruthlessness. His 1978 defence of historical materialism did what the classical tradition had declared unnecessary and the hostile critics had declared impossible: it stated the theory as a set of theses, each with its evidence and its vulnerabilities exposed. The development thesis held that productive forces tend to grow across history. The primacy thesis held that production relations are explained by the level of those forces. Around these Cohen built a structure of distinctions, between forces and relations, between economic structure and superstructure, between a relation’s existing and its being explained, that allowed defenders and critics to locate their disagreements on a shared map for the first time.

The classical reception of Cohen’s book is a study in ambivalence. Perry Anderson, whose own work stood at the summit of the Western Marxist tradition, treated the book with a respect bordering on alarm, recognising a reconstruction that preserved the theory by transforming its presentation beyond recognition. Others objected that the transformation was the loss: a historical materialism restated as theses about technology and its effects had misplaced the class struggle that the classical tradition regarded as the theory’s soul, and Cohen’s famous concession that class struggle operates as the mechanism of transitions struck them as a demotion dressed as a clarification. Cohen’s own later trajectory deepened the puzzle. The philosopher who had defended functional explanation spent his maturity writing normative theory, attacking Nozick’s self-ownership thesis, defending equality against Rawls’s incentives argument, and confessing in lectures that his confidence in historical materialism’s central theses had ebbed. A classical reader sees the school’s whole story in that one career: clarity purchased, doctrine dissolved, politics transposed into moral philosophy. An analytical reader sees a mind following arguments across forty years without asking permission from its own earlier books, which is what the method promised at the start.

Why does Cohen embarrass both camps in the comparison?

Because his career supplies each side with its best exhibit and its worst problem at once: the analytical side must explain why its clearest product ended as egalitarian moral philosophy rather than Marxist social theory, while the classical side must explain why its tradition produced no reconstruction of comparable rigour and why its own writers teach from his distinctions.

The functional explanation controversy shows Cohen’s method at its most instructive. Elster charged that explaining production relations by their tendency to develop the productive forces was teleology unless a mechanism connected the tendency to the outcome, comparable to natural selection in biology. Cohen replied that functional explanations can be legitimate before their mechanisms are known, as Darwin’s were before genetics, provided the pattern they capture is robust and an elaboration is in principle available. The exchange is often summarised as a win for Elster, since Cohen spent later years supplying elaborations. A fairer summary is that the exchange disciplined both positions: functional explanation survived as a form, burdened with a duty to elaborate; the mechanism demand survived as a duty, softened by the admission that duties can be discharged later. Classical writers watching the exchange drew their own moral. Two analytical philosophers, sharing premises, tribunal and vocabulary, had spent a decade on the logic of one explanatory form without touching the historical substance the form was meant to serve, and the opportunity cost of rigour, measured in archives unopened and cases unstudied, is a cost the classical tradition exists to protest. The protest has force. It also has a limit, since the classical tradition’s own decades produced controversies, over structural causality and the relative autonomy of the state, whose ratio of archives to assertions was no better.

Elster and the demand for mechanisms

Jon Elster carried the analytical temper to its most uncompromising expression, and his Making Sense of Marx remains the school’s most exhilarating and exasperating book. Elster read the entire corpus as a court reads a dossier, claim by claim, keeping the theory of alienation in revised form, discarding the dialectics of nature, dismantling the labour theory of value, and praising, with real warmth, the Marx of the historical writings whose eye for mechanism and motive he regarded as unsurpassed. The book’s method was its message: no claim received credit for its author’s intentions, its place in a system, or its service to a movement. Each claim was asked what it asserted, whether the assertion was true, and what would show it false. Readers formed by the classical tradition experienced the procedure as philistinism raised to a principle. Readers formed by Elster experienced the classical alternative, in which a claim’s depth excused its vagueness, as the true philistinism, a refusal to think at the speed thought requires.

Elster’s positive doctrine, methodological individualism joined to rational choice theory, is where the modelling objection finds its clearest target, and Elster, to his credit, never hid the target. He held that social phenomena are in principle explicable by the actions and interactions of individuals, and that game theory supplies the grammar for the interactions. The classical objection replies that the grammar decides the metaphysics: a science whose sentences must have individuals as subjects will never state the claims, about forms, structures and totalities, that distinguish Marxism from liberal social theory. Elster’s rejoinder was to deny that such claims, once their metaphors are cashed, assert anything beyond complicated facts about individuals, and to challenge the classical tradition to produce one claim that survives scrutiny while resisting individualist restatement. The challenge was rarely taken up in the form offered, which the analytical side records as a victory and the classical side records as a refusal to play a rigged game, rigged because the rules of restatement exclude in advance the relational predicates the classical claims are built from. Here the generative disagreement shows its purest form, a dispute about what counts as an answer, conducted by two parties each sincerely unable to see the other’s procedure as inquiry rather than evasion.

What did Elster keep from Marx after the sifting finished?

More than his reputation suggests, since the sifted residue included the theory of ideology as motivated irrationality, the analysis of exploitation in revised form, the account of alienation stripped of essentialism, and a deep admiration for the historical monographs.

The discard pile was larger, dialectics, the value theory, the falling rate of profit as a law, yet the keeping matters for the comparison because it shows the analytical method operating as salvage rather than demolition.

Elster’s later distance from Marxism, and from rational choice orthodoxy itself, adds a final twist that both camps exploit. The scourge of functional explanation became a student of emotions, norms and irrationality, conceding that the rational actor of his models was a special case of the human material. Classical writers cite the concession as the framework confessing its poverty. Analytical writers cite the same concession as the method working on its own founder, evidence that a tradition of self-correction had operated exactly as advertised. The comparison cannot award the point to either side without begging the generative question, and so it records the twist as the dispute in miniature: one biography, two morals, and the choice between morals determined by a premise about method that the biography itself cannot supply.

Roemer and the reconstruction of exploitation

John Roemer’s contribution gave the analytical school its most technically accomplished social theory and the classical objection its most convenient specimen. Trained as an economist, Roemer set out to discover which parts of Marxian economics survived translation into the general equilibrium and game-theoretic language of his discipline. The answer he reached in 1982 rearranged the comparison’s landscape. Exploitation could be defined without the labour theory of value, through the withdrawal counterfactual described in the grid section above. Class could be derived from asset distributions under specified conditions. Capitalism’s moral indictment could be relocated from the labour market, where the wage bargain’s formal voluntariness had always embarrassed a crude coercion story, to the prior distribution of assets, where no voluntariness story gets started. Each result was a theorem, checkable by any competent reader, and the classical tradition, which had treated Marxian economics as a citadel, found the citadel’s rooms rearranged by a guest using the house’s own keys.

The classical response, from Wood and others, fastened on the relocation. If exploitation is unjust asset distribution, then Marxism’s economics dissolves into a branch of distributive ethics, and the theory of capitalism becomes a theory of unfair starting positions, equally applicable to any economy with unequal endowments. What disappears is the analysis of production as a social relation with its own dynamics: the drive to lengthen and intensify labour, the transformation of the labour process by machinery, the reproduction of the capital relation through the wage itself. Roemer’s defenders replied that disappearance is the wrong word for a deliberate abstraction, that the models isolate the distributive skeleton so its logic can be inspected, and that nothing prevents reattaching the flesh of production in further work. The reply is reasonable as methodology and incomplete as history, since the further work, by Roemer and by the school generally, moved toward ethics and institutional design rather than back toward the labour process. A research programme is judged by where its energy flows, and the energy flowed away from production, which is the classical point restated in sociological form.

Why did Roemer’s models persuade economists and alarm historians?

Because the two audiences value opposite virtues in a representation: economists prize the isolation of a mechanism in a form whose assumptions can be varied one at a time, while historians prize the preservation of a context in which mechanisms are entangled and dated.

A model that strips context to expose logic looks, to the first audience, like rigour and, to the second, like the removal of everything that made the event happen.

Roemer’s later proposal for market socialism, developed with Pranab Bardhan and others, extended the trajectory into institutional engineering: equalise assets through a coupon stock market, let markets coordinate, and exploitation in his defined sense disappears while efficiency survives. Whatever the proposal’s merits, its place in the comparison is diagnostic. A Marxism that culminates in a design for fairer shareholding has travelled a measurable distance from a Marxism that culminates in the abolition of the capital relation, and the distance was covered by a sequence of individually defensible analytical steps. Classical critics did not need to allege bad faith; the bibliography made the argument. Analytical defenders answered that distance travelled under argument is called learning in every other field, and that a tradition unable to distinguish learning from apostasy will preserve its doctrines by embalming them. Both speeches are by heart at this point in the literature, and the comparison’s contribution is to locate their source: the speeches disagree because the speakers disagree about whether a method may be trusted to revise the doctrine that adopted it, which is the generative disagreement wearing political dress.

Wright and the empirical wager

Erik Olin Wright wagered that the analytical reconstruction of class could survive contact with data, and he spent a career collecting the contact. His class structure research programme surveyed thousands of respondents across several countries, located them in a typology built from assets in means of production, organisational authority and skills, and tested propositions about income, consciousness and political allegiance that the classical tradition had asserted from the armchair of theory. The programme’s findings were mixed in the honourable way: the typology predicted income robustly, predicted consciousness weakly and unevenly, and generated anomalies, the contradictory locations behaved contradictorily, that kept a generation of sociologists employed. Wright reported the weaknesses with a candour that made his programme the analytical school’s best advertisement, since a Marxist research programme visibly capable of disappointing its director is a different creature from the doctrine machines the school’s founders had rebelled against.

The classical critique of Wright is the comparison’s most concrete, because it concerns identifiable choices. A survey fixes a moment; class, on the relational account, is a process extended in time, so the instrument is guaranteed to photograph the corpse of its object. A questionnaire individuates respondents; class, on the relational account, exists between people, in workplaces and unions and households, so the instrument’s unit of analysis excludes its object by design. Wright heard the critique, partly from friendly quarters, and his work migrated across his career toward case studies, institutional analysis and finally the study of real utopias, cooperatives, participatory budgets, basic income experiments, examined as existing alternatives within capitalism. The migration can be read classically, as the empirical conscience of the school discovering the limits of its founding instrument. It can be read analytically, as a research programme maturing from measurement to design. The readings share the facts and divide over the moral, exactly as the generative disagreement predicts, and Wright himself, who engaged Thompson’s tradition with respect and cited the relational critique without rancour, may be the figure in the whole comparison who came closest to practising the combination this article recommends, even though his starting definitions remained, to the end, the asset-based ones the classical side rejects.

What did the class surveys actually establish?

They established that position within production, measured by assets and authority, predicts income and material interest with a strength no rival stratification scheme matched casually, and that the path from position to consciousness runs through organisations, experiences and defeats that a cross-section cannot see.

The first finding vindicated the analytical wager; the second vindicated the classical warning, and the programme’s honesty consisted in publishing both.

Wright’s death in 2019 closed the school’s most productive empirical career, and obituaries across the divide acknowledged the wager’s nobility even while disputing its terms. For the comparison, his work supplies a control case against exaggerations on both sides. Against the classical exaggeration that analytical method is scholasticism, Wright’s decades of fieldwork and data stand as refutation. Against the analytical exaggeration that classical categories are untestable poetry, Wright’s own anomalies stand as evidence that the relational critics had identified real absences in the instrument, absences the data themselves kept announcing. A comparison that cannot hold both halves of that sentence has chosen a camp, and this article’s method, steelman before verdict, exists to prevent the choice from being made by inattention.

The classical lineage: Lukács, Korsch and the standpoint

The classical current’s founding documents were written in the aftermath of revolution and defeat, and their circumstances explain their emphases. Lukács’s History and Class Consciousness appeared in 1923, and Karl Korsch published Marxism and Philosophy in the same year, and both books argued, against the orthodoxy of the socialist parties, that Marxism was being ruined by its conversion into a positive science of society modelled on natural science. For Lukács, the ruin had a name, reification: the contemplative stance of bourgeois thought, which confronts social forms as things governed by laws, had infected the workers’ movement itself, so that Marxists studied capitalism the way its own economists did, from outside, as a mechanism. The antidote was the standpoint of the proletariat, the class positioned inside production in a way that lets it grasp society as a totality of human relations and therefore as transformable. Method, on this account, is not a technique a researcher selects. It is the self-consciousness of a historical process, and its validity is bound to the class whose practice it expresses.

The analytical reader’s difficulties with this founding text are genuine and should not be caricatured as incomprehension. The standpoint argument appears to validate a method by the social position of its holder, which reverses the order of justification the analytical temper requires: a method is validated by its results, whoever holds it. Lukács’s claims about totality appear, to the same reader, to license any particular assertion by reference to a whole that is never itself exhibited. Yet the classical reader’s difficulties with the analytical alternative are equally genuine. A method validated only by results, where results are counted by standards the method supplies, has sealed itself against the possibility that its standards are the problem, and Lukács’s insistence that the knower is inside the known, formed by the object and forming it, names a reflexivity that individualist social science keeps rediscovering under other names and filing as a technical complication. The standpoint doctrine in its strong form, that one class’s position grants epistemic privilege, has few defenders left in any camp. The weaker doctrine, that social position shapes what investigators notice, which questions they fund and which anomalies they forgive, is close to a commonplace of the sociology of knowledge, and the distance between the camps on this row is smaller in practice than the founding rhetoric suggests.

What survives of the standpoint argument after its strong form falls?

The insight that inquiry into society is conducted from inside society, by investigators whose categories, careers and curiosities are products of the relations under study, so a method that pictures the investigator as a spectator of a machine has misdescribed the epistemic situation before the first model is built.

Analytical philosophy of social science concedes the point as reflexivity while denying that it licenses a separate logic.

Korsch’s contribution, less read and in some ways sharper, was to historicise the theory itself: Marxism, he argued, passes through phases, its concepts altering as the movement that carries them alters, so that treating the doctrine as a timeless system betrays its content in the act of defending it. The argument anticipates, from inside the classical camp, a difficulty for any comparison like this one, which must freeze two currents into positions for the length of an article while knowing that the freezing falsifies. The article’s defence is pragmatic and is stated here once: the grid is an instrument for readers, its rows are ideal types abstracted from careers that moved, and the movement is restored wherever the prose follows individuals, Cohen toward ethics, Wright toward utopias, Lukács himself toward recantation and accommodation under party discipline. A comparison that denied its own idealisations would be more classical in method and less useful in fact, a trade the analytical side of the author’s desk declined and the classical side is invited to note.

Gramsci, the Frankfurt School and the cultural turn

Antonio Gramsci, writing in prison under fascism, extended the classical inheritance in a direction that complicates the grid productively. His concepts of hegemony, of the integral state, of organic intellectuals, analyse how a class rules through consent organised in civil society as much as through coercion, and the analysis is resolutely relational and historical: hegemony is a process, forever being made and unmade, not an asset a class holds. Analytical readers have found Gramsci the most congenial of the classical founders, because his arguments proceed by distinctions and cases that invite restatement, and several analytical writers attempted exactly that restatement for hegemony, with results the classical side received as promising dilutions. The episode matters because it shows the fork operating at the level of reception: the same text, read through one method, yields a research programme of mechanisms of consent; read through the other, yields a warning that consent detached from the historical bloc that organises it becomes a free-floating variable, measurable and meaningless.

The Frankfurt School pushed the classical inheritance toward its most pessimistic and most philosophical expression. Horkheimer’s distinction between traditional and critical theory restated the generative disagreement as a definition: traditional theory models itself on the sciences and takes its object as given, while critical theory knows its object as historically produced and its own activity as part of the object’s fate. Adorno’s negative dialectics radicalised the suspicion of concepts themselves, arguing that identification, the subsumption of particulars under concepts that analytical clarity requires, does a quiet violence to its objects, leaving a remainder that thought must honour by refusing closure. An analytical reader will find in Adorno the limiting case of everything the school opposed, prose that performs its resistance to paraphrase. A classical reader will answer that the performance is the argument, that a theory of a damaged world stated in undamaged prose would falsify itself. The comparison records the standoff without pretending to dissolve it, noting only that Adorno’s own practice, in the empirical studies of authoritarianism conducted with social scientists, was more hybrid than his philosophy advertised, a hybridity the combination section of this article treats as a precedent rather than an embarrassment.

Why does the dispute over prose style carry theoretical weight?

Because each side’s style enacts its epistemology, so the argument about how Marxist theory should be written is the argument about what it knows, conducted in another register.

Analytical prose wagers that an idea which survives plain statement was real; classical prose at its most difficult wagers that plain statement is where certain ideas go to be replaced by simpler ones.

Althusser occupies an unstable position in the classical camp, and the instability is informative. His structural Marxism shared the analytical school’s anti-humanism and its respect for science, while defending a theoretical practice, the production of concepts through a distinctive problematic, that the analytical school regarded as the obscurantism it existed to fight. The celebrated exchange in which E. P. Thompson’s The Poverty of Theory assaulted Althusserianism from a classical-humanist position, while analytical writers assaulted the same target from the opposite flank, produced the comparison’s richest three-cornered fight. Thompson and Elster agreed about almost nothing except that Althusser’s system was an evasion of evidence and history, and their agreement, reached from incompatible premises, suggests that the generative fork, though real, does not exhaust the map: a shared commitment to evidence can unite a dialectical historian and a rational choice theorist against a structuralism both regarded as theology. The grid survives the complication because the grid compares methods, not alliances, and alliances in this literature have always been promiscuous.

Colletti and the classical camp’s internal quarrel

Lucio Colletti is the classical tradition’s most inconvenient member for any simple telling of the comparison, because Colletti attacked, from within Marxism and with dialectical credentials, the doctrine of dialectical materialism that party orthodoxy treated as the theory’s crown. His distinction between contradiction and real opposition, drawn from Kant, argued that nature and logic know oppositions without contradiction, and that the Hegelian dialectic of nature was a metaphysical survival Marx’s materialism should discard. The argument brought Colletti, by an Italian road, to a position adjacent to the analytical school’s: suspicion of a special dialectical logic governing all domains. Yet Colletti remained a classical writer in his defence of the critique of political economy as an analysis of forms, and his late pessimism about the tradition’s future was framed in terms no analytical writer would use. His career demonstrates that the generative disagreement runs through camps as well as between them, and that the grid’s rows describe pressures and tendencies whose grip on any individual writer is partial and contested.

The Italian tradition around Colletti, descending from Galvano Della Volpe, had in fact practised a kind of rigour the analytical school might have recognised: Della Volpe’s logic of the concrete, his insistence on determinate abstraction and on testing concepts against historical material, was a protest against Hegelian vagueness mounted a generation before the September Group. That the two protests, Roman and Oxonian, barely engaged each other is one of the comparison’s quiet scandals, attributable to language, to institutions and to the Cold War’s sorting of intellectuals into camps whose reading lists did not overlap. A reader assembling the full picture should therefore treat the analytical and classical labels as the names of a late and partial crystallisation, the moment when an older, messier set of quarrels about rigour and totality acquired institutions, journals and a generational story. The crystallisation is real, the school existed, the meetings happened, the books answered each other, but the comparison falsifies its subject if it projects the crystallised camps backward onto a century in which the same arguments wore other uniforms.

What does Colletti’s case do to the eight differences grid?

It warns the reader that the grid maps a fault line, not two armies, since the most damaging attack on dialectical logic as a universal method came from a Marxist formed inside the classical tradition, while several analytical writers retained functional and holistic explanatory habits their own manifestos should have forbidden.

Fault lines run through authors and even through single books, which is why the grid’s rows are stated as positions rather than as memberships.

The lesson generalises into a rule for using this article. When a writer is cited for a position, the citation fixes a text and a date, not a soul. Cohen of 1978 and Cohen of 2000 stand on different rows of several differences. Wright migrated across the class row’s border in spirit while keeping his definitions. Elster’s later work on irrationality strained his own methodological individualism from inside. The classical camp’s members likewise migrated: Thompson’s empiricism sat awkwardly with Lukács’s totality, Wood’s materialism with Adorno’s negativity. A comparison honest about migration is more useful than a comparison that wins by freezing its subjects at their most quotable extremes, and the freezing temptation is itself a methodological choice, analytical in its love of stable objects, which the article acknowledges as it commits it.

The value-form tradition and the meaning of retention

The classical retention of the labour theory of value deserves a fuller hearing than polemics usually grant, because the retained theory, in its strongest contemporary form, is not the doctrine the analytical school refuted. Isaak Rubin, writing in the nineteen twenties, argued that value is a social form: the theory explains why the products of private labours confront their producers as bearers of a property, value, that seems to belong to things. Later scholarship in this line, associated with Hans-Georg Backhaus and with the German reconstruction of Marx’s monetary theory, reads Capital as a dialectical presentation of forms, money derived necessarily from the commodity, capital from money, each form analysed for what it presupposes and conceals. On this reading, the quantitative questions that exercised the transformation debate, whether prices of production can be derived from values with aggregate equalities preserved, concern a subordinate layer of the theory, and the analytical verdict on that layer, largely negative, leaves the form analysis standing.

The analytical rejoinder is the school’s characteristic move, and it is powerful: then say so, in plain terms, and surrender the rest. If the labour theory survives as a theory of social form, its defenders should stop teaching it as a theory of price determination, should mark the boundary between the surviving and the surrendered parts, and should accept that the tradition’s textbooks, primers and party schools taught the surrendered parts as the essence for a century. The rejoinder draws blood because it is just. Form theorists who reply that the quantitative layer was always subordinate must explain why generations of Marxists experienced its refutation as a mortal threat, and the explanation, that the tradition itself could not distinguish its layers until critics forced the distinction, concedes the analytical school’s founding charge, that obscurity had protected confusion, in the very act of resisting its application. The comparison scores the value row as divided, and the division has a precise content: the analytical side won the argument about what the theory asserts, the classical side retains a defensible answer about what the theory explains, and the retained answer is smaller, stranger and more interesting than the doctrine either side started with.

Is the value-form reading a rescue or a replacement?

It is a rescue that replaces, since the theory it saves, an account of why social labour appears as value, differs in its central question from the theory the tradition transmitted, an account of what determines exchange ratios, even though both travel under one name.

Recognising the replacement inside the rescue is the analytical contribution; recognising the worth of the rescued theory is the classical one.

A parallel structure recurs across the grid, and naming it helps the reader more than another example. Again and again, the analytical examination forced a doctrine to specify which of its fused claims it meant, and the specification divided each doctrine into a part that failed testing and a part that survived by becoming more precise about its object. Historical materialism divided into a development thesis that weakened and a research heuristic that persists. Exploitation divided into a labour-accounting claim that fell and a distributive claim that travelled. The dialectic divided into a logic that few defend and a sensitivity to form, process and internal relation that many practise without the name. The classical tradition experiences this pattern as dissolution, the whole dismembered into defensible fragments that no longer compose a world view. The analytical school experiences the same pattern as maturation, a world view growing into a set of theories. The experiences are both accurate reports of one process, and the choice between the words dissolution and maturation is, once more, the generative disagreement deciding what a theory is for.

Dialectics examined: contradiction, totality and mediation

Since the whole comparison turns on a method the classical side calls dialectical, fairness requires stating what the term covers in competent hands, because analytical polemics often attack a version no competent hand would sign. Dialectical analysis, in the tradition descending from Marx’s critique of political economy, names a cluster of commitments rather than a syllogism machine. Categories are treated as historically determinate, valid for a form of society and misleading outside it. Relations are treated as internal where the relata are constituted by the relation, as worker and capitalist are constituted by the capital relation rather than meeting as preformed atoms. Processes are treated as developing through tensions internal to forms, the contradiction between use value and exchange value inside the commodity being the founding example. Wholes are treated as structured, so that an element’s behaviour depends on its place, and appearances are treated as real but inverted, forms in which underlying relations show themselves disguised, as the wage shows the sale of labour power disguised as payment for labour performed.

Each commitment can be assessed separately, and the analytical school’s actual practice, as opposed to its manifestos, assessed them very differently. Historical determinacy of categories was accepted without fuss; no analytical writer modelled feudal dues with the categories of capital. Internal relations were resisted in name and used in practice, since Roemer’s classes are constituted by asset relations and Wright’s positions by organisational relations, neither thinkable as atoms. Development through internal tension was translated, by analytical writers on crisis and transition, into models of endogenous instability, a translation the classical side may regard as lossy while conceding that something recognisable crossed over. The structured whole survived as the concept of an economic structure in Cohen’s reconstruction. Appearance and essence survived as the distinction between surface forms and underlying mechanisms, which Elster’s ideology theory uses constantly. What did not survive, anywhere in the analytical literature, is the claim that these commitments compose a logic, a method of inference with its own validity, exempt from the demand that each step be statable and checkable. The residue of the dispute is therefore narrower than the rhetoric: not whether Marx’s characteristic sensitivities are valuable, both sides in practice grant most of them, but whether they amount to a proprietary method that licenses conclusions ordinary argument cannot reach and protects them from ordinary refutation.

Which dialectical commitment proved hardest to translate?

The doctrine of contradiction as a motor, since translation into endogenous instability captures systems that destabilise themselves but loses the stronger classical claim that a form both requires and undermines its own conditions in a single movement, so that its development and its crisis are one process viewed at different speeds.

Whether the stronger claim asserts anything beyond the translation remains the comparison’s most technical open question.

Classical writers, pressed on the logic claim, often retreat to a defensible perimeter: dialectics as a heuristic, a discipline of attention that directs inquiry toward relations, histories and internal tensions, and away from static classification. The retreat is honourable and consequential, because a heuristic makes no proprietary claim and can be combined with any standards of evidence, which is precisely the combination this article recommends. What the retreat surrenders is the founding classical insistence that the method is inseparable from the content and unavailable to outsiders. A heuristic of attention is available to anyone, including the analytical school’s own best practitioners, whose work is full of relational and historical sensitivity deployed inside an individualist frame. The comparison therefore records a quiet convergence at the level of practice beneath a loud divergence at the level of doctrine, and it treats the convergence as genuine progress rather than as a trick: traditions often settle their disputes by absorbing each other’s tools while keeping their flags, and the flags matter less than the tools to readers who come to the dispute for understanding rather than for membership.

The self-ownership debate and the normative turn

The analytical school’s normative turn produced the comparison’s most technically refined debate, the argument over self-ownership. Nozick had built libertarianism on the premise that persons own themselves, so that taxation of earnings resembles forced labour. Cohen’s counterattack accepted the premise for argument’s sake and showed that self-ownership, combined with unequal world ownership, generates outcomes its defenders should find uncomfortable, while a separate line of his work argued that Marxists had themselves relied on self-ownership tacitly, condemning the capitalist for stealing the worker’s labour in a vocabulary that presupposed the worker owned it. The resulting tangle, in which Marxist exploitation talk appeared to rest on a libertarian premise, delighted the school’s critics and instructed its members. Resolutions proliferated: exploitation redefined to avoid the premise, the premise accepted and the world ownership equalised, the premise rejected in favour of an egalitarianism that needs no ownership talk at all. Cohen’s own resting place, that equality is the value and self-ownership a dispensable instrument, marked the school’s arrival at a moral philosophy continuous with the broader egalitarian literature and discontinuous with anything the classical tradition would recognise as its own.

Classical writers watched the debate with a distinctive double reaction. As Marxists, they noted that a decade of ingenuity had been spent on a concept, self-ownership, that belongs to the juridical surface of bourgeois society, and that the tradition’s own categories, labour power as commodity, the wage form, the production of surplus, had been spectators at their own trial. As readers, the better classical critics conceded that the debate exposed a real incoherence in popular Marxist moralising, which does rely on theft imagery it cannot cash. The episode is the normative row of the grid in narrative form: analytical candour about values produced genuine discoveries, and the discoveries were purchased by relocating the theory’s centre of gravity from production to distribution, from history to moral philosophy. Whether the purchase was wise depends on what one thinks Marxism is, which is the identity question the next section takes up directly, and which the generative disagreement ensures cannot be answered neutrally.

Did the self-ownership debate change any minds outside the school?

It changed the terms on which egalitarians argued with libertarians across political philosophy, since Cohen’s strategy of accepting an opponent’s premise and deriving unwelcome consequences became a standard exercise, taught to students who never learn its Marxist provenance.

Inside the comparison, that migration of a result beyond its tradition counts as evidence for the analytical wager that clarity travels.

The debate also illustrates a risk the classical tradition predicted with accuracy. A research programme that migrates into a neighbouring discipline competes under the neighbour’s standards of prestige, and those standards reward the problems the neighbour finds central. Political philosophy in the analytical mode prizes puzzles of distribution under idealised conditions. Marxism’s home problems, crisis, class formation, the state, imperialism, the transformation of production, are not puzzles of that kind, and a generation formed by solving the neighbour’s puzzles loses the taste and eventually the equipment for its own. The school’s bibliography across two decades shows the drift in aggregate even where individuals resisted it, and the drift is the retreat from class argument stated at the level of the profession rather than the proposition. None of this shows the normative work was bad. Much of it was excellent. It shows that excellence is not the only criterion by which a tradition judges its allocations, since a tradition is also a division of labour across generations, and the classical complaint is finally about the division, about who was left minding the home problems while the school’s talent argued about self-ownership in the journals of another field.

Is analytical Marxism still Marxism?

The identity question shadows every row of the grid, and the comparison owes it a direct hearing. Three criteria compete. A doctrinal criterion counts a position as Marxist when it retains designated theses, the labour theory of value, the primacy of class struggle, the designated agent, and by that criterion the analytical school’s mature positions fail severally and jointly. A genealogical criterion counts descent: a position is Marxist when it arises from serious engagement with Marx’s texts and problems, by that criterion the school qualifies easily, since few bodies of work in the century engaged the texts more closely. A methodological criterion, the classical tradition’s own when Lukács is speaking, counts fidelity to the dialectical method as the essence, and by that criterion the school fails by definition and by intention. The criteria deliver different verdicts because they encode different answers to the generative question, doctrinal criteria treating the theory as a bundle of claims, the methodological criterion treating it as a way of knowing, the genealogical treating it as a tradition of problems. A reader’s choice among criteria is therefore not a preliminary to the comparison but a restatement of its central fork, and the article, having traced the fork through eight differences, declines to smuggle in a ninth decision disguised as a definition.

What can be said without begging the question is sociological. The school’s members divided over the label themselves. Some, Elster most plainly, concluded that what remained true in Marx did not amount to a Marxism and said so. Others, Cohen most poignantly, described a loyalty divided between the tradition that formed them and conclusions they could not unsay, keeping the label as an address rather than a creed. Wright used the label programmatically to the end, arguing that a Marxism rebuilt on defensible foundations was the only kind worth the name. The classical tradition’s gatekeepers, applying doctrinal and methodological criteria together, mostly read the school out, while continuing to cite its reconstructions when convenient, a combination that suggests the boundary is policed rhetorically and crossed practically by everyone. For students, the practical advice is to treat the label as a question the literature argues about, not as a fact the literature reports, and to specify, whenever the word Marxist does work in an essay, which criterion is in force. That advice is analytical in form and neutral in effect, since classical writers gain as much as anyone from criteria stated in the open.

Why does the identity question resist settlement by evidence?

Because the disputants are not disagreeing about a matter of fact that evidence could display, but about which facts about a tradition, its doctrines, its descent, its method, constitute membership in it.

Evidence about texts and careers is abundant and shared; the weighting of that evidence is the dispute, and weightings are chosen, not discovered, so the argument continues at the level of choice.

The identity question also has a forward-looking edge that the comparison should not miss. Traditions survive by the work done under their names, and the analytical episode altered what the name Marxism can mean for later generations in two opposite directions. It demonstrated that Marxist problems could command the attention of first-rate philosophers and social scientists working to general standards, a demonstration whose prestige effects outlast the school. It also associated the name, in the minds of a cohort of students, with a doctrine that its own best examiners had largely surrendered, an association whose costs the classical tradition itemises bitterly. Which effect dominates depends on the counterfactual, what those talents and those students would have done in the school’s absence, and counterfactuals of that scale are exactly the kind of claim both methods handle badly, the analytical for want of a model, the classical for want of a control history. The article records the standoff as data about the limits of both instruments, a rare row in which the grid’s honest entry is that neither side’s tools can grade the other’s homework.

Brenner, the transition debate and the meeting over evidence

The Brenner debate deserves its own section because it is the comparison’s best evidence that the two tempers can meet over archives and numbers when the question is posed at the right scale. The question was why the crisis of the fourteenth century issued in serfdom’s dissolution in western Europe and its intensification in the east, why capitalism emerged from agrarian relations in England rather than from the more commercialised economies elsewhere. Brenner’s answer gave causal primacy to class relations and the outcomes of class struggles, against demographic and commercial models that treated population and trade as self-moving forces. The debate that followed, across Past and Present and in volumes of replies, drew in historians with no stake in the Marxist label and obliged every participant to specify mechanisms, sequences and comparisons. Analytical readers found in Brenner a classical argument built to their standards of evidence. Classical readers found in the debate’s reception a vindication of their insistence that relations, not aggregates, carry causal weight.

The episode cuts both ways, which is why it is valuable. It shows a classical thesis surviving, indeed dominating, an evidential contest conducted without dialectical protection, which weakens the analytical suspicion that classical claims need their method’s shelter. It also shows the classical thesis winning precisely by behaving analytically for the duration, distinguishing claims, naming mechanisms, accepting that specified evidence would tell against it, which weakens the classical claim that such behaviour falsifies the object. Partisans prefer one moral; the comparison draws the joint one, that the opposition between the tempers is real at the level of doctrine and porous at the level of good practice, and that the porosity is where the living research in both traditions has always happened. Brenner himself, a classical writer by formation and allegiance, is read in graduate seminars beside Roemer, and the syllabus that places them together teaches the comparison’s lesson better than any summary: the fork is genuine, the inheritance is shared, and the work is judged, in the end, by readers who hold the two standards in one head and refuse to excuse either.

What made the transition debate a fair test between the tempers?

The question was concrete enough for evidence to bite, a divergence of outcomes across regions sharing demographic and commercial pressures, yet structural enough for the rival methods to generate genuinely different explanations rather than different descriptions.

Most disputes in the comparison fail one side of that condition, being either too abstract for evidence or too narrow for method to matter.

A second meeting over evidence deserves mention beside Brenner’s: the standard of living debate among historians of industrialisation, where Marxist and non-Marxist scholars contested wages, heights, mortality and consumption across generations of archival work. Analytical habits, index construction, sensitivity to composition effects, disciplined the contest; classical habits, attention to the relations behind the aggregates, to what a wage concealed about dependence and command, kept the contest honest about its object. The debate ended without a victor’s parade, in a refined picture of early industrial living standards that neither temper would have produced alone. The comparison cites it as a precedent for the combination it recommends, and as a reminder that the reader’s interest, which is understanding capitalism’s history, is served by instruments in combination even when the instruments’ owners are at war.

The September Group and the sociology of a school

Institutions shape arguments, and the comparison gains realism by attending to the school’s. The analytical Marxists met as a group, informally at first from 1979, in annual September gatherings that gave the circle its nickname, with a rule of discussion that became its legend: no bullshit, a phrase Cohen used as the group’s motto and later as the title of an essay on its ethos. Papers circulated in advance, claims were pressed until they clarified or collapsed, and seniority conferred no immunity. Participants included, at various times, Cohen, Elster, Roemer, Wright, Van Parijs, Przeworski, and visitors from neighbouring fields, with Robert Brenner and others at the edges supplying the classical counterweight inside the room. The group’s sociology explains features of its output that pure intellectual history misses: the shared vocabulary, the rapid internal criticism that kept positions moving, the collective migration across problems from historical materialism to exploitation to equality, as the room’s attention swung and carried its members’ agendas with it.

The classical tradition’s institutions were of a different kind: parties and their schools in the earlier period, journals and university enclaves later, with New Left Review as the commanding Anglophone instance, a journal whose editorial collective under Anderson pursued a deliberate strategy of importing continental theory to renovate a British Marxism judged provincial. The two institutional forms bred two relations to doctrine. A seminar of peers with a no-bullshit rule manufactures revision as a by-product of conversation, since every position must survive colleagues who gain prestige by denting it. A journal with a renovation strategy manufactures synthesis and importation, valuing the essay that reframes a field over the note that corrects a premise. Neither institution is a conspiracy and neither is innocent: each selects for the virtues its tradition celebrates and against the virtues it neglects, which is why the comparison between the currents is also, unavoidably, a comparison between ways of organising intellectual life. Readers choosing a side are choosing, whether they know it or not, a seminar room or an editorial office as the place where their own thinking will be corrected.

What did the no-bullshit rule exclude besides obscurity?

It excluded, in practice, certain kinds of speech the classical tradition values: appeal to the whole as a reason for tolerating a part’s vagueness, historical narrative offered as self-validating, and the citation of a founder’s authority as a move in argument.

Excluding those moves made the room’s arguments faster and cleaner while guaranteeing that the classical objection, which is partly stated in the excluded idioms, would be heard there as noise.

The rule’s costs surfaced in the group’s treatment of its own founding question. Because the modelling objection could not be stated, in the room’s idiom, as anything but a request for a better model, it was processed as a technical challenge and answered with technical improvements, while its actual content, that the idiom itself was the issue, went undiscussed in the form its proponents intended. Institutional analysis thus converges with the article’s verdict from an unexpected direction: the school’s failure to answer the objection was not an oversight of individuals but a predictable output of its discussion rules, which filtered out the objection’s genre. Classical institutions have the mirror defect, filtering out counterexample and quantification by an idiom in which precision reads as naivety. A reader who wants the benefits of both rooms must build a private institution of one, reading each tradition in the other’s presence, which is laborious and is also, the comparison submits, the only honest way the dispute can be studied at all.

Preferences, fetishism and the depth of the givenness problem

The modelling objection’s second assumption, given preferences, connects to the oldest stratum of Marx’s theory, and the connection deserves excavation because it shows the objection is not a debating point but a structural feature of the theory being modelled. Commodity fetishism, in the first chapter of Capital, names the process by which social relations between producers appear as relations between things, endowed with value as a natural property. The analysis implies that the agents of a market society systematically misrecognise their own relations, taking as given, natural and prior a set of arrangements that their collective practice produces. Preferences, on this account, are not merely influenced by society in the loose sense that upbringing matters. They are formed inside the fetishised forms, so that what an agent wants, what an agent takes to be a good, a cost, a fair price or a natural hierarchy, bears the stamp of the relations the theory exists to criticise. A model that begins with those wants as data has therefore not made a neutral simplification. It has taken the phenomenon’s surface, the formed preferences of formed agents, as its foundation, and built the critique’s object out of the critique’s target.

Analytical writers were not unaware of the problem, and the school’s treatments of ideology and adaptive preferences show the awareness at work. Elster’s account of sour grapes, preferences adapting to the feasible set, imported a mechanism by which relations shape wants, and his work on ideology treated beliefs as motivated by position in ways that echo the fetishism analysis with the metaphysics removed. Cohen’s writing on the structure of proletarian unfreedom likewise examined how choices made rationally by each worker compose an outcome no worker chooses. These are real engagements, and the classical side does its case no good by ignoring them. The objection survives them in a narrowed form: engagement at the level of mechanisms, added to a framework whose primitives remain individual wants and endowments, treats formation as an influence upon agents who exist prior to it, whereas the fetishism claim is that the agents of market society, as bearers of preferences of the relevant kind, are themselves precipitates of the relations. The distance between influence and constitution is the distance between the two methods stated at the level of philosophical anthropology, and it is not obvious that any formalism can cross it, since formalisms require elements whose identity survives the relations into which they enter.

Can a model represent agents constituted by their relations?

Only by cheating gracefully, in the judgement this article reaches: by treating constitution as a prior stage whose outputs become the model’s given agents, which represents the history of formation while denying its continuing operation.

The cheat is often worth performing, since a staged model teaches more than no model, yet the classical point stands that the representation misplaces the phenomenon it stages.

The givenness problem also explains why the dispute recurs in fields far from Marxism, a recurrence the next section surveys. Any inquiry whose object includes the formation of the inquirers and their categories faces the same fork: proceed with given elements and accept the misplacement, or honour the formation and accept the loss of formal control. Psychoanalysis faced it over the formation of desire, anthropology over the categories of kinship, theology over the formation of the believing subject. The Marxist case is the best documented because the school quarrel forced both horns into print with unusual clarity, and readers from those neighbouring fields who find their own disputes pre-enacted in this one are not imagining the resemblance. The comparison’s value as a case study in the general problem of method is a theme the article presses in its closing sections, since it may outlast the substantive dispute that generated it.

The general question: may a tradition claim a distinctive logic?

Abstracted from Marxism, the generative disagreement asks whether any research tradition may claim a logic of its own, exempting its arguments from the standards that govern inquiry generally. The question recurs wherever a body of thought feels itself misjudged by foreign criteria. Psychoanalysis claimed that its evidence, dreams, slips, transference, could be assessed only by the method that produced it, and critics replied that a method answerable only to itself is a machine for manufacturing confirmation. Interpretive social science claimed, in some of its versions, that human action must be understood from within, by categories foreign to causal explanation, and its critics replied that understanding which cannot be wrong cannot be knowledge. Theology has claimed, across centuries, that its object requires a mode of reasoning proportioned to revelation, and its critics have replied with the analytical school’s exact sentence in older dress, that a claim which cannot be stated clearly may not be a claim. The Marxist dispute is the most instructive instance because both sides wrote at the highest level and because the stakes, a theory of society and a politics, forced the methodological argument out of the philosophy seminar and into the open.

The comparison yields a transferable test for such claims, assembled from the two sides’ strongest points. A tradition claiming a distinctive method should be asked three questions. What does the method let its holders see that general standards miss, stated as findings rather than as promises? Where has the method been wrong by its own lights, and what in it permitted the error to be recognised? What would its holders accept as a fair external check, and if the answer is nothing, what distinguishes the method from an immunity? The analytical tradition inside Marxism scores well on the second and third questions and must work for the first. The classical tradition scores well on the first, its findings about form, fetishism and totality are genuine exhibits, and must work for the second and third, where its record of self-recognised error is thin and its acceptance of external checks grudging. A tradition that could answer all three fluently would have earned a distinctive method in the only sense worth wanting, as a disciplined specialisation rather than an exemption, and neither Marxist current, the article judges, reached that standard across the whole of its practice, though individuals in both came close.

Why is the demand for a fair external check not itself a rigged standard?

Because the check demanded is not that the tradition pass foreign tests on foreign questions, but that it name some encounter with the world, on its own questions, in which it could discover itself mistaken.

A method that can describe its own possible failure has connected itself to reality; a method that cannot has connected itself only to its holders.

That test also disciplines the analytical side’s triumphalism. General standards are not a view from nowhere; the demand for mechanisms, for individual-level foundations, for statement in ordinary logical form, are themselves commitments with histories, and a school that cannot say what would count against methodological individualism as a framework has its own immunity problem at one remove. Elster, to his credit, approached the question by treating individualism as a regulative ideal rather than a metaphysical discovery, a concession whose importance the classical side underrated. The fairest summary of the general question is therefore double-edged: distinctive logics claiming exemption fail the test, and universal logics claiming innocence fail a milder version of it, so the defensible position is a universalism that knows its own premises are premises, which is, stated abstractly, the combination the article has recommended since its middle sections.

Misreadings the comparison must correct

Five misreadings recur in summaries of the dispute, and correcting them is a service to readers arriving from them. The first holds that the quarrel was about jargon, a style war between clear writers and obscure ones. Style was the symptom and sometimes the weapon, never the substance: the substance is whether a distinctive method exists, and plenty of clear classical prose, Thompson’s and Wood’s among it, argues the classical side, while plenty of forbidding technical prose argues the analytical one. The second misreading holds that the analytical school set out to destroy Marxism, a fifth column of liberals in Marxist dress. The record refutes it: the founders were defenders, several were socialists of long standing, and their abandonments were published with reasons and often with regret, a pattern no infiltration hypothesis explains. The third misreading, from the other flank, holds that classical Marxists reject clarity as such, preferring mist to light. The tradition’s best writers are formidably clear about everything except the points where they believe clarity of the analytical kind falsifies, and the belief, right or wrong, is an argument, not a taste for fog.

The fourth misreading holds that formalisation is politically neutral, a tool equally available to any politics. The modelling objection is precisely the denial of this, and the school’s trajectory is the denial’s best illustration, since the tools, applied faithfully, kept delivering their users to the same political neighbourhood. Tools are not neutral when they select what can be said, and every method selects. The fifth misreading, the subtlest, holds that the school’s trajectory settles the methodological question, proving the classical case by the analytical school’s own history. It does not, because the trajectory is compatible with the analytical reading, that honest examination dissolved a weak doctrine, as well as with the classical reading, that a loaded method dissolved a strong one. Trajectory is evidence that requires the modelling objection to become an argument, which is why this article built its verdict on the objection rather than on the biography of the school. Readers who keep the five corrections in mind will find the primary literature less confusing than its reputation, since much of the confusion is manufactured by summaries written from inside one camp’s misreading of the other.

Which misreading does the most damage to students?

The first, the jargon reading, since it sends students to the dispute expecting a lesson in writing and leaves them defenceless before the actual argument about method, evidence and social ontology.

A student who grasps that the quarrel concerns what may count as a reason has been equipped for every later encounter with the literature; a student expecting stylistic tips has been equipped for none.

A sixth misreading, worth adding for researchers, holds that the dispute ended. It did not end; it dispersed. The questions the school posed about exploitation, class and equality migrated into mainstream political philosophy and sociology, where they are pursued by scholars who may never cite the quarrel that generated them, while the classical questions about form, totality and fetishism persist in Marx scholarship and in the critical theory that descends from it. The two literatures cite each other rarely, which is the institutional form of an unsettled argument: not agreement, not refutation, but a parting of the shelves. Readers of this comparison are therefore not studying a closed episode. They are studying the headwaters of two streams that still run, mostly out of each other’s sight, through the contemporary study of society, and the generative disagreement runs in both streams as an underground river, surfacing whenever a modeller meets a dialectician over a fresh object and each discovers, with a sense of recognition, that the old argument has begun again in new vocabulary.

The state, the party and imperialism under the fork

The derivation claimed for the generative disagreement can be tested against topics the grid does not list, and the article runs the test the earlier sections promised. Take the state. Classical theory, in its several variants, treats the capitalist state as a form of the capital relation, its autonomy relative and structured, its policies bounded by the reproduction of accumulation, whether the mechanism is personnel, finance or the silent discipline of investment decisions. Analytical treatments decompose the claim: which actors, in which ministries and parties, with what incentives, produce the outcomes attributed to the structure, and Przeworski’s models of social democratic governance showed governments of the left administering capitalism because the investment decisions of private owners punish alternatives, a mechanism any classical writer recognises as the structural constraint stated in another grammar. The translation is impressive and the classical residue is identifiable: the model explains why governments retreat, given the structure of ownership, while the classical theory asks why that structure presents itself to governments as a fact of nature rather than as the stake of politics, a question about legitimation and form that the model’s actors, choosing within the structure, cannot pose from inside it. The fork reappears, precisely as predicted, inside the translation’s remainder.

Take the party. Classical theory designates the revolutionary party as the bearer of class consciousness to a class that spontaneous struggle leaves at the level of trade union demands, a doctrine whose epistemology is the standpoint argument institutionalised. Analytical treatments ask the questions the doctrine forbids: by what mechanism does the party acquire the consciousness it bears, what checks its claim against the class’s own expressed judgements, and what in its structure prevents the bearer from becoming a substitute that rules in the name of the absent class. The historical record of parties that answered those questions badly is long, and analytical writers drew on it with effect. The classical residue, again identifiable, is the observation that organisation is not an optional addition to class agency but its medium, that a class without durable organisation is not an agent at all, and that models of individual political choice systematically underrate the collective goods, identities and disciplines that organisation produces. Each side’s residue is the other’s research programme, and a reader who can state both residues has understood the state and the party better than a reader armed with either tradition alone.

Take imperialism. Classical theory, from Lenin through the dependency writers, treats expansion and hierarchy among states as expressions of capital’s structure, monopoly, finance and the export of capital binding peripheries into a world system whose relations constitute the units. Analytical treatments ask for the mechanisms and the counterfactuals: which firms, which states, which decisions, and whether the outcomes attributed to imperialism follow from asset distributions and incentives that a model can vary. The analytical pressure improved the classical theory where it landed, forcing distinctions between expansion as policy and hierarchy as structure. The classical residue is the world system’s constitutive claim, that the units themselves, core and periphery states, national capitals, are products of the relation between them, so that a method beginning from given units and their preferences re-enacts, at the level of nations, the givenness problem the fetishism analysis identified at the level of persons. Three topics, three confirmations of the derivation, each with the same shape: translation succeeds across a wide middle, and the untranslated remainder is always the constitutive claim, the assertion that the elements are made by the relations. A derivation that predicts the location of disagreement in fresh territory has earned the name this article gives it, generative, since it generates not only the eight listed differences but the form of every successor dispute.

Does the derivation leave anything between the currents unexplained?

Personal and institutional loyalties explain much of the dispute’s heat and some of its alliances, yet loyalties do not explain the content of the positions taken, which is what the derivation claims.

Where a disagreement between the currents cannot be decomposed into the fork, the article’s claim would be damaged; the standing candidates, over the state, the party and imperialism, decompose on inspection, as this section has shown.

The test also clarifies what a genuine reconciliation would require. It would not require splitting differences, half a dialectic and half a model in every paragraph. It would require a theory of constitution, an account of how relations produce the agents whose interactions reproduce the relations, stated with analytical explicitness and accepted by classical writers as capturing their constitutive claim rather than staging it. Fragments of such a theory exist in the literature on structuration and in critical realism, whose founder Roy Bhaskar engaged Marxism directly and whose transformational model of social activity, society as both condition and outcome of action, was built to hold exactly the two truths the fork divides. The analytical reception of critical realism was cool, finding its prose a relapse into the idioms the school existed to cure. The classical reception was divided, welcoming the ontology and suspecting the system building. The episode is recorded here as the comparison’s most promising unfinished business: a candidate bridge, inspected by both banks, crossed by neither army, used quietly by individual travellers ever since.

A short history of the quarrel in five scenes

The dispute has a chronology, and a compressed narrative fixes the dramatis personae in the reader’s memory before the verdict. The first scene is prehistory: the nineteen sixties, when Althusserian structuralism dominated Marxist theory in the universities and its opponents, Thompson above all, fought it with historical scholarship and polemic. Into that fight the analytical school had not yet entered, yet its founding grievances, obscurity as authority, quotation as argument, were the grievances of Thompson’s side, a kinship the later quarrel obscured. The second scene is foundation: Cohen’s 1978 defence, the September meetings from 1979, Roemer’s and Elster’s great books of the early nineteen eighties, Wright’s class programme launched, a school visibly constructing a Marxism to general standards and visibly proud of the construction. Admiration and alarm divided the older tradition, Anderson’s respect and Wood’s suspicion marking the poles.

The third scene is confrontation: the middle and later nineteen eighties, when the classical counterattack took organised form, Wood’s essays on rational choice and the retreat from class, Sayer’s and others’ defences of historical method, the exchanges over Thompson’s and Elster’s rival readings of Marx’s materialism. The modelling objection was stated in this scene, in the essays this article has treated as its principal statements, and the school’s response, better models and sharper concepts, was also given in this scene, so that by its end the dialectic of the dispute, objection at the level of framework, answer at the level of technique, was fully formed and has repeated since without essential variation. The fourth scene is migration: the nineteen nineties, when the school’s centre of gravity moved into normative philosophy, Cohen on equality, Roemer on market socialism and later on equality of opportunity, Van Parijs on basic income, and the label analytical Marxism faded from the work of its own producers, not with a renunciation but with a change of letterhead. The classical tradition, meanwhile, survived the collapse of the socialist states with its categories intact and its confidence damaged, turning increasingly to the exegesis of Marx’s texts and to cultural and ecological themes where its methods faced less competition.

Why did no decisive public debate between the camps occur?

Because a decisive debate requires shared rules for what would count as deciding, and the absence of such rules is the dispute’s subject, so the camps argued past each other in parallel literatures whose occasional intersections, reviews, conferences, a shared platform, produced clarification of positions rather than movement.

The quarrel was settled the way unsettleable quarrels are, by demography and migration rather than by verdict.

The fifth scene is dispersal, the condition described in the misreadings section: the questions separated and travelled, the analytical questions into philosophy and sociology’s mainstream, the classical questions into Marxology and critical theory, with occasional reunions when a writer like Wright or Brenner worked both sides of the street. A reader entering the literature in this fifth scene needs the map this article supplies more than a reader of the third scene did, because the signposts have been removed: the journals no longer announce the quarrel, the syllabuses assign the texts without the controversy, and the generative disagreement must be reconstructed from the works themselves. That reconstruction, performed patiently, is also the best training the dispute offers, since it teaches the reader to detect a method dispute inside a doctrinal one, a skill that transfers to every field where schools divide and the division is misreported as a list of conclusions.

Justice, ideology and the quarrel over Marx’s morals

One sub-dispute deserves expansion because it shows the analytical method discovering something the classical tradition had every reason to hide from itself. The question was whether Marx condemned capitalism as unjust. Wood’s influential reading held that he did not and could not, consistently: justice, for Marx, is a standard internal to the mode of production in question, measuring whether transactions conform to the relations they arise from, and by that standard the wage transaction is just, since the worker receives the value of labour power as the system’s own rules define it. Marx’s condemnations, on this reading, rest on other values, freedom, self-realisation, community, and on the system’s historical obsolescence, not on injustice. Geras answered with a patient accumulation of texts in which Marx calls the appropriation of surplus labour theft, robbery and embezzlement, vocabulary that presupposes a standard of justice the official doctrine disowns, and concluded that Marx held, inconsistently and fruitfully, both that capitalism is just by its internal standard and unjust by a trans-historical one his theory required and his meta-theory forbade.

The classical stake in Wood’s reading was protective: if Marxism needs no theory of justice, it cannot be refuted by one, and its critique remains scientific where liberal critique is moral. The analytical stake in Geras’s was foundational: if the tradition’s central condemnation rests on an unowned moral premise, then owning it, stating the theory of justice the condemnation requires, is the school’s charter in miniature. The dispute’s resolution, insofar as scholarship resolves, favoured Geras on the texts and Wood on the doctrine, a split verdict that pleased neither camp and instructed both. Its larger importance for the comparison is the demonstration that analytical methods could settle a question inside the tradition’s own archive, by evidence both sides accepted, when the question was posed at the level of texts rather than frameworks. The modelling objection governs framework questions; it does not govern all questions, and the school’s victories on textual and logical ground are permanent acquisitions that no classical revival will dislodge, because they were won under rules the classical side, in those instances, shared.

What does the justice quarrel reveal about ideology as a concept?

It reveals that the tradition used ideology in two senses it rarely distinguished, false belief that serves power and the whole system of categories a society thinks with, and that analytical pressure forced the distinction into the open, after which ideology critique could specify whether it was charging error, function or form.

Each charge requires different evidence, and much classical writing had drawn strength from sliding among them.

Ideology is in fact the topic where the two methods most obviously need each other, and the article pauses on it as a final worked example before the verdict. The analytical treatment, Elster’s above all, analyses ideology as motivated belief formation: actors adopt beliefs because of their position, through mechanisms of dissonance reduction, wishful thinking and selective exposure, each mechanism testable in principle and some in practice. The classical treatment analyses ideology as form: the categories of everyday economic life, the wage as payment for labour, the market as a meeting of equals, are themselves ideological, not because anyone errs but because the relations present themselves inverted, so that seeing them aright requires a critique of the forms of appearance and not merely of the beliefs they cause. A complete account of a concrete ideology, of meritocracy, say, or of the naturalness of competition, needs both: the forms that present the social order as fair and the mechanisms by which situated agents come to inhabit those forms as their own convictions. Neither tradition produced the complete account alone, each having defined the other’s half as the dispensable part, and the reader who assembles the halves has performed, on a single topic, the combination this article recommends as its practical conclusion.

Van Parijs, basic income and the school’s furthest shore

Philippe Van Parijs carried the analytical reconstruction to its most policy-concrete conclusion, and his trajectory compresses the comparison into one bibliography. Beginning from a Marxist engagement with exploitation and unemployment, Van Parijs argued, in work culminating in Real Freedom for All in 1995, that justice requires the highest sustainable unconditional basic income, funded from the value of scarce assets including jobs themselves, reconceived as gifts whose holders owe their scarcity rents to society. The argument is analytical in every fibre: premises stated, the concept of exploitation rebuilt as unequal access to assets, the conclusion derived and defended against the objection, pressed by critics, that the idle surfer on a grant exploits the working taxpayer. Van Parijs’s reply, that employment itself is an asset whose rent the employed already capture, turned the exploitation concept against its intuitive users with a elegance the school’s founders would have applauded.

The classical reading of the trajectory writes itself, and parts of it write themselves justly. A Marxism whose practical horizon is a grant administered by a capitalist state, justified by a theory of asset fairness, has exchanged the transformation of production relations for a redistribution of their fruits, and the exchange was driven by the framework: begin from assets and distributions, and the best imaginable world is a fair distribution of assets, since the framework contains no other variable to optimise. Yet the analytical reading also writes itself, and justly. Van Parijs addressed the actual unfreedom of workers tied to jobs they hate by the absence of an exit option, and his proposal, whatever its lineage, would alter the balance of power in every workplace more directly than many programmes that keep the classical vocabulary. The comparison declines to grade the proposal and instead exhibits it as the school’s furthest shore, the point at which a reader must decide whether the voyage was discovery or drift. The decision, predictably and by the article’s whole argument, will be made by the reader’s answer to the generative question, since discovery and drift are the same wake viewed from different theories of what the ship was for.

Does basic income belong inside a comparison of methods?

Yes, as evidence about methods rather than about policy, since the proposal’s shape, a distributive instrument derived from an asset-based framework, illustrates how a method’s primitives become a politics’ horizon.

The classical tradition’s alternative horizon, transformation of the relations themselves, is invisible from inside the distributive frame, and that invisibility, not the grant’s merits, is the comparison’s datum.

A companion figure completes the normative shore. Roemer’s later work on equality of opportunity carried analytical Marxism’s tools into the design of policies that compensate for circumstances while respecting responsibility, a framework adopted by economists and philosophers with no Marxist allegiance. The migration of tools beyond the tradition is the analytical wager’s clearest vindication and the classical warning’s clearest illustration, in a single fact. The tools travelled because they were built to general standards. They travelled away from Marxism because general standards are owned by the disciplines that set them, and a tradition that submits its tools to foreign standards will find its tools naturalised abroad and its name retired at home. Both sentences are true, and the comparison’s refusal to choose between them is not indecision but accuracy, since the choice requires a premise about what traditions are for that the evidence of the case cannot supply.

Crisis theory and the fate of economic laws

Economic crisis is the topic on which the classical tradition’s confidence was highest and the analytical school’s scepticism most consequential, and the row it forms, though absent from the grid of eight, tests the derivation once more. Classical crisis theory, in its several variants, locates a tendency toward breakdown or stagnation inside accumulation itself: the falling rate of profit from rising mechanisation, the realisation problems of departments producing means of production and consumption, the disproportionalities of unplanned investment. The theories share a form the classical method loves, a contradiction internal to the capital relation unfolding through its own development. Analytical examination separated the variants, formalised each, and reported damages: the falling rate argument, once the Okishio theorem showed that viable technical change chosen by capitalists raises rather than lowers the general profit rate under stated conditions, could not stand in its simple form, and the rescue attempts, temporal valuations, disequilibrium dynamics, divided the classical camp among themselves. Elster’s verdict, that Marx had no adequate theory of crisis, was the school’s bluntest, and even writers who rejected it accepted the discipline it imposed, that a crisis theory must specify its mechanism and its scope conditions rather than invoke contradiction as a universal solvent.

The classical counterattack on this terrain is instructive because it partly succeeded, which keeps the comparison honest. Defenders of the temporal single-system interpretation argued that the Okishio result depends on simultaneous valuation assumptions that misrepresent Marx’s own procedure, so that the analytical refutation refuted a reconstruction rather than the theory. Whatever the merits, the argument’s form deserves notice: the classical side won ground by becoming more precise about textual and formal detail, by fighting on the analytical field with analytical weapons, and the episode joins the Brenner debate as evidence that the border between the tempers is porous in practice. The analytical side, for its part, never produced a crisis theory of its own to set against the damaged classical ones, its models of instability remaining exercises rather than accounts of the historical crises the tradition had studied, a silence the classical side records as the framework’s characteristic limit: a method excellent at auditing theories of large processes and poor at generating them.

Why do crisis theories resist analytical settlement?

Because a crisis is a singular historical event in which many mechanisms interact under conditions that will not repeat, so the demand for a general mechanism with scope conditions, reasonable as methodology, may exceed what the object offers, while the classical alternative, a narrative of contradictions maturing, risks explaining every crisis by the same story and therefore explaining the timing and form of none.

The standing result on this terrain mirrors the grid’s pattern. Analytical examination permanently raised the standard a crisis theory must meet, and permanently damaged the simplest versions. Classical insistence permanently preserved the question, why a system of decentralised investment should periodically fail to reproduce itself, against a mainstream economics that had defined the question away. A reader equipped by both sides reads the literature of financial crises with two questions in hand, what mechanism failed and what in the structure made its failure systemic, and the pairing of the questions is the combination working at the level of a single topic, as it worked for ideology above.

The evidence, row by row, without the rhetoric

Strip the comparison of its polemical inheritance and ask, row by row, what a fair examiner can actually conclude, and the result is more modest and more useful than either camp’s publicity. On method, the examiner concludes that clarity is a genuine good whose pursuit produced genuine discoveries, and that frameworks encode conclusions whose audit is a genuine duty the classical side performed; neither conclusion cancels the other. On value, the examiner concludes that the labour theory fails as a general theory of price formation and survives, transformed, as an account of social form, and that both halves of that sentence must be taught together or the history is falsified. On exploitation, the examiner concludes that the concept has two legitimate objects, a distributive pattern and a production relation, that confusion between them wasted decades, and that Roemer’s service was to force the distinction even where his own preference between the objects is rejected. On class, the examiner concludes that asset-based typologies measure what they claim and that what they claim is not the whole of what the tradition meant, a conclusion Wright’s own career ratifies from inside.

On explanation, the examiner concludes for the mechanism demand as a rule of method while noting that the stock of completed mechanistic explanations of macro-historical change is small in every tradition, so the rule functions as a standard of honesty more than as a description of achievement. On normative theory, the examiner concludes that the tradition’s values were always operative and that stating them improved the arguments, while recording that the migration of talent into normative theory starved other problems, a cost visible in the school’s bibliography. On the motor of history, the examiner concludes that the evidence underdetermines the primacy question and that the best historical work uses forces and struggle as complementary variables whose weights vary by case, which is a conclusion neither camp’s theory predicts and both camps’ historians practise. On the agent, the examiner concludes that the classical designation failed empirically in its strong form and that no successor designation has succeeded, leaving the politics of transformation with an unsolved problem that both traditions bequeathed to their heirs.

What kind of conclusion is a split verdict on every row?

It is the conclusion the evidence supports, and its split character is information rather than failure, since it shows that the two currents were strongest where the other was weakest with a regularity that itself confirms the generative claim: a single fork, consistently taken, produces complementary strengths and complementary blindnesses across every topic it touches.

That regularity deserves one more statement, because it converts the comparison from a scorecard into an instrument. If the strengths and blindnesses were randomly distributed, the currents would be mere collections of opinions and the reader could shop among them. Because the distribution is systematic, generated by the fork, the reader can predict, for any fresh topic, where each current will be strong and where it will be blind, and can therefore read both literatures strategically, assigning each the questions it is built to answer and auditing each at the points its method cannot see. Strategy of that kind is what scholars actually do with rival traditions when the traditions are alive, and the comparison’s final service is to make the strategy explicit for a dispute whose participants, committed to victory, could not state it themselves. The verdict that follows is accordingly narrow. It does not crown a tradition. It identifies the single argument in the whole dispute that meets both sides’ standards for an unanswered objection, and it attaches to that identification the condition under which it would be withdrawn.

How dialectical materialism was codified, and why it matters

The classical current’s method arrived with an institutional history that shaped what the analytical school thought it was refusing. Marx left no treatise on method; Engels, in Anti-Dühring and in manuscripts published after his death as the dialectics of nature, generalised the critique of political economy into a doctrine of universal laws, the unity and struggle of opposites, the transformation of quantity into quality, the negation of the negation. Soviet codification under Stalin compressed the doctrine into a catechism, the 1938 chapter on dialectical and historical materialism in the party’s official history, memorised by generations and enforced as the philosophy of a state. Western classical writers defined themselves against that codification as much as against capitalism: Lukács’s dialectics was a rescue of Marx from the catechism, an insistence that method meant the grasp of totality by a class, not a list of laws governing acorns and empires. The full story belongs to the guide to dialectical materialism as a doctrine, and this article links it for readers who need the codification’s detail.

The codification matters to the comparison because the analytical school’s refusal of a distinctive method was aimed, in the first instance, at the catechism’s descendant doctrines and at their Althusserian university cousins, not at Lukács’s subtleties. When Elster dismissed dialectics, the dialectics in his sights were the laws that explained everything and predicted nothing, the verbal machinery by which any outcome and its opposite could both be derived after the fact. Classical defenders replied that refuting the catechism leaves the serious doctrine untouched, and that the school’s polemic gained its ease by choosing the weakest bearer of the name. Both charges are fair, and their fairness defines a rule the comparison has followed throughout: each side is represented by its strongest statements, Cohen and Roemer on one flank, Lukács and Wood on the other, with the catechisms and the caricatures named only to be set aside. A comparison conducted at the level of weakest bearers would be shorter and worthless, since any tradition can be refuted by its catechism and any method vindicated by its masterpieces.

Why did the catechism version of dialectics prove so durable?

Because it served institutions that needed a philosophy teachable to cadres, testable in examinations and applicable as a stamp of approval to decisions taken on other grounds, and a method that certifies is more useful to such institutions than a method that inquires.

Its durability is therefore evidence about organisations, not about the doctrine’s truth, a distinction both camps in the comparison accept when it suits them.

The analytical school’s own relation to Engels is a small parable of the fork. Analytical writers treated the dialectics of nature as an embarrassment to be amputated, Engels the systematiser having misled the tradition Marx the investigator founded. Classical writers divided: Colletti amputated with them, from inside; others defended Engels as the necessary complement, the theory’s ontology. The amputation question, whether Marxism includes a philosophy of nature, looks parochial and is not, since it asks whether the distinctive method, if it exists, governs all domains or only the social one. A method for society alone, historically specific categories for historically specific objects, is a far more defensible claim than a universal dialectics, and the classical tradition’s strongest modern statements quietly adopt the restriction, defending dialectics as the logic of social forms while abandoning its empire over acorns. The restriction concedes half the analytical case, the half aimed at the catechism, while fortifying the other half, the constitutive claims about social objects, on defensible ground. Watching a tradition retreat to its strongest perimeter is watching the comparison’s method work in real time, and the article notes the manoeuvre with respect, since it is the classical analogue of the analytical school’s own retreats to defensible cores under fire.

Analytical philosophy and the tribunal the school chose

The analytical current’s standards did not fall from the sky; they were the standards of postwar analytical philosophy, formed by logical empiricism’s aftermath, by ordinary language philosophy’s suspicion of grand systems, and by a philosophy of science that prized falsifiability, distinction and the piecemeal testing of claims. Cohen’s Oxford, Elster’s Paris and Oslo, Roemer’s Berkeley were variations on one intellectual climate, in which a philosopher’s reputation was made by the counterexample that felled a system and unmade by the paragraph a seminar could not parse. The school’s wager was that Marxism, submitted to that climate, would emerge leaner and stronger. Its critics’ counter-wager was that the climate itself was a local weather system, the ideology of a profession in a particular place and period, mistaking its house rules for the laws of thought. The counter-wager has a sociology of knowledge behind it that the analytical side never fully answered: why should the seminar rooms of Oxford in 1975 supply the tribunal for a theory of world history, and what in the tribunal’s own constitution guarantees its neutrality between the theories brought before it?

The school’s implicit answer was that the tribunal’s rules are not neutral but are corrigible, that distinction, evidence and counterexample have vindicated themselves across every field that adopted them, and that a theory refusing the tribunal must explain its refusal without using the immunity the refusal creates. That answer is strong where the tribunal’s rules are thin, the demand that claims be statable and evidence relevant, and weaker where the rules thicken into a specific social ontology, methodological individualism, which is a substantive theory of society wearing the tribunal’s robe. The comparison has pressed exactly that seam throughout: the analytical school’s thin rules command assent that its thick ontology must earn, and much confusion in the literature comes from the school’s defenders claiming the thin rules’ authority for the thick ontology’s conclusions, and from classical critics rejecting the thin rules because they reject the thick ontology smuggled beneath them. Sorting the school’s output by that seam, thin-rule work on one side, the clarifications, the mechanism demand, the normative candour, and thick-ontology work on the other, the models of class and exploitation among endowed agents, predicts with embarrassing accuracy which parts the classical objection touches and which it cannot reach. A student armed with the seam can read the school’s bibliography as two literatures interleaved, and can accept the first literature entire while reserving judgement on the second, which is the most defensible position the comparison has to offer a newcomer.

Is methodological individualism a rule of clarity or a theory of society?

It presents itself as the first and functions as the second, since requiring every explanation to terminate in individuals forbids, in advance, explanations in which relations, forms or structures carry irreducible causal weight, which is a claim about what society contains and not merely about how claims should be phrased.

Naming the double function is the classical objection’s most portable product.

The tribunal question also explains the asymmetry of the two literatures’ reading habits. Analytical writers read classical texts as arguments to be extracted, often with profit, since Lukács and Thompson survive extraction better than their reputations among analytical students suggest. Classical writers read analytical texts as symptoms to be interpreted, locating the class position or the professional incentive behind the argument, a reading habit that occasionally illuminates and routinely exasperates, since a symptom reading never has to meet the argument it explains. The assessment of the school as a whole, gathered in the companion article on the criticisms of analytical Marxism, catalogues both habits and their costs, and readers who want the polemical record in full should follow that link after finishing the comparison, which has deliberately subordinated polemic to structure.

The transformation problem as a controlled experiment

Intellectual disputes rarely offer controlled experiments, yet the transformation problem comes close, and the comparison uses it as one. The problem is precise: Marx’s third volume transforms values into prices of production while asserting that total price equals total value and total profit equals total surplus value, and generations of critics, from Böhm-Bawerk onward, charged that the transformation as Marx sketched it cannot sustain both equalities. Here the object is narrow, the claims are formal, and both methods must show their work. The analytical treatment, building on a long line of mathematical Marxian economics, demonstrated the difficulty rigorously and explored the conditions under which versions of the equalities survive, generally at the cost of redefining the terms. The demonstration is the analytical method’s ideal case: a fused claim separated, a formalism applied, a verdict delivered that no amount of totality talk can dissolve, because the claim itself was arithmetical.

The classical responses divide in a way that illuminates the whole comparison. One response accepts the formal field and fights on it, the temporal single-system interpretation arguing that sequential valuation, inputs valued at the prices of their purchase period, restores the equalities and matches Marx’s text. Another response changes the question, the value-form reading holding that the equalities were never the theory’s load-bearing claims, since the theory concerns the form of value and its monetary expression rather than an accounting identity between two systems of magnitude. The first response concedes the tribunal and disputes the verdict. The second disputes the tribunal’s jurisdiction over the real theory. A reader can see the generative fork operating with unusual clarity, because the two responses correspond exactly to the two things a tradition can do when a formal claim fails: repair the formalism or redescribe the claim’s place in the whole. Neither response is dishonest. They differ in what they take the theory to be, a set of propositions that must each survive, or a structured inquiry whose parts bear unequal weight. The comparison’s own judgement is that the second response, to be legitimate, must pay the analytical price: it must state, in advance of the next failure, which claims are load-bearing and which are scaffolding, since a whole whose every failed part is retroactively declared scaffolding has made itself unfalsifiable by bookkeeping.

What does the transformation problem teach about the two methods?

It teaches that the analytical method is sovereign wherever a claim is formal and separable, and that the classical method’s characteristic move, redescribing a part in light of the whole, is legitimate exactly when the redescription is principled and prospective rather than improvised after each defeat.

The problem thus grades both traditions at once, which is why teachers of the dispute assign it early.

The experiment has a second lesson, about the sociology of refutation. The transformation problem was posed by critics in the eighteen nineties, answered and re-posed in every generation since, and its persistence shows that traditions do not die of their unsolved problems; they organise around them, building careers, schools and identities on the contested ground. Analytical writers sometimes wrote as if a demonstrated inconsistency should end a research programme, and the history of their own programme, which organised itself around the difficulties of historical materialism for decades, refutes the expectation in their own case. Classical writers sometimes wrote as if persistence proved vitality, and the history of the catechism, which persisted by institutional enforcement, refutes that expectation in theirs. Persistence proves only that a problem is fertile or a institution is strong, and telling fertility from enforcement is historical work that neither method performs automatically. The comparison therefore treats longevity, its own subject’s and its traditions’, as neutral evidence, and advises students to ask of any long-lived dispute not why it has not been settled but what work its unsettledness is doing for the people who maintain it.

Historical materialism under both examinations

Cohen’s reconstruction made historical materialism the most examined theory in the comparison, and the double examination it received, analytical from inside the school, classical from outside, is a model of how the two tempers interact at their best. The analytical examination separated the development thesis, the primacy thesis and the correspondence between relations and forces, then pressed each for its mechanism and its evidence. Elster’s scepticism about the development thesis, its want of a motor comparable to selection, and Wright and Levine’s attempts to rebuild the theory on class struggle as the motor, mark an internal debate of high quality in which the theory’s defenders and critics shared premises and texts. The classical examination objected to the reconstruction’s shape: a theory of history centred on technology and its effects had, in the objectors’ view, displaced the theory’s actual centre, the historically specific relations through which any technology is developed, adopted or suppressed, and had converted a critique of political economy into a philosophy of history of exactly the kind Marx’s materialism was supposed to dissolve.

The Brenner debate enters here as the classical examination’s evidence, since Brenner’s cases purport to show relations and struggles deciding outcomes that a forces-first theory predicts wrongly. Cohen’s defenders replied that the primacy thesis operates at a level of abstraction where Brenner’s cases do not touch it, forces explaining the long-run menu within which struggles choose. The reply’s level-of-abstraction move is the analytical analogue of the classical whole move, a redescription that protects a claim by relocating it, and the comparison notes the symmetry without embarrassment: both traditions protect their cores by adjusting the relation between core and case, and the honest question in each instance is whether the adjustment was principled in advance or improvised under fire. On the development thesis, the adjustment history is mixed, and the school’s own later doubts, Cohen’s included, suggest that the thesis in its strong form did not survive the examination its author designed for it. What survived is a deflated historical materialism, a standing instruction to look for the material and class conditions of any social order before its ideas, which functions in good historical work of every allegiance as a heuristic of proven fertility. The pattern of the value row repeats: a doctrine divides under examination into a strong claim that weakens and a discipline of attention that persists, and the traditions divide over whether the persistence redeems the weakening.

Which thesis of historical materialism fared worst under examination?

The development thesis in its strong form, the claim that productive forces tend autonomously to grow and to select the relations that house them, since it lacked both a mechanism that survived Elster’s scrutiny and a record that survived Brenner’s cases.

The primacy of material conditions as a research instruction fared best, being the form in which the theory entered general historical practice.

A coda on this section concerns the uses of defeat. The analytical school’s treatment of historical materialism is sometimes summarised as a demolition, yet Cohen’s book remains the gateway through which students enter the theory sympathetically, assigned by teachers who reject its conclusions because no rival states the theory well enough to reject. A reconstruction so clear that its refutation becomes possible is a gift to the tradition it wounds, and the classical tradition’s failure to produce its own statement of comparable clarity is the standing reproach the analytical episode leaves behind. Conversely, the classical examination’s insistence that the reconstruction had changed the theory’s centre has entered the common sense of the field, so that Cohen is read through Brenner as often as against him. Each tradition thus curates the other’s best work for purposes the other did not intend, a cross-adoption that suggests the dispute’s real legacy is a shared library rather than a victor’s canon.

Standpoint, feminism and the fork outside Marxism

The generative disagreement reappears wherever a subordinated group’s experience is claimed as a source of knowledge, and socialist feminism fought the Marxist version of the quarrel under its own flag. Standpoint theory, developed by Nancy Hartsock explicitly on the model of Lukács, argued that the sexual division of labour gives women an epistemic position from which the abstractions of male-dominated thought, its separations of mental and manual, public and private, subject and object, become visible as the ideology of a standpoint rather than the structure of the world. The argument is the classical method transposed: a social position constitutes a possible knowledge, the whole is grasped from a part that the whole oppresses, and the categories of the dominant inquiry are indicted as partial in their very form. Analytical feminists and analytical Marxists replied in the school’s idiom: state the claims the standpoint allegedly reveals, and let them be tested by any competent inquirer, since a truth available only from a position cannot be checked by those outside it and therefore cannot be taught, only professed.

The exchange matters to this comparison for two reasons. It shows the fork operating independently of its Marxist birthplace, confirming that the generative disagreement names a general structure in the theory of knowledge rather than a local feud. And it produced, in the feminist literature, some of the most sophisticated attempts at the combination this article recommends, since standpoint theorists under analytical pressure learned to state their claims as hypotheses about what differently situated inquirers systematically miss, a form in which the claims become testable without surrendering their point, that social position organises attention and evidence before argument begins. The analytical side, in the same exchange, learned to treat the sociology of its own tribunal as a legitimate question rather than an insult. When two traditions educate each other under pressure, the dispute has done the work disputes are for, and the Marxist case is best read as the largest and best documented instance of a general education that several fields underwent in the same decades.

What does the feminist parallel prove about the Marxist dispute?

It proves that the fork is not caused by anything peculiar to Marx’s texts, since it opened identically around a different tradition’s claims, and it suggests that the fork’s persistence across fields reflects a real tension in social knowledge between the situated character of inquiry and the general character of validity, a tension no field has dissolved.

Ecology supplies a second parallel worth a paragraph. Marxist ecology divides between analytical treatments, which model resource use, discounting and collective action with the school’s instruments, and classical treatments, which read the metabolic relation between society and nature through the categories of form and totality, finding in capital’s drive to accumulate a structural rift in the metabolism itself. The analytical work clarifies policy and exposes lazy catastrophism. The classical work preserves the claim that the rift is constitutive, that a system requiring endless accumulation on a finite planet has its crisis written into its form rather than into its managers’ choices. The modelling objection applies to the first literature in its familiar shape, since models with given preferences and discount rates build the contested temporality of capital into their arithmetic, and the clarity defence applies to the second, since a rift asserted without mechanisms is a metaphor in search of a research programme. A reader who has followed the comparison this far can generate both critiques unaided, which is the test of whether the article’s central claim, that one disagreement generates the rest, has been understood rather than memorised.

How each current reads a text

Methods show themselves in reading habits, and the two currents read Marx differently enough that their students effectively study different books. The analytical reading is extractive and reconstructive: it asks what claim a passage makes, whether the claim is true, and how the argument looks with its premises numbered, treating the text as a colleague’s draft to be improved in the margin. Cohen’s seminars were famous for the treatment, in which a page of Capital might occupy a term, every sentence audited. The classical reading is symptomatic and architectural: it asks what problem a passage solves in the economy of the whole work, what its categories presuppose, where the presentation inverts or doubles back, treating the text as a structure whose meaning lives in its articulations. The value-form scholarship is this reading at its most developed, reconstructing the derivation of money from the commodity as a logical presentation whose order is itself an argument.

Each reading habit has a characteristic blindness that the other names correctly. Extraction loses the architecture, and with it the possibility that a passage’s function is presentational rather than assertoric, so that a claim extracted and refuted may never have been asserted in the extracted form. Architecture loses the audit, and with it the possibility that a passage is simply confused, its place in the whole a rationalisation constructed by charitable readers across generations. Marx’s text, which contains both assertoric argument and dialectical presentation inextricably mixed, punishes each habit in turn, and the mature scholarship of recent decades, whatever its allegiance, tends to combine them, reconstructing arguments while respecting presentation, a combination at the level of reading that mirrors the combination this article recommends at the level of theory. Students should therefore distrust any course that teaches only one way of reading the founding texts, since the dispute between the currents is, among other things, a dispute between two literacies, and literacy in both is the closest thing to a settlement the comparison can imagine being taught.

Which reading habit should a beginner acquire first?

The analytical habit, in this article’s judgement, since extraction and audit are prerequisites for knowing what a text claims before its architecture can be responsibly described, and since the classical habit acquired first tends to immunise confusions as depths.

The judgement is pedagogical rather than philosophical, and classical teachers who reverse it can point to beginners who extracted claims Marx never made.

The reading dispute also governs the two currents’ opposed attitudes to Marx’s authority. For the analytical school, Marx is an author, and authorship confers no authority beyond the arguments; a claim of Marx’s stands exactly as tall as its evidence, and the school’s willingness to report that the master was wrong, in print and without anguish, was its most scandalous break with the tradition’s manners. For the classical tradition, Marx is the founder of a method, and the method’s authority is of a different kind, closer to a discipline’s founder in the sciences, whose texts are read as the record of a discovery procedure that later work extends rather than audits sentence by sentence. The difference in manners encodes the generative disagreement once more: if the theory is its claims, the founder is an author; if the theory is its method, the founder is a legislator of inquiry, and disrespect toward the text is a different offence in the two courts. Readers who understand why each court finds the other’s manners scandalous have understood the comparison at the level of practice, which is where methods finally live.

Hegemony and the classical strength in cultural analysis

The classical tradition’s most fertile export, hegemony, deserves a fuller treatment in a comparison that has stressed the analytical virtues, because hegemony is the concept where the classical method’s superiority for certain objects is hardest to deny. Gramsci’s insight, that a dominant class rules substantially by organising the common sense of the society, its sense of what is natural, possible and respectable, required for its development exactly the instruments the classical method supplies: historical narrative, attention to institutions of civil society, schools, churches, press, associations, and a conception of consciousness as formed within relations rather than chosen by agents. The research programme that descended from the insight, through cultural studies and the history of popular culture, produced findings about how consent is manufactured and contested that no asset distribution or preference schedule could have generated, because the object, a society’s common sense, exists only at the level of the whole the classical method is built to grasp.

Analytical treatments of the same territory exist and prove the rule by their shape. Consent can be modelled as an equilibrium of beliefs, ideology as motivated reasoning, common sense as a distribution of preferences with media effects as shifters, and each model captures a mechanism the classical narratives leave impressionistic. What the models cannot capture is the content’s historical specificity, why this common sense, with these metaphors and these silences, belongs to this social order and would be unintelligible in another, since the models’ variables are deliberately content-free. The classical narrative, conversely, captures the content and waves at the mechanism, attributing consent’s stability to hegemony in a functional grammar that Elster’s critique permanently taxed. The two literatures on consent thus stand as the comparison’s cleanest exhibit of complementarity, each holding half of an explanation whose halves are shaped exactly as the generative fork predicts, mechanism without content on one shelf, content without mechanism on the other, and the reader’s task is the assembly neither shelf performs.

Why is common sense a difficult object for formal modelling?

Because its elements are meanings, whose identity depends on histories and associations that a model’s variables must strip away to become variables, so the formalisation preserves the distribution of assent while discarding the thing assented to.

A model can count how many hold a belief; the classical question is what the belief, in its specific imagery and omissions, does to the holders’ perception of their relations.

The hegemony case also disciplines the classical side’s pride. A concept that explains consent wherever consent appears risks explaining nothing, since stability confirms hegemony and revolt confirms its crisis, and the tradition’s weaker work used the concept as exactly that universal key. Analytical pressure, applied by critics of cultural studies from inside and outside Marxism, forced the concept’s serious users to specify mechanisms of transmission and sites of contestation, after which hegemony research became measurably better history. The pattern is by this point familiar enough to state as a law of the comparison: analytical pressure improves classical work by taxing its functionalism, classical pressure improves analytical work by taxing its formalism, and the improvements occur despite the traditions’ official positions, through the osmotic contact of scholars who read across the shelf division their institutions maintain. The article’s recommendation of combination is therefore not a utopian proposal but a description of the field’s actual best practice, formalised into advice.

Przeworski and the modelling of political choice

Adam Przeworski’s work supplies the analytical school’s most politically serious modelling, and the comparison examines it because it answers, partially, the charge that models cannot touch real politics. His analysis of social democratic parties modelled the electoral dilemma with austere clarity: workers are never a numerical majority of the electorate in the required sense, so a party of the working class must choose between class purity with permanent minority status and broad appeals that dilute the class programme and demobilise the base. The model organised a century of European political history into an intelligible pattern and generated the bleak conclusion that participation in electoral politics transforms socialist parties by a logic internal to the choice situation, independent of leaders’ intentions or betrayals. Classical writers, whose tradition explained the same transformations by betrayal and bureaucratisation, were obliged to concede the structural point even while contesting its completeness, since a dilemma that operates regardless of intentions is exactly the kind of claim their own structuralism should have taught them to respect.

The residue, predictably, is constitution. The model takes the identities, worker, middle class voter, as given by occupation and asks what parties should do; the classical analysis asks how those identities are themselves formed and reformed by parties, unions, press and struggle, so that the choice situation the model solves is partly the product of previous rounds of the very politics being modelled. Przeworski, a modeller of unusual historical learning, knew the residue intimately and wrote narrative history alongside his models, a double practice that again instantiates the combination in a single career. His later work on democracy and development, statistical and comparative, travelled furthest of all the school’s output into the mainstream of political science, where its Marxist provenance is a footnote. The comparison draws its by this stage standard double moral: the analytical instrument, honestly used, produced knowledge the classical tradition needed and partly absorbed; the instrument’s triumphs occurred at the price of the questions it could not ask, which the classical tradition kept in trust, mostly unread by the audiences the instrument won.

What does the electoral dilemma model leave unmodelled?

The formation of the electorate’s identities and the transformation of the parties’ goals across repeated rounds of the game, since a party that dilutes its programme to win alters what its voters want and what winning means, so the players and the prizes are products of the play.

Classical theory begins exactly there, which is why the two accounts of social democracy read as descriptions of different objects.

The school’s political science legacy extends beyond Przeworski. Analytical class analysis entered mainstream stratification research, where Wright’s schema competes with neo-Weberian schemes in the journals, judged by predictive performance in a forum the classical tradition disdains and cannot ignore, since the forum’s verdicts shape what the next generation of sociologists takes class to be. Analytical Marxism’s concepts also entered development studies and the theory of the state through scholars who absorbed the school’s standards without its label. Dispersal, the fifth scene of the quarrel’s history, is thus not a euphemism for death. It is the normal fate of a successful research programme in a decentralised academy, absorption into the common stock, and the classical tradition’s contrasting fate, preservation of a named orthodoxy in enclaves, is what ordinarily happens to a programme whose claims resist absorption because they are constitutive rather than portable. Each fate has its dignity and its pathos, and the comparison, having traced both, is ready to deliver the narrow verdict it has been constructing since the grid.

The state debate as prehistory: Miliband, Poulantzas and the fork before the fork

A decade before the analytical school formed, the Miliband and Poulantzas debate over the capitalist state rehearsed the generative disagreement in a single arena, and reading it as prehistory sharpens the comparison’s claim. Ralph Miliband’s The State in Capitalist Society, published in 1969, argued empirically: the state serves capital because its personnel are drawn from, educated with and interlocked with the capitalist class, and because business dominates the environment of pressure within which governments act. The argument proceeded by evidence about persons and institutions, auditable item by item. Nicos Poulantzas replied that the argument’s very form betrayed the theory: a Marxist account of the state must begin from the state’s structural place in the reproduction of capital relations, its relative autonomy inscribed in its institutional separation from production, and personnel are the last place to look, since the structure would produce capitalist outcomes through saints or through strangers. Miliband countered that a structure which operates regardless of persons explains everything and nothing, and demanded the mechanisms by which structural imperatives reach decisions.

The exchange maps onto the later quarrel with uncanny completeness. Miliband’s demand for mechanisms anticipates Elster’s by a decade, from inside a classical formation and a historian’s practice. Poulantzas’s insistence on form and relative autonomy anticipates the constitutive objection, that the elements, here ministers and officials, are constituted by the relations the analysis must start from. Neither man belonged to the later camps, Miliband a classical socialist historian, Poulantzas an Althusserian structuralist, yet the fork opened between them along the same line, because the line is not a property of the analytical school’s invention but a permanent fault in Marxist theorising that the school later named and occupied. The comparison draws a moral about priority: the analytical school did not create the demand for mechanisms or the suspicion of structure talk; it inherited a quarrel already two generations old, gave one side of it a philosophy department’s artillery, and thereby changed its balance without changing its terms. Classical writers who date the corruption from 1979 misdate it; the corruption, if it is one, is coeval with the tradition’s first attempts to say how structures act.

What did the state debate settle before the school existed?

It settled that the two explanatory styles could each expose the other’s characteristic evasion, instrumentalism’s blindness to form and structuralism’s immunity to evidence, without either style eliminating the other, since the state’s operation genuinely involves both the persons who staff it and the constraints that survive every change of staff.

Later state theory that endures, on all sides, is written in awareness of both exposures.

The debate also supplies the comparison’s favourite pedagogical exercise, because its texts are short, its stakes are clear and its mapping onto the grid is exact enough for students to perform unaided. Assign the exchange before the grid, and students discover the fork in the wild; assign the grid first, and they merely apply it. The difference between discovery and application is the difference between the two pedagogies the currents themselves favour, the classical preference for encountering a totality in a concrete case and the analytical preference for stating the principle before the instances, so that even the teaching of the dispute re-enacts it. A teacher aware of the re-enactment can alternate the order across a course and let the alternation itself become a topic, which is the closest a classroom can come to neutrality between methods whose dispute is partly about what neutrality in teaching would mean.

Game theory examined as the school’s emblematic instrument

No instrument carries more symbolic freight in the comparison than game theory, and a section that examines the instrument itself, rather than its applications, clarifies what the modelling objection does and does not claim. Game theory studies strategic interaction: agents with preferences over outcomes choose actions in light of others’ expected choices, and the analysis finds equilibria, configurations from which no agent benefits by departing alone. Its attractions for the analytical school were genuine and multiple. Class conflict, coalition formation, collective action and the dilemmas of socialist parties all have strategic shapes that prose handles loosely and games handle exactly. Olson’s logic of collective action, which game-theoretic reasoning formalised, posed the free rider problem that any theory of class agency must answer: why should a worker bear the costs of organisation when the benefits accrue to workers generally? The classical tradition had answered with solidarity, consciousness and organisation as historical products. The game-theoretic answer began from the problem’s structure and asked what institutions, selective incentives, repeated interaction, enforcement, could solve it, converting a mystery of consciousness into a design problem with solutions that history could be checked against.

The classical objection to the instrument is the general objection at its most specific, and it has three layers that should be distinguished because they have different strengths. The first layer is the givenness of the players: a game begins with agents whose identities and preferences are fixed for its duration, while class politics is substantially about the formation and transformation of the players, workers becoming a class, parties becoming machines, identities dissolving and reforming across the play itself. The second layer is the givenness of the game: the available actions and their payoffs are specified by the analyst, while political struggle is characteristically a struggle over the game, over what actions exist, what counts as winning and who may play, so that the most important moves are those that rewrite the matrix the analyst must hold still. The third layer is the equilibrium concept itself: a theory that explains outcomes as equilibria has a built-in bias toward stability, toward showing why things stay, and must import its account of change from outside, from shocks and parameter shifts, whereas the classical theory wants change generated from inside the form. Each layer is a version of the constitutive claim, and each has a partial analytical answer: players can be modelled across stages, games can be nested inside larger games, equilibria can be multiple and fragile. The answers share the fate described in the fetishism section, they stage constitution as a prior phase, and the staging is often illuminating and never complete.

Does the free rider problem refute the classical theory of class agency?

It refutes the version in which class position generates collective action automatically, a version serious classical writers had already abandoned, and it disciplines the surviving version by demanding the mechanisms, organisation, solidarity norms, selective benefits, through which position is converted into action.

The classical theory that emerges from the discipline, Thompson’s and Brenner’s included, is stronger than the one that entered it.

Fairness to the instrument requires a closing observation that classical polemic rarely grants. Game theory’s practitioners inside the school were its own best critics on precisely the constitutive points: Przeworski wrote history beside his games, Elster catalogued the irrationalities his models excluded, Roemer flagged the abstractions his theorems required. The image of modellers innocent of their instruments’ limits is a polemical convenience. What the school lacked was not awareness but a forum in which the awareness could become a research programme rather than a caveat, since a caveat acknowledged and then set aside, paper after paper, functions in the literature exactly like an unanswered objection. That is the institutional form of the article’s verdict, and it is stated here in the instrument’s section because the pattern is clearest where the instrument is most admired: the school saw the objection, understood it, and organised its work so that the objection never had to be met, which is a more interesting failure than blindness and a more instructive one for readers who organise inquiries of their own.

Archives against models: the two evidentiary imaginations

Beneath the dispute about method sits a dispute about evidence, and surfacing it explains several otherwise puzzling features of the literature, including why the two sides so rarely cite the same exhibits. The classical evidentiary imagination is archival and narrative: evidence is the document, the memoir, the record of a meeting, the sequence of a struggle reconstructed until the actors’ categories become intelligible, and a claim is supported when the reconstruction shows the process at work in a case thick enough to bear it. Thompson’s histories are the imagination’s masterpieces, and their standard of proof is saturation, the point at which further documents stop altering the picture. The analytical evidentiary imagination is comparative and formal: evidence is the variation across cases, the estimate that survives controls, the model whose implications match a pattern no rival model predicts, and a claim is supported when it survives attempts to make it fail. Wright’s surveys are the masterpieces of this second imagination, and their test is exposure, the degree to which the claim was placed at risk.

Each imagination has a characteristic fraud it cannot detect. Saturation cannot detect the case that the archive was built to hide, the voices unrecorded because the record keepers shared the relations under study, and classical historiography’s confidence in its documents has repeatedly been corrected by the recovery of subaltern archives its method had no instruction to seek. Exposure cannot detect the variable that the dataset was built without, the relation or meaning that no questionnaire item operationalised, and analytical social science’s confidence in its estimates has repeatedly been corrected by ethnography showing that the measured attitude and the lived relation were different objects sharing a name. The two frauds are the constitutive objection and the mechanism objection translated into the philosophy of evidence, which is further confirmation, if the comparison still needs it, that one fork generates the dispute at every level at which it can be posed. A mature researcher in either tradition learns the other’s fraud as a checklist item, and the learning, again, happens through osmotic contact rather than through official conversion.

Why do the two sides cite different exhibits in the same argument?

Because an exhibit is evidence only inside an imagination that specifies what would count against the claim, and the imaginations specify differently, saturation against exposure, so each side experiences the other’s best exhibits as illustrations rather than as tests.

Agreeing on what would count as a test is therefore prior to agreeing on any result, and it is the agreement the generative disagreement prevents.

The evidentiary divide also governs the traditions’ opposed relations to quantification, which students often mistake for the whole dispute. Classical writers are not innumerate; the tradition’s economists and historians count constantly, and Thompson’s pages are dense with wages and prices. Analytical writers are not hostile to narrative; Cohen’s examples and Elster’s cases are miniature narratives chosen with a storyteller’s care. The divide is over what counting and narrating are for. In the analytical imagination, a number is a claim’s exposure to the world, valuable in proportion to the risk it runs. In the classical imagination, a number is a document like any other, produced by relations, valuable in proportion to the criticism of its production that accompanies it, so that a wage series is read the way a memoir is read, for what its categories conceal as much as for what its figures report. Neither relation to number is a mistake. Each becomes a mistake when it forgets the other, the estimate read without source criticism, the document read without aggregation, and the field’s best work, in both traditions, is marked by the double reading performed as a habit.

The dispute’s bequest to readers of other disputes

The comparison closes its substantive work by generalising its instrument one final time, because the reader who has followed the argument this far owns something transferable and should be told its name. School disputes in every field present themselves as lists of conclusions, and are almost always generated by a prior disagreement about method, evidence or the constitution of the object. The reader’s skill, trained on this case, is excavation: refuse the list, find the fork, derive the list from the fork, and test the derivation against fresh topics the disputants have not addressed. Where the derivation succeeds, the dispute becomes intelligible as a structure, its moves predictable, its strongest arguments locatable, its verdicts gradable row by row. Where the derivation fails, the failure is itself a finding, that the school is a coalition rather than a system, held together by institutions or enmities rather than by a premise, and coalitions require a different analysis, sociological rather than logical. Either way the excavation outperforms the list, and the performance is the article’s final argument for its central claim, offered not as doctrine but as a tool the reader can test on the next dispute encountered, in this field or any other.

Excavation has its own temptations, which honesty requires naming before the verdict section. A derived structure is satisfying, and satisfaction is a solvent of doubt: the excavator may force recalcitrant material into the fork, treating every anomaly as a disguised confirmation, which is the classical sin of totality committed by the comparison itself. The article has guarded against the sin in the only way available, by exhibiting the material that strains its claim, Colletti’s internal dissent, Thompson and Elster’s alliance against Althusser, Geras’s result that neither programme predicted, and by scoring rows as divided or underdetermined wherever the evidence refused a cleaner entry. A reader who finds further strain is invited to weigh it against the derivation rather than to excuse it, since the claim of this article, like the verdict it delivers, is offered with its defeater attached: a contrast between the currents which the fork cannot generate, established in a case the article has not pre-digested, would damage the generative claim exactly as the specified model would damage the verdict on the modelling objection. Symmetry of that kind, the examiner submitting to the examination, is the comparison’s last lesson in method, and it is drawn equally from the analytical school’s candour and from the classical tradition’s insistence that no standpoint, including the comparer’s, stands outside its object.

Rational choice Marxism as a label and a provocation

The school’s critics fastened on one wing of it, rational choice Marxism, and the label’s career deserves attention because it shaped how the comparison was received by readers who never opened the books. The phrase named the work of Elster, Roemer and Przeworski insofar as it used the theory of rational action and its game-theoretic extensions, and it was popularised as much by opponents as by practitioners, since it compressed the modelling objection into a slogan: Marxism rebuilt on the bourgeois science of choice. Defenders objected that the label confused a wing with the whole, Cohen’s philosophy and Wright’s sociology being Marxist and analytical without being rational choice theory in any strict sense, and that the slogan let critics attack the school’s most exposed position while ignoring its centre. The objection is just, and the comparison has accordingly treated the modelling objection as an objection to a programme within the school rather than to the school entire, a precision the polemical literature rarely observes.

Yet the label stuck for a reason the defenders understate. Rational choice work was the school’s most visible product in the social sciences, the part that neighbouring disciplines cited, taught and absorbed, and a school’s public identity is set by its exports rather than by its manifestos. The exports carried the individualist ontology into every seminar that adopted them, so that the classical objection, aimed at the exports, reached the school at the point of its widest influence. There is a general lesson here about how traditions are judged: by their travelling parts. The classical tradition is likewise judged abroad by its travelling parts, the vocabulary of hegemony and ideology in cultural studies, rather than by its most careful historical work, and each tradition experiences the other’s judgement-by-export as a slander while practising it daily. The comparison’s discipline of citing texts and dates is the only known antidote, and it is labour-intensive, which is why slander by export remains the literature’s default mode and why review essays in this field should be read with the suspicion usually reserved for advertising.

Why did the school’s exports travel further than its name?

Because exports built to general standards are detachable from their origins by design, so the better the analytical work succeeded on its own terms, the less it needed the Marxist label that had sheltered its making.

Detachability is the analytical wager’s triumph and its tragedy in a single property, and the classical tradition’s non-detachable exports exhibit the opposite pairing of virtues and costs.

The label’s history also illuminates the school’s internal diversity, which the grid’s rows flatten and which a responsible comparison must restore before the verdict hardens into caricature. Between Elster’s austere individualism and Wright’s empirically saturated class analysis there is a distance as great as between either and Thompson; between Cohen’s Kantian egalitarianism and Roemer’s market socialism there is a policy gulf; between Van Parijs’s basic income and Przeworski’s electoral pessimism there is a temperamental one. The school was a conversation, not a creed, and its members’ disagreements, Cohen against Elster on functional explanation, Wright against Roemer on the weight of domination, Elster against everyone on dialectics, were conducted with a ferocity that outsiders mistook for unity because the ferocity was grammatical, a shared insistence on statement and counterexample, rather than doctrinal. The classical tradition was likewise a conversation misread as a creed, its internal distances, Thompson against Althusser, Colletti against the dialectics of nature, Wood against the cultural turn’s excesses, as great as its distance from the school. The grid survives these restorations because it compares positions on questions, and positions can be shared, split and migrated among persons without the questions losing their shape.

The philosophy of science backdrop: paradigms, programmes and falsification

The comparison has used the vocabulary of refutation and research programmes throughout, and a short section on the philosophy of science that supplies the vocabulary will steady the verdict’s foundations. Karl Popper’s falsificationism gave the analytical temper its simplest charter: a theory is scientific insofar as it forbids observations, and a tradition that protects its theories from forbidding anything has left science for metaphysics. The charter’s application to Marxism was an old story by the time the school formed, Popper himself having indicted historicism by name, and the school’s originality lay in accepting the indictment’s standards while contesting its verdicts piecemeal, rebuilding the theories that could be rebuilt to forbid something definite. Imre Lakatos softened the charter into methodology: research programmes are judged not by single refutations but by their trajectory, progressive when they predict novel facts, degenerating when they only accommodate them, a standard this series applied to Marxism itself in the article on whether the theory is a science. Thomas Kuhn complicated both charters with paradigms, communities whose standards are internal and whose revolutions are conversions as much as arguments, a picture the classical tradition found congenial since it licensed the thought that the analytical tribunal was one paradigm’s court judging another’s.

The verdict of this article was constructed to survive translation among all three charters, and the construction is worth making explicit. In Popperian terms, the modelling objection charges that the school’s models are protected by their assumptions from the observations that would test the contested claims. In Lakatosian terms, it charges that the modelling programme accommodated the classical critique by successive adjustments that predicted no novel fact about class or exploitation, while the programme’s defenders can fairly reply that other branches, class analysis and the theory of electoral dilemmas, did predict and did find. In Kuhnian terms, it records that the two paradigms never shared the standards by which the objection could be jointly graded, so its unanswered status is a fact about the literature’s institutions as much as about its arguments. A verdict that holds in three vocabularies is sturdier than one that needs its home vocabulary to be true, and the sturdiness is deliberate, since the article’s own method, steelmanning across frameworks, is the combination it recommends applied to its own foundations. Readers trained in any of the three charters can therefore use the verdict without buying the others, which is the most a comparative judgement in a method dispute can reasonably offer.

Which charter grades the classical tradition most fairly?

The Lakatosian, since the tradition’s defenders can point to novel content generated across its career, the value-form reconstruction, the Brenner findings, the hegemony research programme, while its critics can point to accommodations that merely restated the core in safer vocabulary, and the argument between those inventories is at least an argument both sides recognise as the right one to have.

A closing observation on the backdrop concerns the word science itself, which both traditions used as an honour and a weapon. The analytical school claimed science for its standards and denied the title to dialectical claims that forbade nothing. The classical tradition claimed science for its object, the laws of motion of a social form, and denied the title to a method that mistook the object’s surface for its structure. The word’s honour value corrupted both claims, since each side’s real achievement, rigour in the one case and depth in the other, is describable without it. The comparison has accordingly used the word sparingly and descriptively, and advises students to translate any sentence in this literature containing it into the specific virtue being claimed, testability or structural fidelity, before deciding whether the sentence is true. Translation of honours into virtues is analytical work in the service of classical suspicions, and it can serve as the reader’s pocket emblem of the combination this article has recommended from its opening sections to its verdict.

Two classrooms: how the dispute is taught where it is taught well

Teaching is where methods reproduce, and the comparison gains a final concrete dimension by asking how the dispute is actually transmitted in the two traditions’ classrooms. The analytical classroom proceeds by the problem set and the counterexample. Students receive a claim, Cohen’s primacy thesis or Roemer’s definition, in a form clean enough to attack, and the seminar’s honour goes to the objection that forces a distinction. Texts are read in extracts, chosen for the argument they contain, and the history of the dispute is taught as a sequence of positions, each with its refutation conditions attached. The classical classroom proceeds by the text and the context. Students receive a chapter, of Capital or of History and Class Consciousness, and read it whole, slowly, with its moment, 1848 or 1923, reconstructed around it, and the honours go to the reading that shows a familiar passage doing unfamiliar work. Arguments are extracted only after the architecture is inhabited, on the principle that a claim torn from its presentation is a claim replaced.

Each classroom manufactures the other’s characteristic complaints about its graduates. Analytical graduates, classical teachers report, can dismantle any position and inhabit none, mistaking the audit of a theory for knowledge of its object and graduating with a toolkit in search of a subject. Classical graduates, analytical teachers report, can inhabit any text and test none, mistaking fluency in a tradition’s idiom for knowledge of the world and graduating with a subject in search of a test. The complaints are caricatures with the usual core of truth, and the institutions that teach the dispute well, a minority in both traditions, are recognisable by their theft from the other classroom: the analytical seminar that assigns a whole book and waits, the classical seminar that sets a counterexample as homework. A student choosing where to study the dispute should therefore inspect syllabuses for theft rather than for allegiance, since theft is the observable sign that a classroom’s method has remained a method, a way of finding out, rather than hardening into the catechism both traditions, at their foundings, existed to oppose.

What single exercise teaches the whole dispute fastest?

Requiring each student to state the modelling objection to the satisfaction of a classical examiner and the mechanism demand to the satisfaction of an analytical one, in consecutive weeks, since the double ventriloquism forces the discovery that each position is a discipline of attention rather than a list of opinions, and students who can perform both rarely relapse into the jargon misreading.

The classrooms also transmit opposed relations to the founder’s texts that outlast any particular syllabus. Analytical training produces readers who treat Marx as a colleague, which is a form of respect the tradition’s elders experienced as insolence and its students experience as oxygen. Classical training, by contrast, forms readers for whom Marx stands as a classic, which is a form of respect the analytical students experience as embalming and its elders experience as the minimum decency owed to a discovery. The comparison has argued throughout that the two respects encode the generative disagreement, claims against method, colleague against classic, and it notes in closing the pedagogical corollary: a curriculum that wants both respects must sequence them, colleague first for the audit, classic second for the architecture, or the reverse with the risks the reading section described, and must tell the students that the sequencing is itself a position in the dispute they are studying. Reflexivity of that order is exhausting to teach and impossible to fake, which is why the dispute, handled honestly, remains one of the best instruments in the humanities for teaching what a method is.

The long shelf: reference points for the reader who continues

A comparison article earns its keep partly by the reading it makes possible afterward, and the shelf below is organised by the dispute’s structure rather than by chronology, in prose rather than in a list, so that each recommendation carries its reason. Begin with the generator: Cohen’s Karl Marx’s Theory of History for the analytical construction at its strongest, read beside the essays in History and Theory that record its author’s later doubts, and Lukács’s History and Class Consciousness for the classical construction, read in the edition whose preface records its author’s self-criticism under party pressure, since the preface is itself a document in the politics of method. Continue with the objection and its targets together: Roemer’s 1982 reconstruction of exploitation and class for the modelling programme at its most accomplished, Wood’s The Retreat from Class for the objection’s principal statement and its political sequel, and Elster’s Making Sense of Marx for the mechanism demand applied without mercy to the whole corpus. Add Wright’s Classes for the empirical wager and Thompson’s great narrative of English working class formation for the relational alternative in narrative form, Brenner’s essays on the transition for the meeting over evidence, and Geras’s Marx and Human Nature for the result neither programme predicted. A reader who works through that shelf in that order will have re-performed the comparison with the primary materials, which is the only sense in which an article like this one can be verified.

The shelf has a second storey for readers whose interest is the general question, whether a tradition may claim a distinctive logic, and it is stocked by analogy rather than by allegiance. The literature on psychoanalysis and its critics supplies the closest parallel case, with the same fork between a method answerable only internally and a tribunal demanding external checks. The hermeneutics debates in anthropology and theology supply others. Reading across the analogies teaches the transferable skill this article has pressed throughout, the excavation of a generative disagreement beneath a list of conclusions, and it protects the reader against the provincialism of supposing that Marxists invented either rigour or totality. The Marxist case remains the best documented because its participants wrote their methodological commitments into their substantive books instead of confining them to prefaces, a habit for which both currents deserve the gratitude of every later field that has had to excavate its own fork from less obliging materials.

Which book on the shelf should a sceptic of both currents read first?

Geras’s study of Marx and human nature, since it demonstrates on a single question that the analytical audit and the classical archive can produce a joint finding neither camp expected, and a sceptic persuaded that the combination can work on one question will read the rest of the shelf with the patience the dispute requires and rarely receives.

With the shelf described, the comparison’s remaining obligations are its verdict, already delivered with its defeater in the section the reader will find after these teaching sections in the article’s order of argument, and its persona guidance, which follows the verdict there. Readers who have arrived at this point by the long road, through the grid, the objection, the ledgers and the deepening sections, hold the dispute as its participants held it, as a structure of pressures rather than a museum of positions, and the article’s work is done if they also hold its double moral without flinching: that clarity is a virtue with a price, that the price is sometimes the object itself, and that knowing the price list, in both currencies, is what understanding a method dispute means.

An external shock: the end of the socialist states and both currents

Between 1989 and 1991 the state systems that claimed Marxist legitimacy in Eastern Europe and the Soviet Union collapsed, and the collapse struck the two currents asymmetrically in ways the comparison should record, since each side’s response exhibited its method under stress. The classical tradition had long distinguished its theory from those states, through the state capitalist and degenerated workers’ state controversies and through the Western Marxist refusal of the catechism, yet the distinction, however sincere, could not fully insulate a tradition whose categories had been the official philosophy of the fallen regimes in vulgarised form. Classical writers responded characteristically: with historical accounts of the collapse that stressed the regimes’ distance from the theory’s emancipatory content, and with a renewed turn to the founding texts, as if the tradition’s survival required returning to the source before the codifications. The response preserved the doctrine by periodising the disaster, a move whose legitimacy the tradition’s own standards must judge, since a theory of history that explains every defeat as a betrayal of the theory has purchased survival at the price the analytical school always named.

The analytical school, by contrast, had largely completed its migration away from the doctrine before the shock arrived, and its members could greet the collapse with the equanimity of scholars whose positions no longer depended on the fallen states in any way. Cohen wrote of the socialist project requiring redesign rather than burial, and turned to the normative questions, what equality demands and what feasibility permits, with the collapse treated as evidence about design constraints. Roemer’s market socialism proposals acquired a new audience precisely because they conceded the information and incentive problems the collapsed systems had exhibited. The equanimity was genuine and it was also, as classical critics noted with some justice, the equanimity of a tradition that had already surrendered the claims the collapse would have tested. A school cannot be refuted by an event its theories no longer describe, and whether that immunity counts as wisdom or as evasion is, predictably, the generative disagreement applied to historiography: the analytical reader calls it having learned in time, the classical reader calls it having left the field before the battle.

How did the collapse grade the two currents’ predictions?

Harshly and unevenly, since the classical tradition’s designated agent and historical direction received no confirmation from the manner of the regimes’ ending, while the analytical school’s scepticism about planning without markets was borne out in outline.

Yet the school’s own constructive hopes, for a redesigned socialism of a market or egalitarian kind, found no historical bearer either, leaving both currents as critics of a world neither had forecast in detail.

The longer consequence of the shock was institutional, and it shaped the fifth scene of dispersal described earlier. University Marxism contracted; the journals thinned; the September Group’s meetings lost their occasion as its members’ agendas completed their migration into philosophy and social science proper. The classical tradition contracted into its enclaves, Marxology, critical theory and the history of political thought, where its methods faced colleagues who shared its premises about texts and totalities. Each current thus survived the shock in the form its method predicted for survival generally: the analytical by absorption into the general disciplines, its questions naturalised and its name retired; the classical by preservation in a guarded precinct, its name retained and its audience narrowed. A reader deciding which survival is preferable is deciding, once again, what a tradition is for, whether it exists to bequeath questions to whoever can use them or to keep a way of seeing intact against a hostile climate, and the comparison, consistent to the end, declines to decide it on the reader’s behalf.

The firm, the manager and a test neither side set

One further test of the derivation deserves a section, because it concerns an object both currents neglected and their neglect is diagnostic. The modern corporation, a hierarchy in which most production is organised by command rather than by contract between independent parties, sits awkwardly inside both frameworks. For the analytical models, the firm is an embarrassment of exactly the classical kind: inside its boundary, exchange stops and authority begins, so that the economy’s characteristic unit is a zone where the voluntary transactions of the models do not occur, a fact economists of the mainstream, from Coase onward, explained by transaction costs while rarely asking what the zone of command implies for the theory of the wage bargain at its gates. For the classical tradition, the firm should have been home ground, the scene of the domination its categories describe, yet classical work on the actual organisation of production, its hierarchies, its technologies of supervision, its internal labour markets, remained thinner than the tradition’s polemics promised, the labour process debate of the nineteen seventies being the honourable exception that proved the thinness by its isolation.

Harry Braverman’s Labor and Monopoly Capital, published in 1974, stands at the crossing point and shows what each current might have made of the object. Braverman’s account of the degradation of work under scientific management was classical in its categories, the separation of conception from execution as the capital relation’s inner tendency, and empirical in its method, built from the records of industries and the testimony of workers, with mechanisms specified at the level of the workshop. The analytical school, forming as the book appeared, largely ignored it, since its questions were not posed in the school’s grammar. The classical tradition celebrated it and built less on it than the celebration implied, since its questions were not framed in the idiom of totality either. The book’s orphan status in both literatures is the comparison’s quiet evidence that the fork, for all its generative power, does not exhaust the space of valuable work, and that a tradition’s questions can blind it to an object as reliably as its answers can. Readers assembling a personal synthesis from this article’s recommendation should therefore add a third habit to the two borrowed ones: a periodic suspicion toward the shared agenda of both currents, whose agreements about what matters are as inherited, and as unaudited, as their disagreements.

Why did the labour process debate fade from both literatures?

Because its questions, about how authority is built into machinery and job design, required an empirical patience and a technical literacy that neither the seminar room’s counterexample culture nor the editorial office’s synthesis culture rewarded, so the debate’s findings were absorbed piecemeal into industrial sociology while its theoretical charge, that production itself is a political relation, waited for readers outside both camps.

The firm also returns the comparison to its home ground by a back door. Inside the corporation, the preferences of the managed are formed by the managing structure as a condition of employment, through training, appraisal and the internalisation of targets, and the formation is documented in the management literature with a frankness that social theory rarely matches. A modeller who begins at the firm’s gate with given preferences has cropped out a formation process that the firm’s own manuals describe in detail. A dialectician who describes the cropping in the vocabulary of totality has yet to explain any particular firm’s actual practices. The object sits there, enormous and documented, waiting for the combined method this article has recommended, and its waiting is perhaps the most practical verdict available on the dispute: the questions that matter most may be those that require both instruments and are therefore orphaned by the division of labour between them.

The verdict and the condition of its reversal

The verdict of this comparison is deliberately narrow, because the preceding sections have shown that broad verdicts in this dispute are method-dependent and therefore worthless as arbitration. On the eight differences, the evidence divides, with the analytical current prevailing on the logic of explanation and on the candour of normative commitment, the classical current prevailing on the audit of frameworks and on the foresight about trajectory, and the remaining rows divided or underdetermined for reasons the grid states. On the question that organises the whole, whether Marxism possesses a distinctive method, the comparison finds that the claim to a distinctive logic, exempt from general standards, was not sustained by its ablest defenders, who retreated under pressure to a heuristic of attention that any inquirer may share, while the counter-claim, that general standards are innocent of substance, was likewise not sustained, since the standards as practised carried an ontology inside them. Between those two failures stands the single judgement the article is willing to sign without qualification. The classical modelling objection, that the analytical school’s constructive models assume voluntary exchange between agents with given preferences and thereby build their conclusions into their premises, was never adequately answered. It was answered for particular models by revision, and it was answered for the school’s logical and conceptual work by the correct observation that the objection does not reach there. As an objection to the modelling programme itself, it stands where Wood and its other principal authors left it.

Every verdict in this run of the series carries its defeater, and this one is stated with the precision the subject deserves. The judgement that the classical modelling objection was never answered would be defeated by an analytical model that derives Marxist conclusions without assuming voluntary exchange between agents with given preferences, which would show the assumptions to be dispensable rather than load-bearing. The defeater is not a rhetorical flourish. It specifies an object whose existence would change the verdict, describes the property the object must have, derivation of the conclusions from premises the classical side accepts, and invites any reader with the technical equipment to attempt the construction. Attempts would themselves be valuable whatever their outcome, since a failed attempt that locates exactly where the contested assumptions become indispensable would teach both traditions more than another decade of position statements, and a successful attempt would be the most important event in Marxist theory since the school’s founding, reuniting rigour with the constitutive claims in a single formalism.

What should a reader do with a verdict that carries its own defeater?

Treat the verdict as a claim about the literature’s past rather than a prophecy about its future, since the defeater describes a possible work, not an impossible one, and the history of this dispute is full of works that arrived from unexpected quarters.

The verdict grades the answers given; the defeater keeps the examination open, which is the only posture either tradition, at its best, ever recommended.

A secondary verdict concerns the comparison basis stated at the outset. The dispute examined here ran between methods as stated and claims as stated, and the article has refused throughout to grade the political commitments of adherents. The refusal costs the comparison some drama and buys it whatever authority it possesses, since the slide from method to motive is the besetting corruption of this literature, practised by classical writers who read analytical method as liberal politics in disguise and by analytical writers who read dialectics as the alibi of authoritarian parties. Both slides have historical material to feed on, and both are fallacies in the strict sense, since a method’s validity is independent of its holders’ biographies even though a method’s career is not. Keeping the two questions apart, validity and career, is the discipline this article has tried to model, and it is offered to readers as the transferable skill the whole comparison exists to teach: in any dispute between schools, grade the arguments by the standards the arguments invoke, grade the careers by the evidence of what the methods did, and never let either grade masquerade as the other.

For students and exam candidates

Hold the comparison as one fork and eight consequences, and you hold the essay structure most examiners reward. Open with the generative disagreement, whether a distinctive Marxist method exists, derive the differences in the order of the grid, and give the modelling objection the central place its strength earns, since answers that treat it as a complaint about mathematics announce a misunderstanding in the first paragraph. State what each side got right before any verdict, and attach the defeater to your verdict, the model whose construction would reverse it, because a judgement with its reversal condition stated reads as analysis rather than allegiance. Keep the eight differences card, with the root column memorised rather than the wording, in a study file such as the VaultBook study notes for this series, and regenerate the grid from the fork in practice until the derivation is faster than recall. Distinguish the method question from the conclusions a method was used to reach in every paragraph where they threaten to merge, since that distinction, marked explicitly, is the single most reliable discriminator between scripts that describe the dispute and scripts that understand it. Candidates who can also name the strongest bearer of each position, Cohen and Roemer against Lukács and Wood, and can say why the catechism versions were set aside, will find the comparison’s questions, in whatever wording an examination chooses, already answered in structure before they are answered in detail.

For teachers and seminar leaders

Teach the dispute as a structured debate with positions assigned against sympathy, since the analytical case is best learned by a student who must defend dialectics and the classical case by a student who must build a model. Set the eight differences grid as the seminar’s map, one row per meeting, and require each row’s discussion to end by tracing the disagreement back to the fork, marking any row where the tracing fails as a finding against this article’s central claim. The modelling objection deserves a double session, one for its statement at full strength from its principal texts and one for the answers attempted, with the class asked to specify the model that would satisfy the defeater before being told that none exists. Pair the readings across the shelf division, Cohen with Lukács, Roemer with Wood, Wright with Thompson, and grade the pairings by the quality of the residues students identify, the untranslated remainder in each translation being the seminar’s most reliable evidence of understanding. Warn the room early against the five misreadings, especially the jargon reading and the trajectory fallacy, and return to the warnings when discussion slides toward biography, since the slide from arguments to biographies is the failure mode of seminars on this dispute as reliably as of the literature itself. A cohort that leaves able to state both sides’ strongest cases, and to say what would change its own mind, has been taught the dispute rather than merely informed about it.

For researchers, historians, and journalists

Treat the comparison as a map of two literatures that cite past each other, and exploit the parting of the shelves. Analytical results on exploitation, class measurement and electoral dilemmas sit in philosophy, sociology and political science journals whose findings Marx scholarship rarely imports; classical results on value form, fetishism and hegemony sit in Marxology and critical theory whose findings the mainstream rarely imports, and synthesis work that carries a result across the divide, with its standards audited at the border, remains scarce and therefore valuable. Verify every quotation and attribution in this literature personally, since the polemical tradition has normalised paraphrase dressed as quotation, and the modelling objection in particular circulates in versions its principal authors would not sign. For historians, the school’s own archive, the September Group’s papers and the journals’ editorial records, is an underused source for the sociology of postwar socialism’s intellectual collapse and reinvention. For journalists, the dispute supplies the grammar behind recurrent public arguments about whether markets are natural, whether preferences are sovereign and whether expertise is neutral, and a reporter who can hear the generative fork inside a policy quarrel will write the quarrel better than one who hears only the conclusions. In all three professions, the standing rule of this article applies: attribute positions to texts and dates, never to camps in the abstract, and treat any summary that cannot name its bearers as unresearched.

Alienation, human nature and the young Marx

The two currents divide over the young Marx as predictably as over everything else, and the division is a compact illustration of the fork applied to biography. The classical tradition, especially in its humanist branch, reads the Economic and Philosophic Manuscripts of 1844 as the theory’s foundation: alienation, the worker’s separation from the product, the process, the species and other people, names the human cost of the capital relation in a vocabulary that joins economics to an account of what human beings are for. Althusser’s classical counter-current declared an epistemological break, the mature Marx of Capital having abandoned the humanist ideology of the manuscripts, a declaration analytical writers found congenial in its verdict and deplorable in method, since it settled a question of interpretation by fiat about science and ideology rather than by reading. Elster’s treatment performed the analytical operation on the concept itself: alienation was separated into objective and subjective components, the objective condition of separation from control over one’s activity and its products, and the subjective feelings that may or may not accompany it, after which the concept could do empirical work, distinguishing the alienated assembly worker who feels nothing from the fulfilled craftsman inside relations of domination.

The classical objection to the operation is the constitutive one in miniature. Alienation, on the humanist reading, is not two facts, a condition and a feeling, that happen to correlate; it is a relation in which the condition deforms the feeling, so that the worker who feels nothing is not a counterexample but the deepest case, the separation having consumed the capacity to register it. Separating the components for measurement therefore dissolves the phenomenon’s most important instance into the residual category of measurement error. The analytical reply is the school’s standing one: a phenomenon whose deepest instances are invisible to its measures may still be real, yet a theory that counts invisibility as confirmation has insured itself against the world, and the insurance premium is paid in credibility with every audience the theory hopes to persuade. Both speeches are correct about the other’s danger, and the literature on alienation since the quarrel, richer in measures and in theory than either side produced alone, suggests that the concept survived by being fought over, each camp’s audit preventing the other’s complacency.

Why does the young Marx embarrass the analytical school less than expected?

Because the manuscripts’ central claims, once separated into condition and feeling, mechanism and meaning, proved partially translatable into the school’s grammar, and the school’s embarrassment is reserved for the untranslatable remainder, the claim that the separation itself is the alienation’s mode of operation, which the school must either model as constitution or decline as mystification.

The human nature question beneath the alienation dispute shows the same structure at greater depth. Classical Marxism, in most of its branches, holds a historical conception of human nature: needs, capacities and forms of individuality are produced by modes of production, so that the acquisitive individual of market society is a historical product rather than a natural given. Analytical models require a thin human nature, agents with preferences and reasoning capacities stable enough to model, and the school’s writers defended the thinness as methodological modesty. Geras, again the tradition’s internal auditor, complicated both positions by arguing that Marx held a concept of human nature, constant needs and capacities, hunger, health, the need for varied activity and for others, beneath the historical variation of their forms, and that historical materialism requires the constant layer to explain why history has the direction and the costs it does. The argument, analytical in method and classical in comfort, is among the comparison’s best exhibits of a result neither camp’s programme predicted, produced by applying the school’s standards to the tradition’s texts with the tradition’s questions in mind. It stands in this article as evidence that the combination recommended throughout is not merely a diplomatic hope but a demonstrated producer of findings, since Geras’s book changed how both camps teach the subject.

Marxian economics after Sraffa and the analytical inheritance

Behind the school’s value verdict stood a technical literature the comparison should acknowledge, because it shows analytical examination operating on Marxian economics before the school had a name. Piero Sraffa’s reconstruction of classical political economy gave economists a rigorous account of prices and distribution in which labour values were computable and, as Ian Steedman argued in Marx after Sraffa, redundant: the physical data of production, with the wage, determine prices and the profit rate without the detour through values, so that the labour theory survives, if at all, as an unnecessary middle term. Steedman’s book, published in 1977, was the analytical school’s founding economic document in everything but membership, and its reception rehearsed the whole comparison in advance: defenders of the theory charged that redundancy in a price system misses the theory’s purpose, the analysis of exploitation and form, while critics answered that a purpose immune to redundancy arguments must be stated as a different theory and defended on different ground. The value-form tradition’s later flourishing is partly the long answer to Steedman, an answer that accepted his arithmetic and rebuilt the theory’s centre elsewhere, exactly the pattern of division under examination that this article has tracked across every row.

The episode also corrects a lazy assumption about the school’s novelty. Analytical Marxism did not import rigour into a field that had never known it; Marxian economics had a mathematical tradition from Dmitriev and Bortkiewicz through the Japanese and Italian schools, and the classical tradition’s own economists modelled reproduction schemes with sophistication. What the school imported was not rigour but jurisdiction: the claim that the standards of the surrounding disciplines, philosophy’s included, governed Marxist claims without remainder, and that no internal court of the tradition could overrule them. The classical objection, properly aimed, targets the jurisdiction rather than the rigour, which is why classical writers who share the rigour, Colletti in philosophy, the temporal interpreters in economics, still refuse the jurisdiction, holding that a tradition may use every instrument of its neighbours while reserving the right to judge the instruments’ fitness for its objects. Stated that way, the dispute narrows to a question any mature discipline faces about its borders, and the comparison’s verdict on the modelling objection can be restated in the same terms: the objection is the border dispute’s sharpest case, the point where an instrument’s fitness for the object was challenged and the challenge was met with improvements to the instrument rather than with a defence of its fitness.

Did Steedman’s redundancy argument refute the labour theory of value?

It refuted the theory as a necessary detour in the determination of prices and profits, a result the subsequent literature has not overturned, while leaving open the theory’s other employments, in exploitation accounting and form analysis, whose defenders must show that the employments need the value categories rather than merely preferring them.

The distinction between refuting a use and refuting a theory is the episode’s permanent lesson.

A further inheritance deserves recording. The analytical school’s standards migrated into Marxian economics so thoroughly that the field’s journals enforce mechanism, specification and formal clarity as house rules, whatever the author’s allegiance, while the classical tradition’s questions survive inside the same journals as the standing agenda of what the models leave out. A young economist entering Marxian economics inherits both, usually without knowing their provenance, and works the border dispute as a daily practice of choosing specifications and defending exclusions. The comparison’s large claim, that the generative disagreement organises the field, is thus testable in the small: open any issue of the field’s journals and the fork is visible in the referee reports the articles survived, one referee demanding the mechanism, another demanding the form, the author negotiating between tribunals that the history recounted here created.

What remains: the dispute as a research instrument

A comparison that ended in verdicts alone would waste its subject, since the dispute’s chief value to a working reader is instrumental: the two currents, taken together, form a checklist for interrogating any social theory, Marxist or not. From the analytical side, take the questions that expose. Can the theory’s claims be stated one by one, and does each survive its statement? Where a pattern is explained by its benefits, what mechanism connects benefit to persistence? Which values does the critique presuppose, and are they owned? What observation would embarrass the theory, and has the theory’s literature ever reported one? From the classical side, take the questions that cross-examine the tools themselves. What do the chosen categories take for granted about their object, and might those takings-for-granted be among the matters in dispute? Are the people and groups under study made what they are by the very relationships being investigated, and if they are, what does a design that freezes them at the start leave out? Does a key definition still contain the phenomenon that made the category worth coining? Where will this approach predictably fail to look, and does its history confirm the prediction? A researcher who runs both lists against a fresh theory, a theory of platforms, of populism, of artificial intelligence and labour, will find the lists generating a research programme between them, which is the strongest evidence this article can offer that the comparison’s subject is alive.

The checklist also explains the article’s confidence in its central claim. The generative disagreement is not a metaphor imposed on the literature for tidiness; it is the observation that the two lists of questions descend from two answers to one question, whether a theory of society may claim a method of its own. Every item on the analytical list presupposes the negative answer, that claims answer to general standards. Every item on the classical list presupposes the affirmative possibility, that an object may be falsified by the standards brought to it, so that the standards themselves require a theory’s audit. A reader who sees the descent can improvise new items for new cases, extending both lists to objects neither tradition imagined, and improvisation of that kind, rather than recall of the eight differences, is the mastery the article has aimed to teach from its first section.

How should a reader choose between the currents in practice?

By refusing the choice at the level of allegiance and making it at the level of the question, since mechanism questions belong to the analytical instrument and constitution questions to the classical one, and most real inquiries contain both kinds braided together.

The practical skill is unbraiding, assigning each strand to the instrument built for it and auditing the joins where the instruments’ answers must be made to agree.

Unbraiding has a discipline of its own, worth stating for the researchers among the article’s readers. The join between a mechanism account and a constitution account is where explanations in social science most often fail silently, the mechanism presupposing agents whose formation the constitution account describes, the constitution account presupposing processes whose operation the mechanism account must verify. Work that respects both sides of the join is recognisable by a stylistic signature: it narrates formation and then models interaction, or models interaction and then interrogates the model’s primitives against the history that produced them, and it flags the seam rather than papering over it. Wright’s late work has the signature, and Brenner’s, and the best of the value-form literature where it meets economic history. Work that lacks the signature is recognisable too, by the confidence with which it treats its starting point as the world’s. The comparison therefore ends its practical teaching where its theoretical teaching began, at the starting point: every analysis starts somewhere, the traditions differ in whether the start is a convenience to be forgotten or a commitment to be audited, and the reader who audits every start, including this article’s, has learned the one lesson both currents, at their best, were trying to teach.

The comparison restated for the reader who arrived late

A summary is owed to readers who use reference articles by their endings, and it is given here in the article’s own terms rather than in the slogans of either camp. Two rival heirs claim what Marx left behind. The first heir treats the legacy as a portfolio of assertions, each to be marked against the same rules of evidence and inference that govern assertions anywhere, and accepts that honest marking writes down much of the portfolio while rescuing the credit of the remainder. The second heir treats the legacy, instead, as a way of seeing, a grasp of an interconnected whole whose members take their character from their places in it, and warns that the marking exercise misdescribes the object before it begins. Eight contrasts unfold from that divergence, touching how inquiry proceeds, what becomes of the value theory, how exploitation and class are to be defined, what an explanation must contain, whether moral philosophy belongs inside the theory, what moves history, and who if anyone is appointed to change it. The preceding sections traced every contrast back to its source and weighed the evidence contrast by contrast, without pretending the weighings could be totalled into a championship.

One argument towers over the rest in this article’s judgement. It is the classical charge against the modelling programme: that representing a capitalist economy through choosers who arrive already endowed, already wanting what they want, and already free to refuse the wage bargain, settles in the setup the very questions the theory was built to ask, about how that freedom is manufactured and how those wants are made. Replies were offered to many local versions of the charge, and the school’s purely logical work stands outside its reach altogether, but the charge addressed to the programme as a programme received no adequate reply. The accompanying verdict is issued with the condition of its own withdrawal spelled out elsewhere in these pages: exhibit a formal construction that reaches the theory’s characteristic results while dispensing with the contested setup, and the judgement lapses, because the setup will have been shown to be scaffolding rather than foundation.

What each heir got right can be carried away in two sentences each. The analytical heir saw that vagueness had become a hiding place for error, that citing a pattern’s usefulness is not yet explaining the pattern, that putting claims into exact shape lets the world vote on them, and that a critique coy about its own values invites exposure by its opponents. The classical heir saw that tools are not innocent, that class and exploitation bleed out their explanatory force when the relations of production are defined away at the start, that a method’s opening moves foretell where its followers will end up with a reliability the followers themselves will be last to notice, and that certain objects, social forms and shared meanings above all, fall apart under the dissecting knife before the dissection can describe them. A reader who can hold all eight sentences together, and who understands why holding them together is difficult, possesses the comparison entire, and possesses also a way of reading any future dispute between a science and a tradition, since the fork examined here is one of the permanent shapes of intellectual life, and Marxism’s version of it is simply the best documented case.

A note on sources and the discipline of attribution

The comparison has insisted throughout that positions be attached to named writers and dated works, and a closing methodological note explains why the insistence is more than scholarly hygiene. In a dispute where each camp maintains a polemical portrait of the other, unattributed summary is the mechanism by which the portraits reproduce. A sentence beginning with the claim that Marxists believe, or that analysts assume, has usually laundered a contested interpretation into a fact of the literature, and the laundering is invisible precisely because the sentence is grammatical and confident. Attribution interrupts the process. To write that Elster argued a proposition in 1985, or that Wood answered it in the essays collected around the retreat from class thesis, is to give the reader a handle by which the claim can be lifted, checked and, where the check fails, corrected. The discipline costs words and pages. It buys the only currency a comparison can spend, which is the reader’s justified confidence that the positions graded in the verdict are the positions their holders actually took.

Attribution has a second function in this particular dispute, because several of its principals revised themselves in public and a comparison that cites a writer without a date silently chooses which of the writer’s selves to grade. Cohen of the late nineteen seventies, reconstructing historical materialism with a defender’s loyalty, and Cohen of the lectures on equality, auditing that loyalty with a philosopher’s regret, are both real, and the school’s story cannot be told by citing either as the man. Wright’s asset typologies and Wright’s real utopias belong to one bibliography and two tempers. Lukács wrote his founding text, recanted it under discipline, and spent a long career negotiating with his own recantation. A reader who meets these writers as dated sequences rather than as statues understands something about traditions that no grid can state: that a current is a population of arguments moving through time under pressures its own method partly creates, and that the comparison of currents is therefore always, in part, the comparison of trajectories. The grid freezes the trajectories for inspection, as a map freezes a river, and the article has used the map while warning, at every stage where the warning mattered, that the water moves.

A third function is defensive, and it concerns the reader’s protection against this article itself. Every summary in the preceding sections, of the modelling objection, of the value-form rescue, of the Brenner evidence, is an interpretation offered by a comparer with a thesis, the generative claim, that the interpretations collectively support. A comparer with a thesis is a witness with an interest, and the citations scattered through the argument are the reader’s means of cross-examination. Where the article has paraphrased, the paraphrase can be tested against the passage; where it has quoted, the quotation is short enough to be checked in a minute and honest enough, the author hopes, to survive the checking. That is the analytical virtue of auditability placed, deliberately, at the service of a comparison whose sympathies are divided, and it is the last practical lesson the dispute teaches: that the defence against a loaded framework is not a framework without loads, which does not exist, but a framework whose loads are labelled, located and left where the reader can weigh them.

Frequently Asked Questions

Q: What is the core disagreement between analytical and classical Marxism?

At root the two currents split over a single issue: whether this theory owns a proprietary way of reasoning. One camp insists it does not, holding that any assertion worthy of belief can be spelled out in plain logical shape, backed by evidence and left open to refutation, with no dialectical exemption available. The other camp insists it does, holding that the concepts involved apply only to a determinate historical order, that the subject matter is an interconnected whole whose members owe their character to their places within it, and that procedure and substance cannot be prised apart. This study calls that fork the generative disagreement, because the eight further contrasts, spanning agency, history, ethics, explanation, class, exploitation and value, unfold from whichever answer a writer gives here.

Q: What are the main differences between the two currents?

The grid presented earlier in this study arranges them in a fixed sequence. Procedure comes first, exact formal construction set against dialectical treatment of a whole. The value theory follows, mostly discarded by one side and defended by the other. Exploitation is then defined in two incompatible ways, by unequal holdings or by command inside production, and class follows suit, sorted by possessions and credentials or by antagonistic relationships. Explanation divides over whether wholes may act without decomposition into individual mechanisms. Moral philosophy is welcomed as a partner by one side and distrusted as ideology by the other. History is powered, depending on the camp, by technical development or by struggle between classes. Agency closes the sequence, with a designated revolutionary class affirmed on one side and refused on the other.

Q: What is the modelling assumptions objection?

The charge is that the school’s formal constructions decide the outcome before the machinery starts. People enter the constructions as owners of bundles of resources, carrying wants that are treated as already settled, and meeting one another through bargains each is free to refuse. Yet the disputed theory denies precisely those starting points: it describes the wage bargain as compulsion wearing contractual dress, grounded in an earlier separation of producers from productive wealth, and it treats wants as manufactured inside the relationships being examined. A construction that begins by granting what the theory contests will predictably fail to reproduce the theory’s results, and reporting that failure as a discovery mistakes the blueprint for the building. The charge aims at method, not at particular results, and this study judges it unanswered.

Q: What did each side get right?

Four credits stand on the analytical side of the ledger. Murkiness had sheltered feeble reasoning, and the demand for plain statement exposed it. Explanations that cite a pattern’s benefits owe a connecting process, and the demand for one permanently raised the standard. Exact statement lets evidence arbitrate, which protects honest work and embarrasses evasive work. A tradition that condemns while disowning its own values needs the values dragged into daylight, as the school’s normative turn dragged them. Four matching credits stand on the classical side. The premises smuggled into a framework decide results, and somebody had to say so early and often. Class emptied of production relations loses the phenomenon it was coined to explain. A school’s opening commitments foretell its destination, and the critics who forecast the analytical destination from its premises were vindicated. Certain objects, constituted by meaning and relation, genuinely suffer under decomposition.

Q: Does formalisation distort the object?

The honest answer is: for some objects, demonstrably, and the cases can be enumerated rather than gestured at. Where the subject matter is made of meanings, relationships and inherited forms, as with the value form, class consciousness, ideology or a society’s shared common sense, stripping context to isolate variables discards the phenomenon and studies a residue. Where the subject matter aggregates, prices, quantities, distributions and demographic movements, exact treatment sharpens rather than damages, and the classical economists of the tradition modelled happily themselves. The serious classical claim was never that equations always lie. It was that choosing an equation is a theoretical decision whose price must be counted object by object, while the serious analytical reply was that refusing equations charges a price of its own, shelter from refutation, which must be counted with equal candour.

Q: Is there a distinctive Marxist method?

Split the claim and the answer clarifies. Asserted as a private logic, a licence to draw conclusions by rules no outsider may inspect, the claim collapsed under examination: its most capable defenders, pressed hard, redescribed their method as a habit of attention, a tendency to look for relationships, histories and internal tensions before looking for aggregates. Asserted as that habit of attention, the method plainly exists and marks its holders, who notice social forms, their presuppositions and their inversions, where differently trained inquirers notice only variables. A habit can be borrowed by anyone, and borrowing it requires no allegiance. An exemption from shared standards was what the founding dispute was about, and no participant in the dispute succeeded in establishing one.

Q: Was the school’s trajectory foreseeable?

It was forecast, in detail and in advance, by the school’s classical opponents, which is a stronger fact than foreseeability in hindsight. Reading the programme’s opening commitments, those critics announced the sequence: exploitation would migrate from production to distribution, class would soften into stratification, moral philosophy would displace historical inquiry, and the designated revolutionary agent would quietly leave the theory. The published record then enacted the sequence step by step, every step defended on its merits by scholars of obvious sincerity. Two morals compete for the episode. On one, a loaded method ground a doctrine down. On the other, honest scrutiny dissolved weak claims. The bare sequence cannot choose between the morals; joined with the modelling objection, which names the loading mechanism, it becomes the classical case’s strongest supporting evidence.

Q: Can the two currents be combined?

Working scholars combine them already, and this study’s practical advice is to do deliberately what the best practice does by instinct. Borrow the analytical habits of separation and exposure: one assertion at a time, a named process beneath every pattern credited to a benefit, values confessed in the open, refutations received as information. Borrow the classical habits of suspicion toward instruments: ask what a framework has granted itself before admiring what it proves, refuse definitions that evict the phenomenon, and treat wants and agents as products of the arrangements under examination. Brenner’s histories and Wright’s later studies each practise a version of the mixture. No school flies the mixture’s flag, perhaps because the flag would have to announce a double allegiance, to clarity and to distrust of clarity, that institutions find hard to recruit under.

Q: What is the retreat from class argument?

Ellen Meiksins Wood’s phrase names a staged withdrawal she traced through the school’s own publications. First the account of exploitation moved house, from the factory floor to the distribution of holdings. A distributive account then needed principles of fairness that any reasonable citizen could endorse, and principles of that reach appoint no particular class as history’s instrument. Class analysis rebuilt from holdings meanwhile disclosed a splintered landscape of positions, no one of them a majority, none plausibly cast as universal agent. A theory of history powered by technical development completed the sequence, since such a motor turns without anybody’s consciousness at the wheel. Wood’s contention was that nobody planned the withdrawal and nobody betrayed anybody: the premises, followed faithfully, led there, because a framework that starts from separate choosers and their settled wants has nowhere to lodge a collective subject made by its place in a whole.

Q: Which side does the evidence favour?

Row by row, the honours divide, and pretending otherwise would falsify the record this study has assembled. Formal scrutiny favours the analytical camp on how explanations must be built, the demand for connecting processes having become the profession’s common standard, and on the value theory read as an account of price formation, where the mathematical results stand. It favours the classical camp on the auditing of frameworks, the modelling charge having survived every reply addressed to the programme as such, and on prophecy, the forecast of the school’s destination having been borne out. Two rows, what powers history and who is appointed to transform it, remain open on the evidence available to either method. The study’s signed judgement is accordingly narrow, confined to the one objection that meets both camps’ standards for an unanswered argument, and issued together with the construction whose appearance would withdraw it.

Q: How does the generative disagreement order the other differences?

By entailment rather than by association. Suppose a writer grants the proprietary method, and a family of positions assembles itself: wholes may explain their parts, the value doctrine stays embedded in a larger analysis, class and exploitation are defined through relationships of command, moral philosophy is suspected as a foreign tribunal, struggle drives history, and one class is cast as its bearer. Suppose the writer refuses the proprietary method, and the opposite family assembles with equal logic: doctrines face judgement separately, definitions are rebuilt from holdings and positions, explanations descend to individuals and their interactions, values are argued as philosophy, technical development takes the driver’s seat, and no bearer is appointed in advance. Because each local stance is downstream of the founding answer, local victories change nothing, which is why the quarrel outlived decades of them.

Q: What should a student separate between a method and the conclusions it is used to reach?

Keep two judgements apart that the literature habitually fuses. A method is neither canonised by one beloved result nor damned by one unwelcome one, so the school’s destination cannot by itself settle whether its procedure was sound, and Cohen’s most elegant constructions cannot by themselves certify the modelling programme. The fusion tempts because conclusions are vivid and methods are abstract, and polemic exploits the temptation by exhibiting a hated destination as if it were a refutation, or a loved one as if it were a vindication. A student who keeps the levels separate can ask, of any episode in the dispute, the two questions this study has kept apart throughout: whether the reasoning holds by the standards it invokes, and what faithful use of the procedure did, across a generation, to the doctrine that adopted it.

Q: Why does the question of a distinctive method come first?

Priority belongs to it because it fixes the rules by which every subsequent question will be argued. Whether the value doctrine is true depends on what may count as its test, and that depends on whether doctrines face judgement separately or only as moments of a whole. Disputants who skip the founding question therefore argue in different grammars while imagining a shared one, each honestly baffled by the other’s refusal to accept a plainly decisive point. The founding question also earns its place by its predictive power: learn a writer’s answer to it and the writer’s positions on class, exploitation and agency can be forecast more reliably than from any summary of the writer’s conclusions, a fact this study has used as the working test of its central claim.

Q: How does the eight differences grid order the comparison?

Its ordering principle is descent from a source, not chronology and not importance. Procedure stands at the head because everything else is procedure’s consequence. The technical heartlands follow, value, exploitation and class, where the consequences are sharpest and the evidence richest. The philosophical rows come next, explanation and moral theory, where the argument concerns what may be said at all. The political rows close the table, the engine of history and its appointed driver, where the consequences matter most and the evidence obliges least. The columns impose a matching discipline: rival positions in single phrases, the shared root named for every row, and an evidence column that is allowed to record a stalemate where the disagreement is constitutive, rather than manufacturing a winner for the sake of the table’s symmetry.

Q: What separates assuming voluntary exchange from deriving it?

The distance between the two is the distance the modelling objection measures. Grant the bargain’s freedom at the outset, as a premise about choosers who may walk away, and the wage relationship’s character is decided before inquiry starts; no theorem proved downstream can unsettle a premise installed upstream. Begin instead from the materials the classical side insists on, an earlier expropriation of producers from productive wealth and wants formed inside the resulting relationships, and then exhibit the free bargain arising out of those materials, as an appearance with a definite structure and definite limits, and the bargain’s freedom has been explained rather than presumed. The first route begs the disputed question. The second route would answer the objection outright, which is why the withdrawal condition attached to this study’s verdict is written as a demand for exactly that derivation.