The direction of historical development

Jairus Banaji History as theory: essays on modes of production and exploitation Historical Materialism books series, Vol 25, Leiden, 2010, pp406, £81

This is the third part of my review of Jairus Banaji’s History as theory. In the first part I set up the book’s political and theoretical context and made some general comments about what Banaji has, I think, established by his work, and some criticisms of the shape of his arguments. In the second part I addressed the objection to ‘teleology’ which Banaji throws around without much explanation in his essays; and I discussed the part of the foundations of historical materialism which is not in itself historical.[1]

In this third part I come in the first place to the reasons for believing that history has some degree of directionality. These reasons will shape the second question, which is what sort of historical periodisation we should be attempting (‘modes of production’, etc). I argue that the best approach is that what is involved is the rise of certain social forms and dynamics; their creation of state and religious forms corresponding to their dynamic; their further rise to apogee; and their subsequent decline in which a variety of attempts are made to hold onto them while managing the effects of decline, or to replace them. Not all attempts at a new order succeed; selection both among them and relative to the declining old order eventually produces a new order and dynamic, which actually succeeds in replacing the old.

In this approach, there are periods of transition and forms transitional between one social order and its successor. It therefore could be that the forms of agrarian relations found in modern India have this character. It follows that Banaji does not in the end succeed in demolishing on the basis of the historical evidence the conceptual foundations of ‘official communism’ and peasant-centric Maoism. Reasons for rejecting these policies will therefore have to be at a level closer to the political concrete. Once we look at this level, simple labour organising at the base will not represent a real strategic alternative to either the ‘official communist’ strategy or that of the Maoists.


Does history have a direction? Marx and Engels certainly thought so; part of Popper’s and similar objections to ‘teleology’ was to deny it did, and to claim that Marxism imposed an imagined direction on history in the hope of a perfect future. I said earlier that to reject the charge of ‘teleology’ is not to prove or even affirm that history has a direction.

There are essentially two reasons given by Marxists for supposing that history has a direction - and they are both based primarily on inferences from the past and present, rather than on hope for the future.

The first, which has been treated elaborately by Gerry Cohen in Karl Marx’s Theory of history: a defence (1978) is that the historical evidence should lead us to suppose that the technology of production tends over the long term to improve. There may be significant periods of stagnation and there may even be local periods of regression in relation to certain technologies (as, for example, in western Europe after the fall of the Roman empire in relation to building and ceramics): but at the end of the day technical improvement tends to be cumulative.

Marx and Engels wrote that “The hand-mill gives you society with the feudal lord; the steam-mill, society with the industrial capitalist.”[2] The tag is striking, but violently inaccurate (feudal, medieval Europe was a society in which water-mills were ubiquitous) and both over-deterministic (a society may have the water-mill without being properly called feudal, as in the case of Tang-dynasty China) and over-specific.

The truth it captures, however, is that the overall productivity of labour and certain specific technologies limit the sorts of social order that are possible. On the one hand, the productivity of food production limits both the absolute population that can be supported, and the proportion of this population that can do anything except agriculture, pastoralism or fishing and their immediate support activities. On the other, the widespread diffusion of the water-mill (retaining the present example) frees up very large amounts of labour time.

These are merely examples, and beyond this point lie complex historical debates both about the evidence for diffusion of technologies, and about cause and effect relations between the adoption of new technologies (in Marxist terms development of the forces of production) and incentive effects of the social order (in Marxist terms the relations of production). For example, did the existence of the institution of slavery (or perhaps the availability of cheap slaves) disincentivise the adoption of labour-saving technology, whether because cheap labour made it economically irrational or because slave-owning aristocrats are not economic ‘rational maximisers’? This is a significant issue in the debate between ‘primitivists’ and ‘modernists’ about the ancient economy.

For Marxists of the schools of Gerry Cohen, or of Chris Harman, such questions are critically important. For these schools it is the human character as homo faber, humans as toolmakers and technical innovators, which in the last analysis is the sole driver of historical change. On this approach, new forces of production on their own demand new relations of production, and it is this contradiction which forces social revolutions. There is clear evidence that this was Marx’s view, in the Preface to the contribution to the critique of political economy. Cohen admitted to difficulty in explaining how new forces of production demand new relations of production.[3] It seems simpler to accept that the extra-European evidence, particularly from premodern China and India, is against such a demand. New forces of production make possible new relations of production. The demand comes from elsewhere in human nature.

This ‘elsewhere’ is primitive communism. Marx’s and Engels’ concern for this category is perfectly clear both in the early work, and in the late work reflected in Marx’s Ethnological notebooks and Engels’ Origin of the family. What they assert - with very varying degrees of explicitness - is that primitive communism grows out of or is a feature of the underlying nature of humans; that class stratification is therefore in a sense unnatural to humans.

The language of Marx and Engels’ early arguments on this issue is Young-Hegelian. The late work is primarily empirical rather than expressed as systematic theory. But the point can be perfectly well made in the terms of evolutionary biology. We have given substantial space in this paper to some of the most systematic arguments of this sort, those of Chris Knight and his co-thinkers. It is unnecessary, however, to agree with the exact detail of these arguments in order to accept the basic point.

Physical evolution by natural selection takes place over very long time-spans: even the rapid-change punctuations of ‘punctuated equilibrium’ last millennia.[4] Humans evolved as scavenger-gatherers and hunter-gatherers and lived in social groups of this sort for at least 98% of the existence of the species. The human social order in question involves low levels of inequality between persons and substantial commitments to cooperation and sharing, in very marked contrast to our nearest genetic relatives, common chimpanzees.

These aspects are observable in present and recent hunter-gatherer societies, so that we are not wholly dependent on the interpretation of the archaeological evidence for early humans. We also have present (recent) evidence in relation to health and other outcomes that humans are ill-adapted to high levels of social inequality, conveniently collected by Richard Wilson and Kate Pickett in The spirit level (2009). We have no reason, therefore, to suppose that humans will physically adapt to class society without either another 200,000 years (or more) of social hierarchy, or the class elite finding some way to genetically engineer deference into the lower orders.

In the historical evidence, this ‘contradiction’, or more broadly ‘conflict’,[5] between human nature and class hierarchy probably underlies the phenomena of episodic slave uprisings, peasant jacqueries, maroon runaway slave communities and other aspects of ‘the art of not being governed’,[6] religious and other forms of utopian communist critique of the existing order, and forms of partial egalitarianism within ruling elites.

The conflict also sets up a profound incentive for individuals and family groups to hope to join, or at least to live like, the class elite: to climb the greasy pole of social stratification (or, at least, not to slide down it). This in turn logically implies an incentive to introduce forms of technical and division-of-labour innovation. In their early stages, until generalised, such innovations can improve the relative position of the introducer within the society.

These incentives in turn imply that class forms of society - however they originally came into existence - will support denser populations, technical superiority, and hence superior military potential to forms of pre-class society. Meanwhile, both individual/family aspirations and the internal collective conflicts of class forms of society, which flow directly from the conflict between class inequality and the evolved character of the species, drive class societies to expand at the expense of neighbouring - or, in capitalism, distant - social groups having lower population density and/or military capability. By doing so internal class conflict within the conquering society is mitigated at the expense of the conquered, whether by exporting population, by taking and redistributing forms of tribute, or by both.

There is therefore an underlying tendency for class societies to displace pre-class societies. This tendency is also and necessarily a tendency for forms both of technology and of social division of labour (class orderings) which support higher population densities and superior military potential to displace ‘lower’ forms.

All the while, internal conflicts are not overcome but merely mitigated. When taken together with growing forces of production, the social dynamics therefore imply a long-term dynamic or directionality within societies towards less onerous forms of the subjection of subordinate classes to ruling classes.

This, too, is a selective product. Within class society as such, less onerous forms of the subjection of subordinate classes imply more potential for the aspiration among these classes to live like the elite to drive innovation. They probably, though this has to be more tentative, also imply more military capability. For a single example, slaves are, apart from their productive work, a pure military liability; medieval villeins could be employed as (low-status) soldiers.[7]

It is legitimate to infer - with appropriate caution - that this long-term tendency towards less onerous forms of subjection ultimately points towards the potential overcoming of the social institution of class.

This potential is presently posed as a short-term political problem because all forms of class society involve a necessary dynamic towards randomly selected innovation (‘growth’) with some bias towards military technique; and, whether war or class society came first (a chicken and egg problem), all forms of class society certainly involve a tendency towards war. As of the 21st century, the destructive powers both of weaponry (nukes) and of the necessary dynamic towards ‘growth’ (global warming), have the potential to destroy the biosphere and the human species in the short term. We need to bring the destructive powers of our collective productive powers under conscious collective control; and to do so we need to overcome the antagonism between human nature and social hierarchy which drives both ‘growth’ and war.


The logic of this argument does imply a sequence of - in broad outline - slavery, feudalism, capitalism. Slavery - in very diverse forms - is present almost as soon as we have written records, and what lies behind it is very obscure.

The ‘Asiatic mode’ which Marx and Engels placed as the first form of class society has recently been renamed the ‘tributary mode’ and in that guise offered to swallow up feudalism. The idea of the ‘Asiatic mode’ suffers from dependence, through tralatician transmissions of categories, from classical Greek, proto-racist characterisation of their big and episodically antagonistic neighbour, the Iranian Achaemenid empire.[8] It also suffers (as Perry Anderson argued) from assimilating to one another a wide range of very different societies widely separated in time. This second objection is even stronger to the “tributary mode”, which tends to swallow up feudalism and represent an evolution towards Weberian and similar concepts of “pre-modern” societies as all much of a muchness.

Archaeologists have in the past written of redistributive “temple states” in early Mesopotamia, an idea which is out of fashion, more recently of “palace states” in the Bronze Age Near East. Though we have now a lot more archaeological information about very early urban societies than was available in the 19th century, we still probably do not know enough to make firm characterisations of the social orders which produced the archaeology. And what written records show us is societies where land could be privately sold and conveyed (even if it was characterised as state-owned in order to legitimise taxes and other state takings)[9] and slavery was practised.

I emphasised very diverse forms of slavery. Notoriously (to simplify grossly) Athens practised private ownership of land and slaves (who were not supposed to be Hellenes), while Sparta ran a system which could be described as collective appropriation of a conquered Hellene people, the Messenians or ‘helots’, and their land. Systems of the Spartan sort have been commonly said not to be ‘really’ slavery. This idea depends on reading off the concept of slavery as a social form from its specific concept in Roman private law. Treating helots or similar ancient peasants tied to the land as - as it were - ‘villeins’ has similar converse problems. Helots were not members of Spartan society in the sense that medieval villeins were members of English or French society or of ‘Christendom’, or even in the sense that medieval peasants in Islamic lands were members of the umma. Orlando Paterson is, I think, right to argue in Slavery and social death (1982) that at the core of slavery is social exclusion.

In this sense some ‘Asiatic mode’ regimes were arguably in their origins and in substance slave social orders. Mesopotamian empires, and following them both the Achaemenids and the Sasanians, transplanted large groups of war captives to dislocate them from their social context. The same practice of transplantation was found in Chin and Han dynasty China. The latter also displayed large-scale use of state penal slavery as a means of extracting surplus. The Hindu sources represent early proto-Hindu northern India as a variant on a slave system of the ‘Spartan’ type, with the twice-born castes as the only real members of the society.[10]

To return to a point made at the end of the second article in this series. To say what I have just said is to say that ‘slavery’ and ‘villeinage’ have to be distinguished by relations to the social order and social division of labour as a whole and not merely by the immediate ‘relations of production’ as they appear by analogy with the employment relation in capitalism. The difference is the downward obligations owed by lords to their villeins (and, similarly, whichever way the causation might run, by lords and by kings to their free fideles). Even if this has little or no representation in secular law, it is present in most of the universalistic religions (catholic, though not orthodox, Christianity, Islam, Buddhism) and has operative effects insofar as landlords and clerisy compete with one another for social surplus drawn from the lower orders and peasants and artisans can, consequently, appeal to parson against squire (or the equivalents) and vice versa.

Hence a ‘slave social order’ in the sense that must be used for historical periodisation does not mean a social order in which the only primary producers are slaves, or the numerically dominant primary producers are slaves. Nor even one in which - as de Ste Croix argued - the primary form of extraction of the surplus which supported the elite was slave production.[11] The point is that the institution of slavery - extracting labour through control by ‘social death’ - shapes the general institutions of the society, the form of the vertical aspect of the division of labour (between those who decide and those who obey) in general.

A slave is not just a slave because of a ‘purely economic’ relation to his or her master. The relation is necessarily juridical, in relation to the society as a whole. It is also necessarily religious: in slave social orders in the sense used here, there are slave cults separate from the cults of the free, and ancient redistributive ‘euergetism’ is linked to the cults of the free and redistributes to the free, not the slaves.[12] A villein, equally, stands in a different relation to his or her lord to a slave not just because of an ‘economic’ relation, but because of a juridical relation given by the society as a whole (here ‘relative subordination’) and a religious relation (here, participation in the common universalistic cult and institutions of redistribution). A proletarian in modern capitalism is not a proletarian in the sense used in Marx’s political strategy (of proletarian self-organisation for political action leading to workers’ power) just because of the wage relation. Rather, she or he is ‘freed’ precisely because the other aspects of slavery and villeinage have been stripped away by capital’s action in its own interests to create its own freedom from the claims of pre-capitalist exploiters and clerisies, and of peasants and artisans, alike. This stripping away is a change in the public institutions.

Public institutions

As I said in the second article, the material division of labour in society, which is the ‘economic base’, cannot only consist of the private-choice or family-based institutions (private property, including slavery and villeinage, private jurisdictions, and so on) and forms of competition between families or firms (market competition, competition for a ‘following’, electoral competition in republican Rome, court factionalism wherever there is a real monarchy). It must also involve public institutions - which we can put very roughly into two classes: the state, and the (commonly religious) institutions of redistribution.

I concluded the second article with the point that, whatever their particular forms, private property, the family, and class division set up competition between families and reasons to aspire to you or your family climbing rather than sliding down the greasy pole. This set-up is therefore prima facie competitive and therefore a dynamic element in social order.

Private-choice economic practices can be experimented with on a small scale and copied, so that we see, as Banaji points out, forms of employment of wage-labour in classical antiquity and other pre-capitalist societies. As those on the other side from Banaji in the debates on antiquity have pointed out, we also see in classical antiquity forms both of attenuated slavery and of semi-free agricultural tenancy, which look like precursors of villeinage.

The state, and religious redistribution, are in contrast prima facie conservative institutions. This is necessary to their character: they are institutions of the whole society. They exist precisely to preserve the society as a whole from the fact that the free play of private-choice institutions will produce distributional changes which end in social collapse. Hence they cannot be subject to small-scale experimentation, but on the contrary have to attempt to preserve the social order from which they were born. This means that they are not merely passive: on the contrary, state and ‘established’ religious-redistributive actors actively resist social, economic, political and religious change.


It is for this reason that a full transition from one general class order (‘mode of production’) to another entails the actual overthrow of the state and the institutions of religious redistribution of the old order; and the construction of new state and redistributive institutions - or, at least, the disarticulation of the old institutions as a coherent entity and the rearticulation of some elements drawn from them into a new form. The creation of feudalism in Europe (over a prolonged period of false starts and transitional forms) required both the overthrow of paganism and urban euergetism (within the late Roman empire) and the overthrow of the state itself (in the western empire: Byzantium, clinging to the old state forms, hung on and shrank until the Turks delivered the coup de grâce in 1453) and the creation of separation of the clerisy from the state and the landlord class (a series of steps down to the Gregorian reform, without clear equivalents in eastern orthodoxy or Islam outside Ottoman Turkey). The creation of feudalism in Japan analogously required both the creation of warrior rule and the triumph of Buddhist monastic and redistributive organisations as dominant religious forms.

The transition to capitalism as something more than interstitial merchant-capitalism analogously required both the subordination of the clerisy and its redistributive activities to the state (Venetian state-controlled local patriarch, Dutch and British Protestantism, French secularism, Meiji state-controlled Shinto) and the overthrow of the legitimate monarchs, heads of the legitimate landlord classes, in favour of subordinated ‘constitutional monarchs’ (true even of the Meiji emperor, if what he was subordinated to was military-industrial-bureaucratic cabals) or republics.

Both the experimental-pragmatic or ‘selective’ quality of changes in the private-choice relations and the resistance of the public-choice institutions to change necessarily mean that there will be long periods of transition before a new class order becomes recognisable. First there will be experimentation within or at the boundaries of the old order; then, as these become more widespread and the old order becomes more problematic, attempts to maintain the old order by strengthening the state, resisting innovation and restoring an imagined image of the past.

The state and the religious-redistributive forms can long outlast the general class order that gave them their form. The later Roman empire and Byzantium forms the classic example of this point. When the state and religious forms were created, slave-taking, slave-trading and slave-holding were commonplace and extended well down the social spectrum. By the later empire, a variety of other forms had displaced this ‘classic slavery’; but it was not clear what would replace it. But equally, European absolutism and the Tokugawa regime, though they actively promoted feudal social hierarchies, were survivals of the feudal social regime, not actually grounded on currently dominant feudal relations of exploitation in the private-choice sphere.

Then the state and religious-redistributive carapace of the old order is broken, and experimentation with new private-choice relations can proceed more rapidly; but much still persists from the old order. Some experiments may prove to be blind alleys, like the Arian Gothic, Vandal and Burgundian kingdoms, the medieval city-states in general (Venice perhaps excepted) or Stalinism. Others are more fruitful, and we gradually begin to see the shape of a new social order emerge; but aspects of the old social order persist within the new, and it is only (as I suggested in the first part) at ‘apogee’ that the social order is fully visibly, distinctively and obviously ‘slave’, ‘feudal’ or ‘capitalist’.

Concrete politics

It is finally now hopefully possible to return very briefly from this high level of abstraction and (as it were) satellite’s eye view of human history, to the concrete political.

Banaji’s argument, as I said in the first part of the review, denies transitions - at least prolonged ones - and transitional forms. It rejects ‘teleology’ without arguing the point, it seems (as far as I can tell from the work) because a theory which gives directionality to history would imply transitions and transitional forms. The political function of this theoretical analysis is to analyse India as fully capitalist without significant pre-capitalist survivals and thereby demolish a priori the ‘official communist’ stages theory and the Maoist theory of peasant war.

Banaji’s alternative approach is labour organising: as he says towards the end of ‘The ironies of Indian Maoism’, “The bulk of the Indian labour force remains unorganised into unions, and it is stupefying to imagine that a revolution against capitalism can succeed while the mass of the workers are in a state of near-complete atomisation.”[13]

But suppose that there is directionality in history and there are transitions and transitional forms. It is then perfectly possible that the phenomena of the putting-out system or ‘formal subsumption of labour to capital’, debt-bondage, indentured labour and so on, which are found in late medieval through to 19th century Europe, and the reinvention of slavery as a colonial institution in the Americas, are indeed transitional forms: certainly, these forms have, in fact, been overthrown - and in countries which remained subordinate (as in Latin America and the Caribbean) as well as in imperialist countries like the UK and USA. Assume, then, that these forms are transitional. We then have to ask what their political implications are?

The first issue is the putting-out system and similar systems of ‘formal subsumption of labour to capital’. In this system, the producers remain formally owners of their means of production, but are in fact controlled by capitalist control of materials supplies, credit, and outlets for their products. The question posed in whether this regime produces what Marc Mulholland has called the proletarian ‘imaginaire’ which led Marx to suppose that the proletariat would tend towards collectivism.[14] The answer is fairly clearly that it does not; and, in fact, it is pretty clear from Marx and Engels’ comments on the urban ‘Straubinger’ of their own day that they did not think it would. The case is a fortiori of sharecroppers and similar strata in the countryside, who have some partial access to petty household cultivation.

Second is the other side of the rural coin: the persistence of pre-capitalist religious forms (in Hindu India, that of antique, pre-feudal, religious forms) and of landlord classes which struggle to maintain some form of juridical subordination of their workforce over and above the ‘dull compulsion of everyday life’ and self-identify as members of one or another sort of religious or martial elite. Where in world history have claims of this sort been overthrown without a jacquerie, peasants’ revolt, Bauernkrieg, or forcible suppression of the old order by external conquerors (as in US-imposed land reform in southern Korea)?

To say this is not to endorse the Naxalites against Banaji’s criticisms - or, for the reasons I gave in the first article, to endorse the ‘official communist’ policy of class alliance with ‘progressive’ capitalists. Banaji is undoubtedly correct that communist policy in India needs to begin with the organisation of the urban proletariat proper: in parties, unions, cooperatives, and so on. But precisely the difficulties of union organising imply - as they did for workers in many countries in the late 19th century - the centrality of political party organisation. And a political party cannot speak only to the concerns of the urban workers, but has to have things to say about ‘agrarian questions’. Banaji’s arguments seem to construct the conclusion that the problems of agrarian labour are simply identical to those of urban labour. It seems unlikely.



1. Except insofar as every report of an experiment or observation is a report of something which happened in the past and, contrary to Popper, all reasoning from such events is a species of induction. The elementary principles of historical source criticism (assessing biases of the witness, closeness to the event described, consistency of evidence, corroboration, antecedent probability of the narrative, and so on) are originally derived from legal approaches to evidence used in court of recent events, and the same approaches also form a substratum of the assessment of observational and experimental evidence in the physical sciences (see B Shapiro A culture of fact New York 2000, and my review of this book, 44 American Journal of Legal History pp445-46). This caveat should be understood wherever I refer to there being ‘non-historical’ evidence or arguments for historical materialism. In doing so my point is simply that this evidence comes from outside the usual purview of the historical profession.

2. Poverty of philosophy second observation: www.marxists.org/archive/marx/works/1847/poverty-philosophy/ch02.htm.

3. There is already considerable artificiality in the argument of Karl Marx’s theory of history, chapters 9 and 10; admissions in History, labour and freedom (Oxford 1988) at various points.

4. S Jay Gould The structure of evolutionary theory (2002).

5. ‘Contradiction’ is to use Hegelian language, but perhaps to overextend it.

6. JC Scott The art of not being governed: an anarchist history of upland south-east Asia New Haven 2009.

7. PR Hyams King, lords and peasants in medieval England Oxford 1980, pp157-59. Note also that Britton (c1290) ed Nichols, i, 195, says that the lord may not kill or maim his villein (unlike slaves in classical Athenian or Roman law) because the king has a competing property right in the villein’s person; Hyams pp135-43 finds 13th century judicial rulings at least in theory consistent with this.

8. ‘Proto-racist’: B Isaac The invention of racism in classical antiquity (Princeton 2004) is admirably but almost tediously cautious in analysing the sources.

9. Examples of later ideological claims to state ownership of land in JG Manning Land and power in Ptolemaic Egypt Cambridge 2003; Gaius Institutes Book 2 §§ 7, 14, 21 (on provincial land in the Roman empire); C Imber Ebu Su’ud (Edinburgh 1997) chapter 5 (on Ottoman land tenure as interpreted in Hanafi-school Islamic law).

10. Whether or how far the representation is accurate is, for present purposes, immaterial.

11. The class struggle in the ancient Greek world New York 1981, chapter 2.

12. P Veyne Bread and circuses London 1990.

13. (2010) ISJ No128 129-148, at 145.

14. ‘Marx, the proletariat and the will to socialism’ (2009) 37 Critique 319-343; ‘“Its patrimony, its unique wealth!” Labour-power, working class consciousness and crises’ (2010) 38 Critique 375-417.