Abstract: Marxist theory has by now largely abandoned the (seriously flawed) notion of the ‘mode of production’, but doing so has only encouraged a trend to abandon much of what was radical about it and naturalize capitalist categories. This article argues a better conceived notion of a mode of production – one that recognizes the primacy of human production, and hence a more sophisticated notion of materialism – might still have something to show us: notably, that capitalism, or at least industrial capitalism, has far more in common with, and is historically more closely linked with, chattel slavery than most of us had ever imagined.

What follows is really just the summary of a much longer argument I hope to develop at greater length elsewhere. A lot of the issues it addresses – the state of Marxist theory, the notion of the mode of production, worldsystems analysis – most anthropologists in the United States have come to think of as somewhat passé. I think though, if well employed, they can still tell us new and surprising things about the world we inhabit. The problem is that they haven’t always been deployed particularly well. This applies especially to the term ‘mode of production’ (MoP), which was in certain ways theoretically quite undeveloped. As a result, when world-systems analysis came along and changed the frame of reference, the concept simply collapsed. One might argue this wasn’t such an entirely bad thing, but the results have been disturbing. Almost immediately on jettisoning the modes of production model, once die-hard Marxists began seeing the market, or even ‘capitalism’, everywhere. Soon one had anthropologists like Jonathan Friedman arguing that ancient slavery is really just a form of capitalism. One could, of course, take the exact same evidence to make the argument precisely the other way around, and argue that modern capitalism is really just a form of slavery, but it never seems to occur to contemporary authors to do this. When even Marxists are naturalizing capitalism, you know there’s something seriously wrong.

Here I want to take a radically different tack. I want to argue that it might be possible to re-imagine the whole concept by seeing ‘modes of production’ not simply as about making and struggling over some kind of material surplus, but, equally, about the mutual fashioning of human beings – the process sometimes referred to in the Marxist tradition as ‘social production’. The moment one does so, all sorts of things leap into focus that might have otherwise remained obscure. For example, one of the most striking things about capitalism is that it is the only mode of production to systematically divide homes and workplaces: that is to say that the making of people and the manufacture of things should properly operate by an entirely different logic in places that have nothing to do with each other. In this, it actually does have certain striking similarities with slavery, so much so, in fact, that one could say that one is, in a certain sense, a transformation of the other.

Observation 1: The concept of the ‘mode of production’ was distinctly under-formulated

As others have noted, Marx used the term ‘mode of production’ quite casually, speaking not only of the capitalist or feudal MoPs but ‘patriarchal’ or ‘Slavonic’ ones, and so on. It only became a rigorous theoretical term when, in the 1950s, Louis Althusser seized on the term as a way of breaking out of the official, evolutionist model that had dominated official Marxism up to his day – that saw history everywhere as proceeding, mechanically, from slavery to feudalism to capitalism – without entirely alienating the very dogmatic French Communist Party of his day.

The resulting formulation, later developed by anthropologists like Meillassoux (1981) or Terray (1969), or historians like Perry Anderson (1974a, 1974b), runs something like this:

A mode of production (MoP) is born of the relation between two factors, the forces of production (FoP) and the relations of production (RoP). The former is largely concerned with factors like the quality of land, level of technological knowledge, availability of machinery and so on. The latter are marked by a relation between two classes, one a class of primary producers, the other an exploiting class. The relation between them is exploitative because, while the primary producers do in fact create enough to reproduce their own lives through their labors, and more to spare, the exploiting class does not, but rather lives at least in part on the surplus extracted from the primary producers. This extraction, in turn, is carried out through one or another form of property arrangement: in the case of slave mode of production, the exploiters directly own the primary producers; in feudalism, both have complex relations to the land, but the lords use direct jural-political means to extract a surplus; in capitalism, the exploiters own the means of production and the primary producers are thus reduced to selling their labor power. The state, in each case, is essentially an apparatus of coercion that backs up these property rights by force.

Societies, or ‘social formations’, as they were called, rarely involve just one MoP. There tends to be a mix. However, one will be predominant. And the exploiting class will be the ruling class, which dominates the state.

Finally, MoPs are assumed to be inherently unstable. Owing to their internal contradictions, they will eventually destroy themselves and turn into something else.

When one looks at actual analyzes, however, what we find is slightly different. For one thing, the ‘forces of production’ are rarely much invoked. Roman slavery and Haitian slavery involved completely different crops, climates, technologies and so on; but no one has ever suggested that they could not, for that reason, both be considered slavery. In fact, the ‘forces’ really only seem to be there in the theory as a gesture to certain passages in Marx, such as where he argues that slavery is a product of the hand mill and feudalism of the water mill (Marx, 1859/1970). So in effect this was just a theory of the social relations through which surpluses were extracted. Second, it proved quite difficult to break out of the evolutionary, Eurocentric mold. The division between slavery, feudalism and capitalism was clearly designed to describe class relations in ancient, medieval and modern Europe, respectively. It was never clear how to apply the approach to other parts of the world. Anthropologists found it especially difficult to figure out how to apply the model to stateless societies. While some coined phrases like the ‘lineage’ or ‘domestic’ mode of production, they never quite seemed to fit. Then there was the question of non-Western states. Marx had argued that empires like China or Mughal India were locked in a timeless ‘Asiatic’ mode of production that lacked the internal dynamism of Western states; aside from being extremely condescending, the way he formulated the concept turned out to be hopelessly contradictory (Anderson, 1974b). Attempts to create alternatives, like the ‘African MoP’ (Coquery-Vidrovitch, 1978), didn’t really catch on. So were all these states simply variations on feudalism, as so many communist parties insisted? Samir Amin (1973, 1985) tried to salvage the situation by proposing that pretty much all non-capitalist states should be subsumed into a single, much broader category, which he called the ‘tributary mode of production’. This, he suggested, would include any system in which the surplus was extracted through political-coercive means. Centralized states like Sung China or the Sassanian empire could be considered highly organized examples; feudalism, as practiced in Europe and perhaps Japan, one particularly disorganized variant. In Europe and the People without History (1982), Eric Wolf took this further in proposing three broad MoPs: the kinship mode of production, which encompassed those stateless societies that were the traditional stomping-grounds of anthropologists; the tributary mode; and then finally capitalism itself. But at this point the concepts had become so diffuse that it became impossible to think of a social formation as a complex mix of different modes of production, except insofar as each new stage incorporated the previous ones, e.g. under tributary states there was still kinship, and under capitalism, state apparatuses that made war and levied taxes (which can be thought of as a continuation of the tributory mode).

Observation 2: The concept of the ‘mode of production’ largely dissolved when removed from the framework of the state

Back in 1974, when Perry Anderson sounded the death-knell of the ‘Asiatic mode’, he called for work to create new concepts to describe states like India or China. One might have imagined this would have been answered by an outpouring of new proposed modes of production. Instead what happened was almost exactly the opposite. The list kept getting shorter and shorter. By the early 1980s, in Wolf, we were back to exactly the same kind of three-part evolutionary sequence Althusser was originally trying to escape – the main difference being that ‘slavery’ had been replaced by ‘kinship’. How could this happen?

Wolf’s book was the first major work of anthropology to try to come to grips with the kind of world-systems analysis being developed by Immanuel Wallerstein and others at the time, and I don’t think this is insignificant. One reason for the collapse of the MoP approach was that it was essentially a theory of the state. For all the fancy terminology, ‘social formations’ just about always turned out to be kingdoms or empires of one sort or another. Hence the theory was thrown into a profound crisis when world-systems analysis completely transformed the unit of analysis. At first this was not entirely clear, because the arguments were mainly about capitalism. Proponents of the mode of production approach insisted that capitalism first emerged from the internal class dynamics of individual states, as wage-labor relations gradually became predominant, ultimately leading to a point where the bourgeoisie could seize control of the state apparatus (as in the English or French revolutions). Wallerstein argued it emerged in the form of a ‘capitalist world-economy’, a broader system of market relations that created an overall division of labor between regions (differentiating a core, periphery and semi-periphery). According to the world-systems approach, what went on within any particular ‘society’ – for example, the rise of wage labor – could only be explained with reference to that larger system.

In principle, this is true of all world-systems – called this not because they encompassed the entire globe, since only capitalism has done that, but because they were spheres of regional interaction that were, in effect, worlds unto themselves.

The holistic emphasis made it impossible simply to substitute ‘worldsystem’ for ‘social formation’ and still argue that any world-system contains a number of different modes of production, of which only one will be dominant. World-systems are assumed to be coherent wholes. As a result, ‘capitalism’ or ‘feudalism’ came to be seen as overall modes of organization for these new, larger, units.

Wallerstein originally proposed three different sorts of world-system, in a formulation that looked suspiciously like yet another of those three-part evolutionary sequences: ‘mini-systems’ (self-sufficient, egalitarian societies), ‘world-empires’ (such as the Achmaenid or Chinese), and ‘world-systems’ linked by trade (which, prior to capitalism, tended eventually to transform into empires, then, usually dissolve). In part the categories were inspired by the Hungarian economist Karl Polanyi’s (1957) distinction between three modes of distribution of wealth: reciprocity (typical of mini-systems), redistribution (typical of empires), and the market (typical of worldsystems). Wallerstein was careful to note that all this was meant as a mere first approximation, to stand as a basis for research until better terms were found, so perhaps it’s not right to make too much of these terms, but one thing stands out. Each was distinguished not by a form of production, but a form of distribution. And it was this larger organization of distribution that gave shape to everything else within each particular universe. This actually suggested a very daunting project of cultural comparison. Since Wallerstein argued that almost all our familiar categories of analysis – class, state, household and so on – are really only meaningful within the existing capitalist world-system, then presumably entirely new terms would have to be invented to look at other ones. If so, then what did different worldsystems have in common? What was the basis for comparison?

Subsequent divisions turn largely on this question. One school of world-systems theorists – the ‘Comparativists’, whose most prominent exponents are Chase-Dunn and Hall (1997) – have tried to refine the terms. First, they had to ditch the notion of mini-systems (basically ‘tribes’), by demonstrating that even in the case of extremely egalitarian societies like that of the Wintu of southern California, there were always regional spheres of interaction, ‘very small world-systems’ as they call them. These smaller systems, though, seemed to lack the cycles of growth and collapse typical of larger, more hierarchical systems like markets and empires. Larger worldsystems, they proposed, tended to be made up of a complex series of overlapping networks; but in the end, the overall organization of all these systems still ends up falling into Wolf’s three categories: kinship, tributary and capitalist (plus one hypothetical socialist one that does not yet exist but might some day). The main difference with Wolf is that they tend to refer to these not as ‘modes of production’ but as ‘modes of accumulation’, which they define as ‘the deep structural logic of production, distribution, exchange, and accumulation’ (1997: 29). It seems a reasonable change in terminology from a world-systems perspective. But it lays bare just how far the term had drifted from its supposed original focus on people actually making things.

Once the terms of comparison have been made this broad, it’s really just a short hop to arguing that we are not dealing with terms of comparison at all, but different functions that one would expect to find in any complex social order. This was the move taken by the ‘Continuationists’ – the prominent names here are Andre Gunder Frank and Barry Gills (Frank, 1993, 1998; Frank and Gills, 1993), Jonathan Friedman and Kajsa Eckholm (Eckholm and Friedman, 1982; Friedman, 1982, 2000) – who argue that, just as any complex society will still have families (‘kinship’), they will also tend to have some sort of government, which means taxes (‘tribute’) and some sort of market system (‘capitalism’). Having done this, it’s easy enough to argue that the very project of comparison is pointless. In fact, there is only one world system. It began in the Middle East some 5000 years ago and fairly quickly came to dominate Afro-Eurasia; for the last couple of thousand years, at least, its center of gravity has been China. According to Gunder Frank, this world system (note, no hyphens now) has seen broad but regular cycles of growth and expansion. This is the basis for his notoriously provocative claim that not only was Europe for a long time a barbarous periphery to the dominant world system – in itself actually a fairly uncontroversial observation by now – but that European dominance in recent centuries was really only the result of a successful campaign of import substitution during a time when the rest of the world system was in its periodic downswing, and that now that it’s time for the boom end of the cycle to reassert itself, the dominance of ‘the West’ may well prove a merely passing phase in a very long history (Frank, 1998).

Observation 3: The main result of the eclipse of the mode of production concept has been a naturalization of capitalism; this becomes particularly evident when looking at the way ‘Continuationists’ treat wage labor and slavery

Friedman, Eckholm and others now openly talk of a capitalist world system that has existed for 5000 years (Andre Gunder Frank would prefer to discard the term ‘capitalism’ entirely, along with all other ‘modes of production’ [1991], but what he describes comes down to pretty much the same thing). The idea that capitalism is as old as civilization is of course a position long popular among capitalists; what now makes it palatable on the Left is largely that it can be seen as an attack on Eurocentrism: if capitalism is now to be considered an accomplishment, then it is deeply arrogant of Euro-American scholars to assume that Europeans invented it a mere 500 years ago. Alternatively, one might see this as a position appropriate for Marxist scholars working in an age when anarchism is rapidly replacing statist ideologies as the standard-bearer of revolutionary struggle: if capitalism appeared together with the state, it would be hard to imagine eliminating one without the other. The problem of course is that, defined so broadly, it becomes very hard to imagine eliminating capitalism at all.

Neither does this position eliminate the privileged position of Europe if you really think about it, because even if the Continuationists argue that the 17^th and 18^th centuries did not witness the birth of capitalism in Western Europe, and thus did not mark some great economic breakthrough, they are still arguing that it marked an equally momentous intellectual breakthrough, with Europeans like Adam Smith discovering the existence of economic laws that they now claim had existed for thousands of years in Asia and Africa but that no one there had previously been able to describe or even, really, notice.

This is actually a more important point than it may seem. The great enemies of the Continuationists are mid-century scholars like Moses Finley (1960a, 1960b, 1973), and Karl Polanyi (1944, 1957, 1968), who argued that authors in ancient and non-Western societies really did understand what was going on in their own societies, and that if they did not speak of something that could be labeled ‘the economy’ it was because nothing exactly parallel to capitalist economic institutions existed. Both come in for particular denunciation and abuse: apparently, for that very reason.

Let me illustrate something of what’s at stake here. Typically, definitions of capitalism focus on one of two features. Some, as in the MoP approach, focus on wage labor. The Continuationists, predictably, prefer the other focus, which looks for the existence of capital: that is, concentrations of wealth employed simply to create more wealth, and in particular, with the open-endness of the process, the drive for endless reinvestment and expansion. If one chose the first, it would be hard to say capitalism has always existed, since, for most of human history, it’s rather difficult to find much that can be described as wage labor. This is not for lack of trying. Continuationists – like most economic historians, actually – tend to define ‘wages’ as broadly as possible: essentially, as any money given to anyone in exchange for services. If you actually spell it out, the formulation is obviously absurd: if so, kings are wage-laborers insofar as they claim to provide protection in exchange for tribute, and the Aga Khan is currently a wage laborer in the employ of the Ismaili community because every year they present him with his weight in gold or diamonds to thank him for his prayers on their behalf. Clearly, ‘wage labor’ (as opposed to, say, fees for professional services) involves a degree of subordination: a laborer has to be to some degree at the command of his or her employer. This is exactly why, through most of history, free men and women tended to avoid it, and why, for most of history, capitalism according to the first definition never emerged.

As Moses Finley noted (1973), the ancient Mediterranean world was marked by a strong feeling of contradiction between political and commercial life. In Rome, most bankers were freed slaves; in Athens, almost all commercial and industrial pursuits were in the hands of non-citizens. The existence of a huge population of chattel slaves – in most ancient cities apparently at least a third of the total population – had a profound effect on labor arrangements. While one does periodically run into evidence of arrangements that, to the modern eye, look like wage labor contracts, on closer examination they almost always actually turn out to be contracts to rent slaves (the slave, in such cases, often received a fixed per diem for food). Free men and women thus avoided anything remotely like wage labor, seeing it as a matter, effectively, of slavery, of renting themselves out (Humphries, 1978: 147, 297n37–8). Working for the city itself could sometimes be considered acceptable, since one was effectively in the employ of a community of which one was oneself a member, but even this was normally kept to a temporary contract basis. In Periclean Athens, permanent employes, even state employes such as police, were invariably slaves.

All this was hardly unique. Remarkably similar things have been documented in, say, 19^th-century Madagascar or Brazil. Reflection on the implications of the idea of renting persons might yield all sorts of insights; similarly, one could consider how institutions that might look to us remarkably like wage labor relations – in that one party worked and another compensated them in some way – might really have had a completely different basis: extended ties of patronage and dependency for example, those complicated statuses that Finley (1964) described as hovering ‘between slave and free’. But, for the Continuationists, as for most economic historians, all this is brushed aside. Friedman (2000), for example, accuses Polanyi, Finley and their followers as being driven by ‘ideological’ motivations in denying the importance of capital and markets in the ancient world. After all, what the actors thought they were doing is largely irrelevant. Capitalism is not a state of mind but a matter of objective structures, which allow wealth and power to be translated into abstract forms in which they can be endlessly expanded and reproduced. If one were to make an objective analysis, one would have to start from the fact that wage laborers, even if they were of servile status, did exist, that they produced objects for sale on the market and that the whole system evinced just the sort of boom–bust cycle structure we’re used to seeing in capitalism. He concludes:

... slavery in Classical Greece is a complex affair involving wage, interest and profit in an elaborate market system that appears to have had cyclical properties of expansion and contraction. This was, in other words, a form of capitalism that is not so different from the more obvious varieties in the modern world. (2000: 152)

For all the pretensions of objectivity though, it’s hard to see this choice as any less ideological. After all, one can define ‘capitalism’ as broadly or narrowly as one likes. It would be easy enough to play the same trick with terms like socialism, communism and fascism, and define them so broadly one could discover them all over ancient Greece or Safavid Persia. Yet somehow no one ever does. Alternatively, one could just as easily turn Friedman’s own example around, define ‘capitalism’ as based on free wage labor, but define ‘slavery’ in the broadest terms possible: say, as any form of labor in which one party is effectively coerced. One could thereby conclude that modern capitalism is really a form of slavery. (One could then go on to argue that the fact that modern capitalists don’t see themselves as coercing others is irrelevant, since we are talking about objective constraining structures and not what the actors think is going on.) Such an argument would not be entirely unprecedented: there’s a reason why so many workers in modern capitalist countries have chosen to refer to themselves as ‘wage slaves’. But no economic historian has ever, to my knowledge, even suggested such a thing. The ideological biases become clear when one considers not just what’s being argued, but the arguments it would never occur to anyone to make.

Thesis 1: The key mistake of the mode of production model was to define ‘production’ simply as the production of material objects; any adequate theory of ‘production’ would have to give at least equal place to the production of people and social relations

The ultimate weakness of MoP approaches, it seems to me, is that they begin from a very naive sort of materialism. ‘Material production’ is assumed to be the production of valuable material objects like food, clothing or gold bullion; all the important business of life is assumed to be moving such objects around and transferring them from one person or class to another.

The approach is usually attributed to Marx – indeed, ‘historical materialism’ of this sort is about the only aspect of Marx’s thought that scholars like Gunder Frank claim is really salvageable (e.g. Frank and Gills, 1993: 106–9). Now, I really don’t see the point of entering into some prolonged debate about whether this represents what Marx ‘really’ meant when he talked about ‘materialism’. Marx’s work, it seems to me, pulls in any number of different directions. But some are decidedly more interesting. Consider this passage from his ethnographic notebooks:

Among the ancients we discover no single inquiry as to which form of landed property etc. is the most productive, which creates maximum wealth. Wealth does not appear as the aim of production, although Cato may well investigate the most profitable cultivation of fields, or Brutus may even lend money at the most favorable rate of interest. The inquiry is always about what kind of property creates the best citizens. Wealth as an end in itself appears only among a few trading peoples – monopolists of the carrying trade – who live in the pores of the ancient world like the Jews in medieval society....

Thus the ancient conception, in which man always appears (in however narrowly national, religious or political a definition) as the aim of production, seems very much more exalted than the modern world, in which production is the aim of man and wealth the aim of production. In fact, however, when the narrow bourgeois form has been peeled away, what is wealth, if not the universality of needs, capacities, enjoyments, productive powers, etc., of individuals, produced in universal exchange? (1965[1854]: 84)

What Marx says here of the ancient Greeks and Romans could, clearly, equally well go for the 15^th century BaKongo kingdom, or the medieval city of Samarkand, or pretty much any non-capitalist society. Always, the production of wealth was seen not as an end in itself but as one subordinate moment in a larger process that ultimately aimed at the production of people. Neither does he suggest that this was just a subjective illusion that we have only learned to see through now that we have developed the science of economics; rather, it is quite the other way around. The ancients had it right. In The German Ideology, Marx had already suggested that the production of objects is always simultaneously the production of people and social relations (as well as new needs: 1970[1846]: 48–50). Here, he observes that the objects are not ultimately the point. Capitalism and ‘economic science’ might confuse us into thinking that the ultimate goal of society is simply the increase of national GDP, the production of more and more wealth, but in reality wealth has no meaning except as a medium for the growth and self-realization of human beings.

The question then becomes: what would a ‘mode of production’ be like if we started from this Marx, rather than, say, the Marx of the Contribution to a Critique of Political Economy? If non-capitalist modes of production are not ultimately about the production of wealth but of people – or, as Marx emphasizes, of certain specific kinds of people – then it’s pretty clear that existing approaches have taken entirely the wrong track. Should we not be examining relations of service, domestic arrangements, educational practices, at least as much as the disposition of wheat harvests and the flow of trade?

I would go even further. What has passed for ‘materialism’ in traditional Marxism – the division between material ‘infrastructure’ and ideal ‘superstructure’ – is itself a perverse form of idealism. Granted, those who practice law, or music, or religion, or finance, or social theory, always do tend to claim that they are dealing with something higher, more abstract, than those who plant onions, blow glass or operate sewing machines. But it’s not really true. The actions involved in the production of law, poetry, etc., are just as material as any others. Once you acknowledge the simple dialectical point that what we take to be self-identical objects are really processes of action, then it becomes pretty obvious that such actions are always (a) motivated by meanings (ideas) and (b) always proceed through a concrete medium (material), and that while all systems of domination seem to propose that ‘No, this is not true, really there is some pure domain of law, or truth, or grace, or theory, or finance capital, that floats above it all’, such claims are, to use an appropriately earthy metaphor, bullshit. As John Holloway (2003) has recently reminded us, it is in the nature of systems of domination to take what are really complex interwoven processes of action and chop them up and redefine them as discrete, self-identical objects – a song, a school, a meal, etc. There’s a simple reason for it. It’s only by chopping and freezing them in this way that one can reduce them to property and be able to say one owns them.

A genuine materialism, then, would not simply privilege a ‘material’ sphere over an ideal one. It would begin by acknowledging that no such ideal sphere actually exists. This, in turn, would make it possible to stop focusing so obsessively on the production of material objects – discrete, selfidentical things that one can own – and start the more difficult work of trying to understand the (equally material) processes by which people create and shape one another.

Thesis 2: If one applies Marx’s analysis of value in capital to the production of people and social relations, one can more easily see some of the mechanisms which obscure the most important forms of labor that exist in most societies, and the real stakes of human existence, thus allowing ‘scientific’ observers to reduce human beings to automatons competing over abstractions like ‘wealth’ or ‘power’

It might be easier to understand what I’m getting at here by considering the work of some anthropologists who have taken roughly the approach I’m endorsing.

I’m referring here to the tradition of what I’ll call ‘anthropological value theory’. Such theory was made possible first and foremost by the insights of feminist social science, which has made it impossible to simply ignore the endless labor of care, maintenance, education and so on, which actually keeps societies running and which has tended to be carried out overwhelmingly by women. Recognizing such forms of action as productive labor, in the Marxian sense, made it easier to see how Marx’s insights might be applied to many of the more egalitarian, stateless societies the MoP approach finds so difficult to deal with. The real pioneer here is Terry Turner (1979, 1984, 1987), with his work on the Kayapo, though there are a number of others working along similar lines (e.g. Myers on the Pintupi [1986], Munn on Gawa [1986], Fajans on the Baining [1997], Sangren on rural Taiwan [1987, 2000], etc.) I have tried to systemize some of their insights myself in a book called Toward an Anthropological Theory of Value (Graeber, 2001).

This approach does, indeed, take it for granted that while any society has to produce food, clothing, shelter and so forth, in most societies the production of such things as houses, manioc, canoes is very much seen as a subordinate moment in larger productive processes aimed at the fashioning of humans. True, the former varieties of production tend to involve physical constraints that are very real and important to take into account. But that does not mean they are simply matters of technical activity. Anthropologists have demonstrated time and time again that even such apparently mundane activities as building or moving about in a house (Bourdieu, 1979) or producing manioc flour (Hugh-Jones, 1979) encode symbolic structures – hot/cold, dry/wet, heaven/earth, male/female – which tend to recur as well in complex rituals, forms of artistic expression or conceptions of the nature of the cosmos as a whole, but which are, ultimately, embedded in those very structures of action themselves. In other words we are never dealing with pure, abstract ideas, any more than we are ever dealing with purely mechanical production. Rather, the very idea that either pure ideas or mindless material action exist is an ideology whose operations need to be investigated.

The latter is an important point because many such societies do make something like this sort of ideal/material distinction, even if it rarely takes exactly the same form. This seems directly related to the fact that just about invariably, some form of exploitation does occur in such societies; and where it does, much as in capitalism, the mechanisms of exploitation tend to be made subtly invisible.

In Marx’s account of capitalism, this happens mainly through the mechanism of wage labor. Money is in fact a representation of abstract labor – the worker’s capacity to produce, which is what his employer buys when he hires him. It is a kind of symbol. In the form of a wage, it becomes a very powerful sort of symbol: a representation which in fact plays a crucial role in bringing into being what it represents – since after all, laborers are only working in order to get paid. It’s also in precisely this transaction that the actual sleight-of-hand on which exploitation is based takes place, since Marx argues that what the capitalist ends up paying for is simply the cost of abstract labor (the cost of reproducing the worker’s capacity to work), which is always going to be less than the value of what the worker can actually produce.

The point Turner makes is that even where there is no single market in labor – as there has not been in most societies in human history – something similar tends to happen. Different kinds of labor still tend to get reflected back in the form of a concrete, material medium, which, like money, is both a representation of the importance of our own actions to ourselves, and simultaneously seen as valuable in itself, and which thus ends up becoming the actual end for which action takes place. Tokens of honor inspire honorable behavior. Really, their value is just that of the actions they represent, but the actors see them as valuable in themselves. Similarly, tokens of piety inspire religious devotion; tokens of wisdom inspire learning and so on. Actually, it’s quite the same in our own society; it’s precisely in those domains of activity where labor is not commoditized that we talk not of abstract ‘value’ but concrete ‘values’. For example, housework and childcare become a matter of ‘family values’, work for the church a matter of religious values, political activism is inspired by the values of idealism, and so on. In either case, certain basic principles seem to apply:

1. Value is the way actors represent the importance of their own actions to themselves as part of some larger whole (or ‘concrete totality’, as Marx liked to put it).

2. This importance is always seen in comparative terms: some forms of value are considered equivalent because they are unique, but normally there are systems of ranking or measurement.

3. Values are always realized through some kind of material token, and generally, in some place other than the place it is primarily produced. In non-capitalist societies this most often involves a distinction between a domestic sphere, in which most of the primary work of people-creation takes place, and some kind of public, political sphere, in which it is realized, but usually in ways which exclude the women and younger people who do the bulk of the work and allow tokens of value to be realized.

Thus Turner argues that among the Kayapo of central Brazil communities are organized as circles, with a ring of households surrounding a public, political space in the center. Forms of value produced largely in the domestic units through the work of producing and socializing people come to be realized through certain forms of public performance (chanting, oratory, keening, etc.) which are extended to elders who are themselves only ‘elders’ because they are the peak of a domestic process of creating and socializing children that takes place just offstage.

This emphasizes that just about always this process of realization of value involves some form of public recognition, but this is not to say that people are simply battling over ‘prestige’; instead, the range of people who are willing to recognize certain forms of value constitutes the extent of what an actor considers ‘society’, in any meaningful sense of the term, to consist of (Graeber, 2001).

What I especially want to stress here though is that, when value is about the production of people, it is always entirely implicated in processes of transformation: families are created, grow and break apart; people are born, mature, reproduce, grow old and die. They are constantly being socialized, trained, educated, mentored towards new roles – a process which is not limited to childhood but lasts until death; they are constantly being attended to and cared for. This is what human life is mainly about; what most people have always spent most of their time worrying about; what our passions, obsessions, loves and intrigues tend to center on; what great novelists and playwrights become famous for describing; what poetry and myth struggle to come to terms with; but which most economic and political theory essentially makes disappear.

Why? It seems to happen, at least in part, because of the very mechanics of value realization. Value tends to be realized in a more public, or anyway political, and hence universalized domain than the domestic one in which it is (largely) created; that sphere is usually treated as if it is to some degree transcendent, that is, as floating above and unaffected by the mundane details of human life (the special domain of women), having to do with timeless verities, eternal principles, absolute power – in a word, of something very like idealist abstractions. Most anthropological value analyzes end up tracing out something of the sort: so Kayapo value tokens end up embodying the abstract value of ‘beauty’, a profound higher unity and completion especially embodied in perfect performances and communal ritual (e.g. Turner, 1987); people practicing kula exchange seek ‘fame’ (Munn, 1986); Berbers of the Morroccan Rif, with their complex exchanges of gifts and blood-feud, pursue the values of honor and baraka, or divine grace ( Jamous, 1981) and so on. All of these are principles which, even when they are not identified with superhuman powers like gods or ancestors, even when they are not seen as literally transcendental principles, are seen as standing above and symbolically opposed to the messiness of ordinary human life and transformation. The same is usually true of the most valued objects, whose power to enchant and attract usually comes from the fact that they represent frozen processes; if one conducts a sufficiently subtle analysis, one tends to discover that the objects that are the ultimate stakes of some field of human endeavor are, in fact, symbolic templates which compress into themselves those patterns of human action which create them (e.g. Battaglia, 1983, 1990).

It seems to me that even beyond the labor that is constantly creating and reshaping human beings, a key unacknowledged form of labor in human societies is precisely that which creates and maintains that illusion of transcendence. In most, both are performed overwhelmingly by women. A nice way to illustrate what I’m talking about here might be to consider the phenomenon of mourning. Rarely do the political careers of important individuals end in death. Often political figures, as ancestors, martyrs, founders of institutions, can be far more important after their death than when they were alive. Mourning, and other acts of memorialization, could then be seen as an essential part of the labor of people-making – with the fact that the dead person is no longer himself playing an active role simply underlining how much of the work of making and maintaining a career is always done by others. Even the most cursory glance at the literature shows that the burden of such labor, here, tends to be very unevenly distributed. This is in fact especially true of the most dramatic forms – cutting off one’s hair, self-mutilation, fasting, wearing drab clothes or sackcloth and ashes, or whatever is considered the culturally appropriate way to make oneself an embodiment of grief, as, essentially, negating oneself to express anguish over the loss of another. Social subordinates mourn their superiors and not the other way around. And pretty much everywhere, the burden of mourning falls disproportionately, and usually overwhelmingly, on women. In many parts of the world, women of a certain age are expected to exist largely as living memorials to some dead male: whether it be Hindu widows who must renounce all the tastiest foods, or Catholic women in the rural Mediterranean who are likely to spend at least half their lives wearing black. Needless to say these women almost never receive the same recognition when they die, least of all from men.

The point though is that symbolic distinctions between high and low do not come from some preexisting ‘symbolic system’; they are continually constructed in action, and the work of doing so is done disproportionately by those who are effectively defining themselves as lower. So with mourning. As Bloch (1982) has emphasized, mourning is also about creating dramatic contrasts between what is considered truly permanent and everything that is corporeal, transitory, afflicted with the possibility of grief and pain, subject to corruption and decay. Mourners, when they cover themselves in dirt or ashes, or engage in other practices of the negation of the self, which seem surprisingly similar across cultures, are also making themselves the embodiment of the transitory, bodily sphere as against another, transcendental one, which is in fact created in large part through their doing so. The dead themselves have become spirits; they are ethereal beings or bodiless abstractions, or perhaps they are embodied in permanent monuments like tombs or beautiful heirlooms, or buildings left in their memory – usually, in fact, it’s a bit of both – but it’s the actions of the mourners, mainly by the dramatic negation of their own bodies and pleasures, that constantly recreate that extremely hierarchical contrast between pure and impure, higher and lower, heaven and earth.

It is sometimes said that the central notion of modernism is that human beings are projects of self-creation. What I am arguing here is that we are indeed processes of creation, but that most of the creation is normally carried out by others. I am also arguing that almost all the most intense desires, passions, commitments and experiences in most people’s lives – family dramas, sexual intrigue, educational accomplishment, honor and public recognition, one’s hopes for one’s children and grandchildren, one’s dreams of posterity after one is dead – have revolved precisely around these processes of the mutual creation of human beings, but that the mechanics of value creation tend to disguise this by positing some higher sphere, whether of economic values, or idealist abstractions. This is essential to the nature of hierarchy (Graeber, 1997) and the more hierarchical the society, the more this tends to happen. Finally, I am suggesting that it is precisely these mechanisms that make it possible for historians and social scientists to create such odd simplifications of human life and human motivations. The labor of creating and maintaining people and social relations (and people are, in large measure, simply the internalized accretion of their relations with others) ends up being relegated, at least tacitly, to the domain of nature – it becomes a matter of demographics or ‘reproduction’ – and the creation of valuable physical objects becomes the be-all and endall of human existence.

Thesis 3: One of the great insights of world-systems analysis is to show how very simple forms of social relation most typical of long-distance relations between people who do not know much about each other are continually introjected within those societies to simplify social relations that need not be that way

Unfortunately, this thesis can’t really be adequately explained, let alone defended, in the space available so let me just summarize it.

Marx was already noting in the passage cited above that commercial relations, in which wealth was the main aim of human activity, appears ‘in the pores of the ancient world’, among those who carry out the trade between societies. This is an insight developed in world-systems analysis, where capitalism is often seen as having developed first in long-distance trading and then gradually wormed its way into ever-more-intimate aspects of communities’ daily life. I would suggest we are dealing here with a much more general principle. One could name a whole series of highly schematic, simplified forms of action, that might be inevitable in dealings between people who don’t understand each other very well, that become introjected in a similar way. The first is probably violence. Violence is veritably unique among forms of action because it is pretty much the only way one can have relatively predictable effects on others’ actions without understanding anything about them. In any other way one might wish to influence others, one has to at least know or figure out who they think they are, what they want, find objectionable, etc. Hit them over the head hard enough, it all becomes irrelevant. Hence it is common to relations between societies, even those not marked by elaborate structural violence within. However, the existence of structural violence – social hierarchies backed up by a systematic threat of force – almost invariably creates forms of ignorance internally: it is no longer necessary to carry out this sort of interpretive work and, generally speaking, those on the top know remarkably little about what those on the bottom think is going on. Here, again, gender relations are probably the most revealing example: with remarkable consistency, across a very wide range of societies, men tend to know almost nothing about women’s lives, work or perspectives, while women tend to know a great deal about men’s – in fact, they are expected to, since a large share of that interpretive labor (if one may call it that) always seems to fall to women, which in turn helps explain why this is not generally considered ‘labor’ at all. And the same tends to apply to relations of caste, class and other forms of social inequality.

Market exchange is another case in point. It’s enough to take a glance at the rich anthropological literature on ‘gift exchange’, or even consider the way objects move within families or circles of friends, to realize how incredibly stripped-down and simplified is a standard commodity transaction in comparison. One need know almost nothing about the other party; all one needs to know is a single thing they want to acquire: gold, or fish, or calicoes. Hence the popularity, in early Greek or Arab travelers’ accounts, of the idea of the ‘silent trade’: in theory, it would be possible to engage in commercial exchange with people about whom one knew nothing at all, whom one never even met, by alternately leaving goods on a beach. The point is again that commercial relations were in many societies typical of relations with foreigners, since they required minimal interpretive work; in dealing with those one knew better, other, more complex forms of exchange usually applied; however, here too, the introjection of commercial relations into dealings with one’s neighbors made it possible to treat them, effectively, like foreigners. Marx’s analysis of capitalism actually gives a central role to this phenomenon: it is a peculiar effect of the market to erase the memory of previous transactions and create, effectively, a veil of ignorance between sellers and buyers, producers and consumers. Those who purchase a commodity usually have no idea who made it and under what conditions it was made; this is of course what results in ‘commodity fetishism’.

Thesis 4: If one reinterprets a ‘mode of production’ to mean a relation between surplus extraction and the creation of human beings, then it is possible to see industrial capitalism as an introjected form of the slave mode of production, with a structurally analogous relation between workplace and domestic sphere

If the notion of ‘mode of production’ is to be salvaged, it has to be seen not merely as a structure for the extraction of some kind of material surplus between classes, but as the way in which such a structure articulates with structures for the creation of people and social relations.

We might start here with the capitalist mode of production, since this was always the case from which the others were extrapolated. As I’ve mentioned, definitions of capitalism tend to start either from exchange or production. In the first case, one tends to see what makes capitalism unique as lying in the unlimited need for growth: where most systems of market exchange are full of actors trying to get what they feel they want or need, capitalism occurs when profit becomes an end in itself and ‘capital’ becomes like a living entity, which constantly seeks to expand; indeed, capitalist firms cannot remain competitive unless they are continually expanding. In the second, the emphasis is on wage labor: capitalism occurs when a significant number of firms are owned or managed by people who hire others to do their bidding in exchange for a direct payment of money, but otherwise have no stake in the enterprise. In the industrial capitalism described by Marx, the two appear together, and are assumed to be connected. I would propose a third. The Industrial Revolution also introduced the first form of economic organization to make a systematic distinction between homes and workplaces, between domestic and economic spheres. (This is what made it possible to begin talking about ‘the economy’ to begin with: the production of people and the production of commodities were to take place in different spaces by entirely different logics.) This split plays a central role in Marx’s analysis as well: for one thing, the market’s veil of ignorance falls precisely between the two. All this was in dramatic contrast to what had existed previously in most of Europe, where very complex systems of ‘life-cycle service’ (Hajnal, 1965, 1982; Laslett, 1972; Wall, 1983) ensured that the majority of young people spent years as apprentices or servants in the households of their social superiors. Once one recognizes this, the similarities with slavery become much easier to see.

I should explain here that the conventional Marxian interpretation of slavery as a mode of production is that slavery makes it possible for one society to effectively steal the productive labor that another society has invested in producing human beings (Lovejoy, 2000; Meillassoux, 1975, 1979, 1991; Terray, 1969). That’s why slaves always have to come from somewhere else (it is only under extraordinary conditions, such as the Southern US cotton boom created by the British Industrial Revolution, that it is economically viable to breed slaves, and even there it was not really sustainable). Human beings, after all, are largely useless as laborers for the first 10 or 15 years of their existence. A slave-owning society is effectively appropriating the years of care and nurture that some other society has invested in creating young men and women capable of work, by kidnapping the products – and then, often as not, working them fairly rapidly to death.

In a way, then, one could say that slavery too involves a separation of domestic sphere and workplace – except in this case the separation is geographic. Human labor produced in Anatolia is realized in a plantation in Italy; human labor produced in what’s now Gabon is realized in Brazil or Jamaica. In this sense, capitalism could be seen as yet another case of introjection. This might seem far-fetched; but in fact the structural similarities are actually quite striking.

The institution of slavery is normally seen to derive from war. If the victor in war spares the life of a captive, he thereby acquires an absolute right to it. The result is often described as a ‘social death’ (e.g. Patterson, 1982): the new slaves are spared literal execution, but henceforth they are also shorn of all previous status within their former communities, they have no right to social relations, no right to kinship or citizenship, or any social relation in fact other than their relation of dependence to a master who thus has the right to order them to do pretty much anything he wants. Now, there have been cases where this is all there is to it, but in the overwhelming majority of known historical cases, this process is mediated by the market. Normally, one is first captured, kidnapped or perhaps reduced to slavery by judicial decision; and then one is sold to foreigners; or perhaps one’s impoverished or debt-ridden parents sell one off directly, but at any rate, money changes hands. Afterwards, slaves remain marketable commodities that can be sold again and again. Once purchased, they are entirely at the orders of their employers. In this sense, as historian Yann MoulierBoutang (1998) has pointed out, they represent precisely what Marx called ‘abstract labor’: what one buys when one buys a slave is the sheer capacity to work, which is also what an employer acquires when he hires a laborer. It’s of course this relation of command that causes free people in most societies to see wage labor as analogous to slavery, and hence to try as much as possible to avoid it.

We can observe the following traits shared by slavery and capitalism:

1. Both rely on a separation of the place of social (re)production of the labor force, and the place where that labor-power is realized in production – in the case of slavery, this is effected by transporting laborers bought or stolen from one society into another one; in capitalism, by separating the domestic sphere (the sphere of social production) from the workplace. In other words, what is effected by physical distance in one is effected by the anonymity of the market in the other.

2. The transfer is effected through exchanging human powers for money: either by selling workers, or hiring them (essentially, allowing them to rent themselves).

3. One effect of that transfer is ‘social death’, in the sense that the community ties, kinship relations and so forth that shaped the worker are, in principle, supposed to have no relevance in the workplace. This is true in capitalism too, at least in principle: a worker’s ethnic identity, social networks, kin ties and the rest should not have any effect on hiring or how one is treated in the office or shop floor, though of course in reality this isn’t true.

4. Most critically, the financial transaction in both cases produces abstract labor, which is pure creative potential. This is created by the effects of command. Abstract labor is the sheer power of creation, to do anything at all. Everyone might be said to control abstract labor in their own person, but in order to extend it further, one has to place others in a position where they will be effectively an extension of one’s will, completely at one’s orders. Slavery, military service and various forms of corvée are the main forms in which this has manifested itself historically. Obviously, this too is something of an unrealized ideal: this is in fact precisely the area of most labor struggle. But it’s worthy of note that feudalism (or manorialism if you prefer) tends towards exactly the opposite principle: the duties owed by liege to lord were very specific and intricately mapped out.

5. A constant ideological accompaniment of this sort of arrangement is an ideology of freedom. As Moses Finley first pointed out (1980), most societies take it for granted that no human is completely free or completely dependent, rather, all have different degrees of rights and obligations. The modern ideal of political liberty, in fact, has historically tended to emerge from societies with extreme forms of chattel slavery (Pericles’ Athens, Jefferson’s Virgina), essentially as a point of contrast. Medieval jurists, for example, assumed every right was someone else’s obligation and vice versa; the modern doctrine of liberty as a property of humans one could possess was developed precisely in Lisbon and Antwerp, the cities that were at the center of the slave trade at the time; and the most common objection to this new notion of liberty at the time was that if one owns one’s freedom, it should then also be possible to sell it (Tuck, 1979). Hence the doctrine of personal liberty – outside the workplace – or even the notion of freedom of contract, that one so often encounters in societies dominated by wage labor, does not really mean we are dealing with a fundamentally different sort of system. It means we are dealing with a transformation. We are dealing with the same terms, differently arranged, so that rather than one class of people being able to imagine themselves as absolutely ‘free’ because others are absolutely unfree, we have the same individuals moving back and forth between these two positions over the course of the week and working day.

So, in effect, a transfer effected just once, by sale, under a regime of slavery is transformed into one that is repeated over and over again under capitalism.

Now, it might seem a bit impertinent to compare the morning commute to the Middle Passage, but structurally they do seem to play exactly the same role. What is accomplished once, and violently and catastrophically, in one variant, is repeated with endless mind-numbing drudgery in the other.

I should emphasize that when I say one mode of production is a transformation of the other, I am talking about the permutation of logical terms. It doesn’t necessarily imply that one grew out of the other, or even that there was any historical connection at all. I am not necessarily taking issue, for example, with the historical argument that capitalism first emerged within the English agricultural sector in the 16^th and 17^th centuries, rather than from long-distance trade (Brenner, 1976, 1979; Dobb, 1947; Wood, 2002). Or perhaps I should be more specific. It seems to me that the ‘Brenner hypothesis’, as it’s called, can account for the first two of the three features that define industrial capitalism as a mode of production: it demonstrates that the emergence of wage labor in the agricultural sector developed hand in hand with structural forces that demanded ever-expanding profits. However, it doesn’t explain the third: the emerging rural proletariats were, in legal principle and usually in practice, servants resident in their employers’ households (see e.g. Kussmaul, 1981). At the same time, this same age of ‘merchant capitalism’ did see a sudden and spectacular revival of the institution of chattel slavery and other forms of forced labor, which had largely vanished in Europe during the late Middle Ages – even though these were legally confined to the colonies. As C.L.R. James argued long ago, rationalized industrial techniques were largely developed on slave plantations, and much of the wealth that funded the Industrial Revolution emerged from the slave trade and even more from industries with servile work forces (Blaut, 1993: 203–5; James, 1938; Williams, 1944). This makes sense. Wage labor relations might have emerged among ‘improving’ landlords during that first period, but the wealthy traders of the time were after ‘abstract labor’ in the easiest form possible; their first impulse was to use slaves. Full, industrial, capitalism might then be said to have emerged only when the two fuzed. One might speculate that one reason large-scale merchants eventually came to apply wage labor at home, even within the industrial sector, was not because slavery or other forms of forced labor proved inefficient as a form of production, but rather because it did not create efficient markets for consumption: one cannot sell much of anything to slaves, and at least at that time it was difficult to keep one’s population of producers and consumers on entirely different continents.

None of this, perhaps, explains the exact connection between wage labor, separation of household and workplace, or the capitalist’s need for unlimited growth. But the theoretical terms I’ve been developing might suggest some directions. The main difference between European firms of this period and commercial enterprises in the Islamic world or East Asia seems to have been that they were not family firms. Especially with the development of the corporate form – the idea that capitalist enterprises were immortal persons free of the need to be born, marry or die – the economic domain was effectively excised from the domain of transformation and the mutual shaping of human beings and came to be seen as something transcendent. This might suggest:

Thesis 5: Capitalism’s unlimited demand for growth and profit is related to the transcendent abstraction of the corporate form. In any society, the dominant forms are considered transcendent from reality in much the way value forms tend to be and when these transcendent forms encounter ‘material’ reality, their demands are absolute

This one, though, I will have to leave as a possible direction for future research.

References