1 Assemblages and Human History

We no longer believe in a primordial totality that once existed, or in a final totality that awaits us at some future date. We no longer believe in the dull gray outlines of a dreary, colorless dialectic of evolution, aimed at forming a harmonious whole out of heterogeneous bits by rounding off their rough edges. We believe only in totalities that are peripheral. And if we discover such a totality alongside various separate parts, it is a whole of these particular parts but does not totalize them; it is a unity of all those particular parts but does not unify them; rather it is added to them as a new part fabricated separately.

Deleuze and Guattari, Anti-Oedipus1

A very important question confronting any serious attempt to think about human history is the nature of the social entities considered to be legitimate agents. One can, of course, include only human beings as agents, either as rational decision-makers (as in micro-economics) or as phenomenological subjects (as in micro-sociology). But if we wish to go beyond this we need a proper concept of a social whole. The very first step in this task is to devise a means to block micro-reductionism, a step usually achieved by the concept of emergent properties, the properties of a whole caused by the interactions between its parts. If a social whole has novel properties that emerge from interactions between people, its reduction to a mere aggregate of many rational decision-makers or many phenomenological experiences is effectively blocked. But this leaves open the possibility of macro-reductionism, as when one rejects the rational actors of micro-economics in favour of society as a whole, a society that fully determines the nature of its members. Blocking macro-reductionism demands a second concept, the concept of relations of exteriority between parts. Unlike wholes in which ‘being part of this whole’ is a defining characteristic of the parts, that is, wholes in which the parts cannot subsist independently of the relations they have with each other (relations of interiority), we need to conceive of emergent wholes in which the parts retain their autonomy, so that they can be detached from one whole and plugged into another one, entering into new interactions. As mentioned in the introduction, the terms ‘exteriority’ and ‘interiority’ are misleading, since they suggest a spatial relation. We may accept the traditional terminology as long as we keep in mind that we are not invoking anything spatial: the relations among parts of the brain, for example, are relations of exteriority despite the fact that these parts are in the interior of the body.

With these two concepts – emergence and exteriority – we can define social wholes, like interpersonal networks or institutional organisations, that cannot be reduced to the persons that compose them, but that do not totalise them either, fusing them into a seamless whole in which their individuality is lost. Take for example the tightly knit communities that inhabit small towns, or neighbourhoods with an ethnic composition in large cities. In these communities an important emergent property is the degree to which their members are linked together. One way of examining this property is to study the network of neighbours, counting the number of direct and indirect links per neighbour, that is, friends and friends of friends, and studying the network’s connectivity. Among the properties that can be ascribed to the network as a whole is the density of its connections, a property that may be informally defined as the degree to which the indirect links of any given member (the friends of his or her friends) know the indirect links of others. Or to put it still more simply, by the degree to which everyone knows everyone else. In such a dense network word of mouth travels fast, particularly when the content of the gossip is the violation of a local norm: an unreciprocated favour, an unpaid bet, an unfulfilled promise, a lie. To the extent that this information is remembered by enough neighbours we can say that the community stores personal reputations. If to this collective memory we add the ability to perform simple punishments, using ridicule or ostracism, the community as a whole can enforce local norms. True, in any one case the neighbours laughing behind my back or refusing to interact with me are concrete persons, but to the extent that the details of their identity do not matter – this group of neighbours is ridiculing me now but the humiliation would be the same if it was a different group of neighbours – it is the community as a whole that performs the punitive functions.

The property of density, and the capacity to store reputations and enforce norms, are non-reducible properties and capacities of the entire community, but neither involves thinking of it as a seamless totality in which the very personal identity of the members is created by their relations: neighbours can pack their things and move to a different community while keeping their identity intact. And a similar point applies to institutional organisations, like bureaucratic agencies, hospitals, universities, or factories. Many organisations are characterised by the possession of an authority structure in which rights and obligations are distributed in a hierarchical way. But the exercise of authority must be backed by legitimacy if enforcement costs are to be kept within bounds. Legitimacy is a property of an organisation’s authority structure even if it depends for its existence on personal beliefs about its source: a legitimising tradition; a set of written regulations and an institutional mission; or even, for small organisations, the charisma of a leader. The degree to which legitimate authority is irreducible to persons can, of course, vary from case to case. In particular, the more an organisation’s resources are linked to an office or role (as opposed to the incumbent of that role) the more irreducible its legitimacy is. Nevertheless, and however centralised and despotic an organisation may be, its members remain ultimately separable from it, their actual degree of autonomy depending on contingent factors about social mobility and the existence of opportunities outside the organisation. It is this type of irreducible social whole produced by relations of exteriority, a whole that does not totalise its parts, that the opening quotation refers to. We can refer to these social wholes as ‘assemblages’.

Our opening quotation, on the other hand, does not employ the concept of emergence (or any of its synonyms, like synergy). There are, however, two compelling reasons to include emergence as part of the definition of the term ‘assemblage’. First of all, without something ensuring the irreducibility of an assemblage, the concept would not be able to replace that of a seamless totality. If the parts of a whole are reducible, then they form an aggregate in which the components merely coexist without generating a new entity. Hence, irreducibility is implicit in the concept of assemblage. Second, making the properties of a whole depend on the interactions between its parts ensures that these properties are not taken to be either necessary or transcendent. When the properties of a given whole are taken as a brute fact, and listed as the unexplained characteristics that the whole must possess in order to be an entity of a given kind, the list of necessary properties swiftly becomes an essence. Essences belong to a different plane of being from the entities whose identity they define, a transcendent plane overflying that which the entities populate. But if the properties are viewed as produced by the interactions between components, and their existence and endurance explained by the continuity of those interactions, then the properties are contingent: if the interactions cease to take place the emergent properties cease to exist. To return to the previous examples, if all the neighbours stop interacting with one another by having conversations (so that gossip about local norm violation can be spread), or if they all stop reciprocating favours or keeping promises, the capacity of the community to store reputations and punish violations ceases to exist. Similarly, if the members of an institutional organisation stop obeying orders and filing reports, the legitimacy of the authority structure is compromised. The right interactions between neighbours or employees must take place on a day-to-day basis for the social assemblages to have the properties and dispositions that they do.

The contributions of the concept of emergence to a theory of assemblages can help us understand the other important characteristic mentioned in the opening quotation: that social wholes must be considered to be peripheral or to exist alongside their parts. This is clearly not a reference to relations in space, as if communities or organisations existed nearby or to one side of the persons that compose them. The reference is not spatial but ontological: the whole exists alongside the parts in the same ontological plane. In other words, the whole is immanent, not transcendent. Communities or organisations are historically individuated entities, as much so as the persons that compose them. While it is true that the term ‘individual’ has come to refer to persons (or organisms in the case of animals and plants), it is not incoherent to speak of individual communities, individual organisations, individual cities, or individual countries. The term ‘individual’ has no preferential affinity for a particular scale (persons or organisms) but refers to any entity that is historically unique. Since all assemblages have this ontological status they all populate the same ontological plane, and this distinguishes assemblage theory from other realist philosophies in which there are strong ontological distinctions between levels of existence, such as genus, species, and organism. In assemblage theory we must insist that the human species be treated as an individual entity, with a date of birth (the event of speciation) and, at least potentially, a date of death (the event of extinction). In other words, the human species as a whole exists ‘alongside’ the human organisms that compose it, alongside them in an ontological plane that is populated exclusively by historically individuated entities.

Having discarded seamless totalities and transcendent essences, we can now return to the question of human history. Historical explanations are inevitably shaped by the ontological commitments of the historians who frame them. These commitments may be roughly divided into two classes along the lines suggested in the opening paragraph, that is, depending on which of the terms of these binary oppositions are favoured: ‘the individual versus society’, ‘agency versus structure’, ‘choice versus order’. Taking the side of the first terms in these dichotomies yields narratives in which the actions or ideas of persons, typically ‘great men’, are the main factors shaping events, situations, or the outcomes of particular struggles. This does not necessarily imply a disbelief in the existence of society as a whole, only a conception of it that makes it into an epiphenomenon: society is a sum or aggregate of many rational agents or many phenomenological experiences shaped by daily routine. Taking the side of the second terms, on the other hand, yields narratives framed in terms of the transformations that enduring social structures have undergone. The best-known example of this view of history is the sequence feudalism–capitalism–socialism. As before, there is no implication here that persons are not real, only that they are mere epiphenomena: persons are socialised as they grow up in families and attend schools, and after they have internalised the values of their societies their obedience to traditional regulations and cultural values can be taken for granted.

The late economic historian Fernand Braudel broke with both of these traditional stances when he set out to study the history of Western economies using the concept of society as a set of sets.2 The characters in his narratives include such diverse entities as communities, organisations, cities, and the urban regions formed by several interacting towns of different sizes. Persons are featured too but not as great men, while larger entities, like kingdoms, empires, world economies, are treated not as abstract social structures but as concrete historical individuals, referred to by a proper name. Speaking of a ‘sets of sets’ is another way of saying that the variety of forms of historical agency (communal agency, organisational agency, urban agency, imperial agency) are related to one another as parts to wholes. Braudel’s is a multi-scaled social reality in which each level of scale has its own relative autonomy and, hence, its own history. Historical narratives cease to be constituted by a single temporal flow – the short timescale at which personal agency operates or the longer timescales at which social structure changes – and becomes a multiplicity of flows, each with its own variable rates of change, its own accelerations and decelerations.

Braudel’s vision can be enriched by replacing his sets, or sets of sets, with the irreducible and decomposable wholes just discussed, that is, with assemblages (and assemblages of assemblages). Let’s illustrate this with a specific example, one that combines Braudel’s data with an ontology of individual entities constraining the field of valid historical actors. In this ontology, an entity like ‘the Market’ would not be an acceptable entity to be incorporated into explanations of historical phenomena because it is not an individual emergent whole but a reified generality. But the marketplaces or bazaars that have existed in every urban centre since antiquity, and more recently in every European town since the eleventh century, are indeed individual entities and can therefore figure as actors in explanations of the rise of Europe, and of the commercial revolution that characterised the early centuries of the second millennium. Equally valid are the regional trading areas that emerged when the towns that housed local marketplaces became linked together by roads and the trade among them reached a threshold of regularity and volume. Regional markets began to play an important economic role in Europe by the fourteenth century and, as historically constituted wholes composed of local marketplaces, they are valid historical actors. So are the national markets that, starting in England in the eighteenth century, came into being by stitching together, sometimes forcefully, many provincial trading areas themselves composed of many regional markets. By the nineteenth century the railway and the telegraph made the creation of national markets a simpler task and they emerged in places like France, Germany, and the United States, playing an important role in the economic history of those countries.3 Today we are witnessing the assembly of continental markets, the European Union being a prime example, but this is still an unfinished historical task, one that could fail if the interacting national markets cease to give rise to an emergent whole.

Up to the level of national markets the main emergent property of these increasingly larger trading areas is synchronised price movements. Braudel uncovers evidence that average wholesale prices (determined mostly by demand and supply) move up and down in unison within urban regions, provinces, or entire countries. But in addition to these commercial ‘sets of sets’, Braudel also examines the very different dynamic created by international trade, a dynamic that has existed in the West since the fourteenth century. The reason why international trade is treated differently is that it tends to link trading areas in which demand and supply are entirely disconnected, creating price differentials that can be used to generate very large profits. Today’s global trade includes many areas in which wage differentials generate the same opportunities. But to the extent that these profits arise by manipulating demand and supply – as opposed to by making decentralised decisions based on the information about demand and supply carried by prices – international trade (and the maritime cities and organisations behind it) constitutes a different economic entity to local, regional, provincial, and national markets. Thus, instead of postulating a single macro-entity (the Market, the Capitalist System), Braudel focuses on dynamics operating at different scales, and is able to capture the heterogeneity of practices and variety of social entities that constitute the real agents of economic history.

In a similar way, other reified generalities, like ‘the State’, should also be replaced. As argued above, in addition to communities, a set of interacting persons can give rise to institutional organisations possessing emergent properties like legitimate authority. Organisations, in turn, can interact to form a larger emergent whole like the industrial network formed by General Motors and its hundreds of suppliers and distributors (car dealerships). And similarly for governments. A typical federal government, for example, is a whole in which many organisations are arranged in a hierarchical way with authority operating at different scales: some have a jurisdiction that extends to the entire country; others have authority only within the boundaries of a province or state; and yet others operate within the limits of an urban centre and its surrounding region. When it comes to the implementation of federal policies, this nested set of overlapping jurisdictions can be a powerful obstacle, many policies becoming distorted and weakened as they move from a decision made into law at the federal level to a set of practices and procedures by different bureaucratic agencies, each possessing its own agenda and internal culture. Since these organisations themselves exist at different scales (national, provincial, urban), the problem of jurisdictional disputes and local veto powers can make the faithful implementation of federal policy highly problematic. This can help explain why many governments are so dysfunctional, an explanation not available to historians who use the concept of ‘the State’ and the implied belief that governments are monolithic entities.

Thus, both ‘the Market’ and ‘the State’ can be eliminated from a realist ontology by a nested set of individual emergent wholes operating at different scales. The expression ‘operating at different scale’, on the other hand, must be used carefully. In particular, it should refer only to relative scale, that is, to scale relative to the part-to-whole relation. Given the fact that any emergent whole always has a larger extension than the parts of which it is composed, this relative usage is unproblematic: communities or organisations are always larger than the persons that compose them. But the same is not true if the term ‘scale’ is used in an absolute sense. If, instead of comparing a community with its own members, we compared the entire set of persons and the entire set of communities inhabiting a given country, we would have to admit that both sets are coextensive, that is, that they occupy the same amount of space: the entire national territory. And a similar point applies to the set of institutional organisations. But even if we relativise the concept we may still disagree on the use of the expression ‘levels of scale’ to distinguish social wholes. Why not use, for example, the expression ‘levels of organisation’, a phrase used by biologists to characterise the part-to-whole relations between individual cells, individual organs, and individual organisms? Because this concept carries with it connotations of increased complexity between levels, and in some cases even teleological implications, as when biological evolution is viewed as involving a drive to greater complexity, from unicellular organisms to multicellular ones. The expression ‘levels of scale’, on the other hand, carries no such connotations: a city is clearly larger than a human being but there is no reason to believe that it possesses a higher degree of complexity than, say, the human brain.

One final point needs to be clarified: when we say that a set of interacting persons gives rise to a community, or that a set of interacting organisations gives rise to a federal government, this should not be taken to imply a temporal sequence, as if a set of previously disconnected persons or organisations had suddenly begun to interact and a whole had abruptly sprouted into being. In a few cases this may indeed be the case, as when people from a variety of war-stricken communities aggregate into a refugee camp and a larger whole emerges from their interactions; or when previously rival industrial organisations aggregate into a cartel forming a larger whole. But in the majority of cases the component parts come into being when a whole has already constituted itself and has begun to use its own emergent capacities to constrain and enable its parts: most people are born into communities that predate their birth, and most new government agencies are born in the context of an already functioning central government. Nevertheless, the ontological requirement of immanence forces us to conceive of the identity of a community or of a central government as being continuously produced by the day-to-day interactions between its parts. So we need to include in a realist ontology not only the processes that produce the identity of a given social whole when it is born, but also the processes that maintain its identity through time. And we must also include the downward causal influence that wholes, once constituted, can exert on their parts.

Let’s pause for a moment to consider how compatible these ideas are with those of Deleuze and Guattari. The first and most obvious sign of incompatibility is that the expression ‘the State’ occurs throughout their work. But this term is often used synonymously with ‘State apparatus’, a term that is less objectionable since it can be taken to refer to the organisational apparatus available to a given government. A more problematic term, one that is also often used in their historical explanations, is ‘social field’ (or less often, ‘the socius’). This term does indeed refer to ‘society as a whole’ and it is therefore not a valid historical actor in the realist ontology being sketched here. It is unclear, for example, just what kind of entity this ‘social field’ is supposed to be. As mentioned in the introduction, Deleuze and Guattari distinguish between different kinds of social wholes: strata and assemblages. The State is classified by them as a stratum.4 A tightly knit community, with its capacity to police its members and punish violations of local norms, would also be a stratum. But an alliance of several communities, such as those involved in a social justice movement, would be considered an assemblage. So we face the problem of whether to treat the ‘social field’ as a stratum or as an assemblage. A different problem is that distinguishing between different kinds of wholes involves ontological commitments that go beyond individual entities. In particular, with the exception of conventionally defined types (like the types of pieces in a chess game), natural kinds are equivalent to essences. As we have already suggested, avoiding this danger involves using a single term, ‘assemblage’, but building into it parameters that can have different settings at different times: for some settings the social whole will be a stratum, for other settings an assemblage (in the original sense).

The term ‘parameter’ comes from scientific models of physical processes. In these models, variables specify the different ways in which the phenomenon being studied can change, while parameters specify the environmental factors that affect the phenomenon. A property like temperature, for example, can be a variable, the internal temperature of a body of water, but it can also be a parameter quantifying the temperature of the water’s surroundings. Parameters are normally kept constant in a laboratory to study an object under repeatable circumstances, but they can also be allowed to vary, causing drastic changes in the phenomenon under study: while for many values of a parameter like temperature only a quantitative change will be produced, at critical points a body of water will spontaneously change qualitatively, abruptly transforming from a liquid to a solid, or from a liquid to a gas. By analogy, we can add parameters to concepts. Adding these control knobs to the concept of assemblage would allow us to eliminate their opposition to strata, with the result that strata and assemblages (in the original sense) would become phases, like the solid and fluid phases of matter. Unlike mutually exclusive binary categories, phases can be transformed into one another, and even coexist as mixtures, like a gel that is a mixture of the solid and liquid phases of different materials. In the case of social wholes, an important parameter, or variable coefficient, is its degree of ‘territorialisation’ and ‘deterritorialisation’. Thus, Deleuze writes: ‘What must be compared in each case are the movements of deterritorialisation and the processes of reterritorialisation which appear in an assemblage. But what do they mean, these words that Félix invents to make them into variable coefficients?’5

Before answering this important question let’s summarise what has been said about assemblages so far:

Let’s move on to characterise the different parameters that must be built into the concept of assemblage, starting with the parameter that quantifies the degree of territorialisation and deterritorialisation of an assemblage. Territorialisation refers not only to the determination of the spatial boundaries of a whole – as in the territory of a community, city, or nation-state – but also to the degree to which an assemblage’s component parts are drawn from a homogeneous repertoire, or the degree to which an assemblage homogenises its own components. When a community is densely connected, we can expect a reduction of personal differences and an increased degree of conformity. But in normal circumstances, this mild degree of territorialisation may be compatible with the acceptance of personal differences. However, when two or more communities engage in ethnic or religious conflict, not only will the geographical boundaries of their neighbourhoods or small towns be policed more intensely, so will the behaviour of their members. The distinction between ‘us’ and ‘them’ will sharpen and any small deviation from local norms will be noticed and punished: are you really a Christian, a Muslim, a Jew? Conflict, in other words, tends to increase the degree of territorialisation of communities, a fact that may be captured conceptually by a change in the setting of the parameter.

The second parameter quantifies an assemblage’s degree of coding and decoding. Coding refers to the role played by special expressive components in an assemblage in fixing the identity of a whole. The two best-known expressive components with a specialised function are chromosomes and languages. Since we are considering here only social wholes, let’s focus on the latter. In institutional organisations, for example, the legitimacy of an authority structure is in most cases related to linguistically coded rituals and regulations: in organisations in which authority is based on tradition, these will tend to be legitimising narratives as part of a sacred text, while in those governed by a rational-legal form of authority they will be written rules, standard procedures, and most importantly, a constitution defining the organisation’s rights and obligations. While all organisations are coded in this way, a state apparatus performs coding operations that affect an entire territory and all the communities and organisations that inhabit it. The more despotic a state apparatus, the more everything becomes coded: dress, food, manners, property, trade. Because many archaic states allowed the communities over which they ruled to keep their own social codes, superimposing on them a dominant code, Deleuze and Guattari refer to this operation as ‘overcoding’.6

What the authors refer to as an ‘assemblage’ is a stratum that has become decoded, that is, one in which the value of the coding parameter is low, as when animal behaviour stops being determined by genes, or when human behaviour ceases to be fully specified by written norms. As they write: ‘Assemblages are already different from strata. They are produced in the strata, but operate in zones where milieus become decoded.’7 Let’s examine in more detail the case of assemblages with high values for both parameters. First of all, as in the case of mathematical parameters, we must define what the parameters stand for. The latter may stand for actual knobs in a piece of laboratory equipment (knobs which may be tuned to change an artificial phenomenon’s environment) or to factors affecting a given natural phenomenon’s environment. So what do the knobs built into the concept of assemblage stand for? They refer to the objective articulatory processes that yield a molar whole from a population of molecular parts. The identity of an objective stratum is therefore determined by a process that Deleuze and Guattari refer to as a ‘double articulation’. As they write:

Each stratum exhibits phenomena constitutive of double articulation. Articulate twice, B-A, BA . . . The first articulation chooses or deducts, from unstable particle-flows, metastable molecular or quasi-molecular units (substances) upon which it imposes a statistical order of connections and successions (forms). The second articulation establishes functional, compact, stable substances (forms), and constructs the molar compounds in which these structures are simultaneously actualized (substances). In a geological stratum, for example, the first articulation is the process of ‘sedimentation’ which deposits units of cyclic sediment according to a statistical order: flysch, with its succession of sandstone and schist. The second articulation is the ‘folding’ that sets up a stable functional structure and effects the passage from sediment to sedimentary rock.8

In this example, the molecular population is made out of small pebbles and the molar whole is the larger sedimentary rock that these pebbles eventually form. We must, however, correct a mistake in this description. The process that creates sedimentary rock proceeds by the sorting out of pebbles of different size and composition, an operation performed by the rivers that transport and deposit the pebbles in homogeneous layers at the bottom of the ocean. Then, these loose accumulations are cemented together and transformed into layers of sedimentary rock. These rocks, in turn, may accumulate on top of one another and then be folded by the clash of tectonic plates to produce a larger emergent entity: a folded mountain range like the Himalayas or the Rocky Mountains. We will see below that this is not the only place where Deleuze and Guattari fail to make a distinction between different scales. But the ease with which the mistake can be corrected shows that the concept of a double articulation is robust against simple errors and, more importantly, capable of multiple variations that accommodate the complexity of actual strata. That is, the first articulation does not have to involve sorting into internally homogeneous sets, although sorting operations do indeed appear in different spheres: the selection pressures exerted by predators and parasites sort out genetic materials and increase their homogeneity, and the use of technical examinations in the recruitment of staff for organisations sorts people out into uniform sets. But the homogenising effects of the first articulation may take place through a variety of mechanisms. A similar situation applies for the second articulation, a consolidation operation that can take a variety of forms, such as the reproductive isolation of an animal community that prevents extraneous genes from other species from flowing into their gene pool.

When we first introduced the parameters standing for these two articulations, the second one, coding, was defined in terms of the effect that a specialised expressive component (genes, words) can have on the identity of a whole. But the geological example just given shows that coding can be present even before the advent of the genetic code. Nevertheless, the appearance of genetic materials capable of information storage and self-replication was indeed a crucial event, an event that changed the very nature of the process of double articulation. While before the rise of living creatures all expression was three-dimensional – the geometry of a crystal, for example, was what expressed its identity – genes are a one-dimensional form of expression, a linear chain of nucleotides, and this linearisation is what allowed material expressivity to gain a degree of autonomy from its material base and become highly specialised. As Deleuze and Guattari put it:

Before, the coding of a stratum was coextensive with that stratum; on the organic stratum, on the other hand, it takes place on an autonomous and independent line that detaches as much as possible from the second and third dimensions . . . The essential thing is the linearity of the nucleic sequence . . . It is the crystal’s subjugation to three-dimensionality, in other words, its index of territoriality, that makes the structure incapable of formally reproducing and expressing itself; only the accessible surface can reproduce itself, since it is the only deterritorializable part. On the contrary, the detachment of a pure line of expression on the organic stratum makes it possible for the organism to attain a much higher threshold of deterritorialization, gives it a mechanism of reproduction covering all the details of its complex spatial structure, and enables it to put all its interior layers topologically in contact with the exterior, or rather with the polarized limit (hence the special role of the living membrane).9

Language emerges in a similar way except that its linearity is now temporal not spatial, involving an even more intense deterritorialisation that makes it independent of its formed materiality. This is what gives language the ability to represent all other strata, to translate ‘all of the flows, particles, codes, and territorialities of the other strata into a sufficiently deterritorialized system of signs . . .’10 And this capacity to represent or translate all other strata is, in turn, what gives language, or more exactly language-based theories, their ‘imperialist pretensions’. In other words, the obsession with language that took place in the twentieth century after the so-called ‘linguistic turn’, forming the basis for the rejection of materialism and the spread of conservative idealism, can be explained within the theory of double articulation as a result of the unique status of this specialised line of expression. Thus explained, the power of language can be accepted while the conceptual obstacle represented by its illegitimate extension is circumvented. In what follows I will use my own version of the concept of assemblage, in which strata are just the phase that results when the two parameters have high values, but these remarks should make it easier to move back and forth to the original version in which strata and assemblages are different kinds of wholes.

Armed with this parametrised concept, we can now move back to the question of human history and a more detailed treatment of the different levels of scale at which historical events take place. Since we reject macro-reductionism as much as micro-reductionism, we must take individual persons seriously, as long as the subjectivity of each person is itself conceived as emerging from the interactions between sub-personal components. From the philosopher David Hume, Deleuze derives a conception of the subject or person as an entity emerging from the interactions of a heterogeneous population of sense impressions, and of the low-intensity replicas of those impressions (ideas). These sub-personal components are assembled through the habitual application of certain operators to the ideas. More specifically, a subject crystallises in the mind through the habitual grouping of ideas via relations of contiguity; their habitual comparison through relations of resemblance; and the habitual perception of constant conjunction of cause and effect that allows one idea (that of the cause) to always evoke another (the effect). Contiguity, causality, and resemblance, as relations of exteriority between ideas, constitute the three kinds of association that transform a mind into a subject.11

Deleuze never gave a full assemblage analysis of subjectivity, but it is possible to derive one from his work on Hume. His most extensive discussion of relations of exteriority occurs, in effect, in his discussion of the empiricists. Deleuze warns us that the history of philosophy gives us an impoverished idea of what the empiricists achieved: the intelligible comes from the sensible. Or in more contemporary terms, theoretical statements are reducible to observation statements, a programme for epistemology that everyone agrees has failed. So what then is their achievement? To have established external relations between the sensual and intellectual components of experience. As he writes,

in effect if relations are external and irreducible to their terms, then the difference cannot be between the sensible and the intelligible, between experience and thought, between sensation and ideas, but only between two sorts of ideas, or two sorts of experiences, that of terms and that of relations . . . In Hume there are ideas and then there are relations between these ideas, relations which may vary without the ideas varying, and then the circumstances, actions, and passions which make these relations vary. A complete ‘Hume assemblage’ which takes on the most varied figures.12

Simplifying somewhat, the sub-personal expressive components of the assemblage would comprise both those that are non-linguistic (sense impressions of varying vividness) and those dependent on language, such as beliefs considered as attitudes towards the meaning of declarative sentences (propositions). Material components would include the routine mental labour performed to assemble ideas into a whole, as well as the biological machinery of sensory organs needed for the production of sensations. Habit itself constitutes the main form of territorialisation, that is, the process that gives a subject its defining boundaries and maintains those boundaries through time. In the model used by Deleuze, habit performs a synthesis of the present and the past in view of a possible future.13 This yields a determinate duration for the lived present of the subject, a fusion of immediately past and present moments, and generates a sense of anticipation, so that habitual repetition of an action can be counted on to yield similar results in the future. A process of deterritorialisation, on the other hand, would be any process that takes the subject back to the state it had prior to the creation of fixed associations between ideas, that is, the state in which ideas and sensations are connected as in a delirium. The onset of madness, high fever, intoxication, sensory deprivation, psychedelic drugs, and a variety of other processes can cause a loss or destabilisation of subjective identity.

Personal identity, on the other hand, has not only a private aspect but also a public one, the public persona that we present to others when interacting with them in a variety of social encounters. Some of these social encounters, like ordinary conversations, are sufficiently ritualised that they themselves may be treated as assemblages. The author who has done the most valuable research on interaction rituals is without doubt the sociologist Erving Goffman, who defines the subject matter of this research as

the class of events which occurs during co-presence and by virtue of co-presence. The ultimate behavioural material are the glances, gestures, positionings, and verbal statements that people continuously feed into the situation, whether intended or not. These are the external signs of orientation and involvement – states of mind and body not ordinarily examined with respect to their social organisation.14

While the most obvious expressive component of this assemblage may be the flow of words itself, there is another one which is not always dependent on language. Every participant in a conversation is expressing his or her persona through facial gesture, posture, dress, choice of subject matter, the deployment of (or failure to deploy) poise and tact, and so on. These components express in a non-linguistic way the image that every participant wants to project to others. The expression of these claims must be done carefully: one must choose an image that cannot be easily discredited by others, but that masks our weaknesses and highlights our strengths. Any conversation will therefore be filled with objective opportunities to express favourable information about oneself, as well as objective risks of unwittingly expressing unfavourable facts. The material components of the assemblage are more straightforward, consisting both of the physical bodies assembled in space, close enough to hear each other and correctly oriented towards one another, as well as the expenditure of attention needed to keep the conversation going and the labour involved in repairing breaches of etiquette or recovering from embarrassing events.15 Some technological inventions, such as the telephone, can change the requirement of co-presence, eliminating some of the material components (spatial proximity) but adding others: the technological device itself, as well as the infrastructure needed to link many such devices. (A similar point applies to the Internet.)

Territorialisation, in this case, is performed by social conventions giving a conversation well-defined borders in space and time. Conversations have a temporal structure defined by rules like turn-taking, as well as by ritual means of initiating and terminating an encounter. The spatial boundaries of these assemblages are clearly defined by the physical requirement of co-presence and by the fact that once the participants ratify each other as legitimate interactors, they can exclude passers-by from intruding into the conversation.16 Any event that takes attention away from the subject of the conversation to focus it elsewhere can destabilise this assemblage. An important example of these are events that threaten in one way or another the public personas of the participants, such as embarrassing revelations or humiliating comments. Goffman discusses critical points of embarrassment after which regaining composure becomes impossible, embarrassment is transmitted to all participants, and the conversation falls apart.17 But other critical events may take place that transform a conversation into a heated discussion, or an intense argument into a fist fight. These should also be considered factors that switch the control parameter into a deterritorialising setting. Other factors promoting deterritorialisation are technological inventions that allow conversations to take place at a distance, blurring their spatial boundaries.

Conversations among neighbours are among the day-to-day interactions that generate the tightly knit communities discussed above. But conversations are also important in maintaining the binding ties that constitute a dispersed network of friends or professional colleagues. Either way, we are dealing with assemblages that can be analysed using the resources offered by social network theory. In the models used in this theory, a node stands for a person, while a line joining two or more nodes stands for a relation or regular interaction among persons. When we examine the emergent properties of a network, the lines or links are often more important than the nodes themselves, a fact that orients social network theory towards relations of exteriority. The links in a network may be characterised in a variety of ways: by their presence or absence, the absences indicating the borders separating one network from another, or defining a clique within a given network; by their strength, determined both by the frequency of interaction among the persons occupying the nodes, as well as by the emotional content of the relation; and finally, by their reciprocity, that is, by the mutuality of obligations entailed by the link. As argued above, one of the most important properties of an interpersonal network is its density, a measure of the degree of connectivity among its indirect links.18 The links in a network must be constantly maintained in good shape, the maintenance labour involved constituting one of the material components. This labour goes beyond merely engaging in frequent conversations. Listening to problems and giving advice in difficult situations, and providing physical help or taking care of other people’s children, are also part of the maintenance process.19 The expressive components of the assemblage include a variety of expressions of solidarity and trust emerging from, and then shaping, interactions. These range from routine acts like having dinner together or going to church, to the sharing of adversity and the displayed willingness to make sacrifices for the community as a whole.20

As in the case of conversations, the value of the territorialisation parameter is closely related to physical proximity. Communities structured by dense networks have historically tended to inhabit the same small town, or the same suburb or ethnic neighbourhood in a large city. These bounded geographical areas are literally a community’s territory and they may be marked and distinguished from others by special expressive signs, from architectural motifs to the way people congregate on the street. Deterritorialising processes include any factor that decreases density, promotes geographical dispersion, or eliminates some rituals, like churchgoing. Social mobility and secularisation are among these processes. The former weakens links by making people less interdependent, by increasing geographical mobility, and by promoting a greater acceptance of difference through less local and more cosmopolitan attitudes. The resulting deterritorialised networks require their members to be more active in the maintenance of links and to invent new forms of communal participation, given that connections will tend to be wider and weaker and that ready-made rituals for the expression of solidarity may not be available.21 The same kind of resourcefulness in the means to maintain linkages may be needed in interpersonal networks deterritorialised by technology. For example, in the early ‘virtual communities’ that emerged on the Internet (such as the Well), members were aware of the loss that a lack of co-presence involved and special meetings or parties were regularly scheduled to compensate for this.22

While in an interpersonal network a particular node may become dominant by being more highly connected to other nodes, the resulting popularity rarely gives the incumbent the capacity to issue commands. This capacity implies the existence of an authority structure, and this, in turn, means that we are dealing with a different assemblage: an institutional organisation. Organisations come in a wide range of scales, with nuclear families at the low end and government bureaucracies and commercial, industrial, or financial corporations at the other. A modern hierarchical organisation may be studied as an assemblage given that the relations between its components are extrinsic: what holds the whole together are relatively impermanent contractual relations through which some persons transfer rights of control over a subset of their actions to other persons. This voluntary submission breaks the symmetry of the relations among persons in an interpersonal network where a high degree of reciprocity is common.23

There are a variety of forms of authority. In small organisations, like religious sects, the charisma of a leader may be enough to legitimise commands, but as soon as the number of members increases past a certain threshold, formal authority becomes necessary, justified by a tradition, as in organised religion, or by actual problem-solving performance, as in the case of bureaucracies.24 In all organisations the automatic obedience to commands on a day-to-day basis constitutes a powerful expression of legitimacy. For the same reason any act of disobedience, particularly when it goes unpunished, threatens this expression and may damage the morale of those who obey. Hence, the expressive role of some forms of punishment that are designed to make an example of transgressors. Punishment, on the other hand, also has a physical aspect, and this points to the material components of the assemblage, related not so much to practices of legitimation as to practices of enforcement. In charismatic and traditional organisations these practices may involve torture, mutilation, confinement, exile. But in modern bureaucracies, as well as in many other members of the population of organisations (prisons, hospitals, factories, schools, barracks), enforcement uses subtler but perhaps more efficient means: a specific use of space, in which dangerous groupings are broken up and individual persons are assigned a relatively fixed place; systematic forms of inspection and monitoring of activity, a practice that shapes and is shaped by the analytical use of space; and finally, a constant use of logistical writing, like the careful keeping of medical or school records, to permanently store the product of monitoring practices.25

As with communities, territoriality in the case of organisations has a strong spatial aspect. Most organisations possess physical premises within which they carry on their activities and which, in some cases, define the extent of their jurisdiction: the effective authority of a doctor ends at the entrance to the hospital and that of the teacher at the entrance to the school. Organisational territories are determined both symbolically, as a legitimate jurisdictional area defined in writing, and materially, by the area over which authority can actually be enforced. But just as in interpersonal networks, processes of territorialisation go beyond the strictly spatial. The routinisation of everyday activities, in the form of the repetition of traditional rituals or the systematic performance of regulated activities, stabilises the identity of organisations and gives them a way to reproduce themselves, as when a commercial organisation opens up a new branch and sends part of its existing staff to provide the institutional memory (the day-to-day routines) of the parent company. Technological innovation, on the other hand, can destabilise this identity. Transportation and communication technologies, for example, can have deterritorialising effects on organisations similar to the effects they have on face-to-face interaction, allowing organisations to break away from the limitations of spatial location. The modern bureaucratic form of authority may have emerged in part thanks to the precision with which the dispersed activities of many branches of an organisation could be coordinated via the railways and the telegraph.26 And a similar point can be made about the transformation that large commercial or industrial corporations underwent in the nineteenth century, as they became nationwide corporations, as well as in the twentieth century when they became international.

As argued above, individual organisations may form larger social entities, like the supplier and distribution networks linked to large industrial firms, or the hierarchy of legislative, judicial, and executive organisations that compose a federal government. Let’s skip this important layer to describe entities operating at an even larger scale, such as cities or nation-states. Neither urban centres nor territorial states when considered as assemblages should be confused with the organisations that make up their government, even if the jurisdictional boundaries of the latter coincide with the geographical boundaries of the former. Cities and nation-states are partly the physical locales in which a variety of differently scaled social agents carry on their day-to-day activities. An urban centre, for example, possesses not only a physical infrastructure and a given geographical setting, but it also houses a diverse population of persons; many interpersonal networks, some dense and well localised, others dispersed and shared with other cities; a plurality of organisations of different sizes and functions, some of which make up larger entities such as industries or sectors. A city assembles the activities of these social entities in a concrete physical locale, allowing them to interact and generate novel properties and capacities.

Cities possess a variety of material and expressive components. On the material side, we must list for each neighbourhood the different buildings in which the daily activities and rituals of the residents are performed and staged (the pub and the church, the shops, the houses, and the local square) as well as the streets connecting these places. In the nineteenth century new material components were added, water and sewage pipes, conduits for the gas that powered early street lighting, and later on electricity cables and telephone wires. Some of these components simply add up to a larger whole, but citywide systems of mechanical transportation and communication can form very complex networks with properties of their own, some of which affect the material form of an urban centre and its surroundings. A good example is locomotives (and their rail networks) which possess such a large mass and are so hard to stop and accelerate that train stops must be located at intervals of two or three miles. This, in turn, can influence the spatial distribution of the small towns that grow around train stations, giving them their characteristic ‘string of beads’ shape.27 On the expressive side, a good example is a city’s skyline, that is, the silhouette cut against the sky by the mass of its buildings and the decorated tops of its churches and public buildings. For centuries these skylines were the first image visitors saw as they approached a city, a recognisable expression of a town’s identity, an effect lost later on as suburbs and industrial hinterlands blurred city boundaries. In some cases, the physical skyline of a town is simply a sum of its parts, but the rhythmic repetition of architectural motifs – minarets, domes and spires, belfries and steeples – and the counterpoint these motifs create with the surrounding landscape may produce emergent expressive effects.28 In the twentieth century skyscrapers and other signature buildings were added to the skyline as a means of making it unique and instantly recognisable, a clear sign that the expressivity of skylines had become the object of deliberate planning.

A variety of territorialising and deterritorialising processes may change the state of a city’s boundaries, making them either more permeable or more rigid, affecting the sense of geographical identity of its inhabitants. Two extreme forms of these boundaries stand out in Western history. In ancient Greek towns a large part of the population spent the summer in their rural homes. This affected the sense of urban identity of their people, as shown by the fact that a town’s residents congregated into neighbourhoods according to their rural place of origin, maintaining their original geographical loyalties.29 European medieval towns, on the other hand, were surrounded by defensive walls, giving a definite spatial boundary to the jurisdiction of a town’s government, as well as a very clear sense of identity to its inhabitants. As Braudel puts it, these highly territorialised cities ‘were the West’s first focus of patriotism – and the patriotism they inspired was long to be more coherent and much more conscious than the territorial kind, which emerged only slowly in the first states’.30 The development of suburbs, residential and industrial, starting in the mid-nineteenth century, blurred the boundaries of urban centres with clear deterritorialising effects. For a while cities managed to hang on to their old identities by retaining their centres (which became home for train stations and, later on, large department stores) but the further extension of suburbs after the Second World War and the differentiation of their land uses (retail, wholesale, manufacturing, office space. government agencies) recreated the complex combinations that characterise a city’s centre. For the first time the suburbanites could conduct their everyday practices entirely outside of the city. This process, in effect, created brand new centres in the suburban band, deterritorialising the identity of cities.31

But centuries before residential suburbs and industrial hinterlands blurred city boundaries, another process was militating against the strong identity of urban centres: a loss of autonomy relative to the emerging territorial states. Once cities were absorbed, mostly through military force, the local patriotism of their citizens was largely diminished. In some areas of Europe strong urban identities were obstacles to the creation of nationwide loyalties. For this reason, the first European territorial states (France, England, and Spain) were born in those areas which had remained poorly urbanised as Europe emerged from the shadow of the collapse of the Roman Empire. The regions that witnessed an intense urbanisation between the years 1000 and 1300 (northern Italy, northern Germany, Flanders, and the Netherlands) delayed the formation of larger territorial assemblages for centuries. But between 1494, when a French army invaded the Italian city-states for the first time, and 1648, the end of the Thirty Years War, most autonomous cities were brought under control. Indeed, the peace treaty that ended that long war, the treaty of Westphalia, is considered the event that gave birth to international law, that is, the legal system in which territorial states were explicitly recognised as legal actors through the concept of ‘sovereignty’.32

As assemblages, territorial states possess a variety of material components. These range from the natural resources contained within their frontiers (mineral deposits like coal, oil, and precious metals, agricultural land of varying fertility) to their human populations (a potential source of taxpayers and of army and navy recruits). The frontiers (and natural boundaries) defining these assemblages also play a material role in relation to other such large entities. More specifically, each kingdom, empire, or nation-state has a given geo-strategic position relative to other territorial entities with which it shares frontiers, as well as material advantages deriving from natural boundaries like coastlines, which may give it access to important sea routes. After the treaty of Westphalia was signed, future wars tended to involve several national actors. This implies, as the historian Paul Kennedy has argued, that geography affected the fate of a nation not just through

such elements as a country’s climate, raw materials, fertility of agriculture, and access to trade routes – important though they all were to its overall prosperity – but rather [via] the critical issue of strategical location during these multilateral wars. Was a particular nation able to concentrate its energies upon one front, or did it have to fight on several? Did it share common borders with weak states, or powerful ones? Was it chiefly a land power, a sea power, or a hybrid, and what advantages and disadvantages did that bring? Could it easily pull out of a great war in Central Europe if it wished to? Could it secure additional resources from overseas?33

There is also a wide range of expressive components of these larger assemblages, from the natural expressivity of their landscapes to the ways in which they express their military might and political sovereignty. The hierarchies of government organisations operating at a national, provincial, and local scales played a key role in determining how nationalist allegiances would be expressed in nation-states through flags and anthems, parades and celebrations. The cities that became national capitals also played an important expressive role, the best example of which is the style of urban design that became fashionable in Europe after the Thirty Years War. This grandiose style, referred to as the ‘Grand Manner’, transformed the new capitals into Baroque displays of the power of their centralised governments: wide avenues were built and lined with trees; sweeping vistas were created, framed by long rows of uniform facades and punctuated by obelisks, triumphal arches, or statues; all the different design elements, including the existing or modified topography, joined in ambitious, geometric, overall patterns.34

National capitals also played a territorialising role, homogenising and exporting to the provinces a variety of cultural materials, from a standard language and currency, to legal codes and medical and educational standards. The spatial aspects of territorialisation were also there: the controllability of the movement of immigrants, goods, money, and, more importantly, foreign troops across a nation’s borders. While the peace treaty of Westphalia gave frontiers a legitimate legal status, the decades that followed its signing witnessed the most intense effort to enforce these legal borders through the systematic construction of fortress towns, perimeter walls, and citadels. In the hands of the brilliant military engineer Sebastian Le Prestre de Vauban, for example, France’s borders became nearly impregnable, maintaining their defensive value until the French Revolution. Vauban built double rows of fortresses on the northern and south-eastern frontiers, so systematically related to each other that one ‘would be within earshot of French fortress guns all the way from the Swiss border to the Channel’.35 The main deterritorialising processes in these assemblages are those that affect the integrity of national frontiers. These could be spatial processes such as the secession of a province, or the loss of a piece of territory to another country. But they could also be border-defying economic processes. As the frontiers of territorial states were becoming solidified after the Thirty Years War, some maritime cities which had resisted integration were creating commercial and financial networks that were truly international. Such a maritime city was Amsterdam, which displayed its lack of attachment to a territory by selling weapons to both sides in that conflict. By the seventeenth century, Amsterdam had become the core of what is today called a world-economy: a large geographical area displaying a high degree of economic coherence as well as an international division of labour.36 A world-economy, in fact, had existed in the West since the fourteenth century, with Venice as its core, but when it acquired global proportions in the seventeenth century it became a powerful deterritorialising process for nation-states, governing economic flows that, to this day, can easily cross political frontiers.

This admittedly simplified description of society as an assemblage of assemblages should serve as a reminder of how misleading it is to view human history as comprising a single temporal flow. Indeed, given that even at the largest scales (territorial states, world-economies) we never reach a point at which we may coherently speak of ‘society as a whole’, the very term ‘society’ should be regarded as a convenient expression lacking a referent. This is perhaps the way to treat terms like ‘the social field’ or ‘the socius’, terms that constantly appear in Deleuze and Guattari’s version of assemblage theory: convenient general expressions that can be replaced when necessary by a description of a concrete social whole. Let’s examine a particular case to show how this can be achieved, that of institutional organisations like prisons, hospitals, schools, barracks, and factories. These are, of course, the ‘species’ of organisations whose mutation during the seventeenth and eighteenth centuries was so thoroughly studied by Michel Foucault. In his book on the subject Deleuze distinguishes the two articulations involved in the production of these social entities this way:

Strata are historical formations . . . As ‘sedimentary beds’ they are made from things and words, from seeing and speaking, from the visible and the sayable, from bands of visibility and fields of sayability, from contents and expressions. We borrow these terms from Hjemslev, but apply them to Foucault in a completely different way, since content is not to be confused here with a signified, nor expression with a signifier. Instead, it involves a new and very rigorous division. The content has both form and substance: for example, the form is prison and the substance is those that are locked up, the prisoners . . . The expression also has a form and a substance: for example, the form is penal law and the substance is ‘delinquency’ in so far as it is the object of statements. Just as penal law as a form of expression defines a field of sayability (the statements of delinquency), so prison as a form of content defines a place of visibility (‘panopticism’, that is to say, a place where at any moment one can see everything without being seen).37

Deleuze is here distinguishing the two articulations roughly along the lines of the discursive (coding) and the non-discursive (territorialisation). Non-discursive practices of visual surveillance and monitoring, performed in buildings specifically designed to facilitate their routine execution, sort the raw materials (human bodies) into criminal, medical, or pedagogic categories; and discursive practices, like those of criminologists, doctors, or teachers who produce a variety of conceptual categories, consolidate those sorted human materials, giving prisons, hospitals, and schools a more stable form and identity. The only problem with this formulation is the absolute use of the micro–macro distinction. For Deleuze and Foucault, the visible and the articulable define a ‘disciplinary age’, that is, a historical period defining a whole ‘society’. But this last move is unnecessary. It only seems to be required because Deleuze’s social ontology is made up of only three levels: the individual, the group, and the social field.38 The communities and interpersonal networks we have been discussing can be considered as ‘groups’, but institutional organisations certainly cannot. And moreover, the transformation that Foucault documents occurred to a population of organisations, not to ‘society as a whole’. In addition, the components of these assemblages (which are strata since they have high values for both parameters) must be specified more clearly. In particular, in addition to the people that are confined – the prisoners processed by prisons, the students processed by schools, the patients processed by hospitals, the workers processed by factories – the people that staff those organisations must also be considered part of the assemblage: not just guards, teachers, doctors, nurses, but the entire administrative staff. These other persons are also subject to discipline and surveillance, even if to a lesser degree.

Let’s examine another example of how enriching a social ontology beyond individuals, groups, and social fields can help us frame historical explanations. The term ‘groups’, as already noted, could be seen as an unproblematic substitute for ‘communities’ or ‘interpersonal networks’. But what about the larger wholes that communities form to pursue collective political action? Are those just larger groups? Or do they possess emergent properties that make them a different social whole? If the latter is correct then we must consider both the properties of the communities composing the movement, such as their degree of solidarity, as well as the properties that can emerge from alliances or coalitions. These coalitions played an important historical role when expressions of political dissent were transformed in the eighteenth century from machine-breaking, physical attacks on tax collectors, and other forms of direct action, to the very different set of displays characteristic of today’s public demonstrations. This is a change in what the historical sociologist Charles Tilly calls repertoires of contention: the set of performances through which collective actors express their claims to political rights. These expressive repertoires changed dramatically during the Industrial Revolution to include public meetings, marches, distinctive slogans, pamphlets, and statements in mass media. Through these means a social justice movement could express that it was respectable, unified, numerous and committed – in short, that it was a legitimate collective maker of claims in the eyes of both its rivals and the government. As Tilly puts it,

Claim making becomes political when governments – or more generally, individuals or organizations that control concentrated means of coercion – become parties to the claims, as claimants, objects of claims, or stake holders. When leaders of two ethnic factions compete for recognition as valid interlocutors for their ethnic category, for example, the governments to which interlocutors would speak inevitably figure as stake holders. Contention occurs everywhere, but contentious politics involves governments, at least as third parties.39

Once a social justice movement manages to win rights from the government – collective negotiation rights, reproductive rights, voting rights – it must then build an organisation to protect those rights and to lobby legislative bodies to influence the writing of any new laws that may impinge on those rights. At that point, the movement becomes a hybrid assemblage of communities and at least one organisation. Are we supposed to capture this complexity through the concept of a group? The limitations of Deleuze’s social ontology are even more visible when it comes to the ‘social field’, a concept used to frame historical explanations in terms of the ‘feudal social field’ or the ‘capitalist social field’. The problem is not, of course, the term ‘field’, since simply changing it to ‘system’ does not make these explanations any better. The term ‘capitalist system’ (or ‘capitalism’) used to belong to the left, but since the 1980s it has also been adopted by the right, the only difference being that while one side demonises it the other side glorifies it. The result is that the concept has degenerated into a term that is part of a morality tale. Deleuze and Guattari have attempted to breathe new life into the concept by redefining it as an axiomatic of decoded flows.40 An axiomatic is, in the field of logic and mathematics, a small body of self-evident truths from which an infinite number of theorems can be derived. Similarly, the authors conceive the ‘capitalist system’ as capable of deriving an infinite number of new entities – technologies, customs, fashions, financial instruments – in a way that is compatible with the overall system. This characterisation aims to contrast the flexibility of ‘capitalist’ coding with the rigid form of coding performed by a state apparatus: fixed codes of behaviour and dress for different social classes; fixed laws based on tradition; fixed repertoires of technology kept closed by fear of innovation. As such it performs a useful theoretical task since the distinction between a closed code and one that is open-ended and versatile is important. This distinction could be captured by modelling ‘capitalism’ as an assemblage, and setting the second parameter to the appropriate values. The question, however, remains: is there such a thing as an economic system that encompasses all of society?

Deleuze and Guattari, belonging as they did to the Marxist tradition, would undoubtedly say ‘yes’. But the economic historians who are the experts on the subject are not so sure. Fernand Braudel, for example, claims that: ‘We should not be too quick to assume that capitalism embraces the whole of western society, that it accounts for every stitch in the social fabric.’41 And he goes on to say that ‘if we are prepared to make an unequivocal distinction between the market economy and capitalism . . . economic solutions could be found which could extend the area of the market and would put at its disposal the economic advantages so far kept to itself by one dominant group of society’.42 These are powerful words. But how can anyone dare to suggest that we must distinguish capitalism from the market economy? These two terms are, for both the left and the right, strictly synonymous. However, a close examination of the history of commercial, financial, and industrial organisations shows that there is indeed a crucial difference, and that ignoring it leads to a distortion of our historical explanations.

More specifically, what Braudel’s research has provided is solid evidence that there have been two very different economic dynamics in the West ever since the fourteenth century: wholesale was never like retail; high finance was never like ordinary credit or money-lending; and large industrial production was always different from small-scale industry. These two dynamics are explained by him as differences in individual economic organisations, not as aspects of an overall social field. In addition, Braudel adds another social assemblage: cities. As we argued above, there is an important distinction between towns trading commodities with each other (in progressively larger trading zones) using the information about demand and supply carried by prices, and maritime ports engaged in international trade of luxury goods linking together markets with disconnected demand and supply. Since, strictly speaking, only trade that uses prices set by impersonal forces can be considered a market economy, Braudel refuses to use that term for any economic activity in which supply and demand are being manipulated, whether this manipulation is performed by taking advantage of price differentials in distant markets, or whether it is an outcome of the exercise of economic power. As the previous quotation shows, he uses the term ‘capitalism’ exclusively for the latter. Using Braudel’s ideas we can try to make sense of Deleuze and Guattari’s insight that economic assemblages in the West underwent a variety of decoding (and deterritorialising) processes.

Let’s start with deterritorialisation. An industrial firm that generates wealth through economies of scale can be said to be deterritorialised in a variety of ways. It most likely has the legal form of a joint stock corporation, an organisational structure in which control of day-to-day operations has become separated from ownership: managers, who move freely from one corporation to another, exercise control, while ownership is dispersed among many stockholders. This is in contrast with small firms run by an entrepreneur who is both the owner and the person who supplies direction for the firm. Entrepreneurs do not move freely between firms but tend to stay put in a single one, their own. Large scale also allows a different type of increased mobility. A corporation tends to internalise a variety of economic functions through vertical integration (buying its suppliers or distributors) or horizontal integration (buying firms in different sectors of the economy). This internalisation, in turn, gives large firms geographical mobility by making them self-sufficient, since they can relocate factories and headquarters to any part of a country (or today the world) that offers them lower wages and taxes. By contrast, small firms, particularly those that exist as part of a network of other small firms and that depend on the agglomeration of talent in a particular geographical area, lack this mobility.

Decoding, on the other hand, displays the opposite asymmetry. There are two ways in which the allocation of scarce resources can be performed: the first uses centralised planning and the issuing of commands, the second uses the information carried by prices to perform large numbers of decentralised decisions. Historically, commands and central plans are associated with ancient government organisations, that is, with highly coded assemblages, while prices and their spontaneous movements are linked with merchants operating at the periphery of those highly coded areas, performing a local decoding of the movement of goods. The spread of prices as the basis for allocative decisions that began in Europe in the eleventh century can therefore be viewed as an intensification of the decoding of flowing commodities. The problem for Deleuze and Guattari is that the kinds of organisations that Braudel would classify as capitalist tend to replace prices by commands and plans. Institutional economists like John Kenneth Galbraith were perhaps the first to argue that this replacement was evidence that large corporations did not belong to a market economy but to what he called ‘the planning system’.43 Organisations in which control and ownership have become separated, and in which commands have replaced prices, are a relatively recent phenomenon: prior to the middle of the nineteenth century, the population of economic organisations was dominated by small firms, the only examples of corporate assemblages being the famous Companies of Indies, which were like a ‘state within the state’.

But Braudel gives us other examples that reach further back in time, questioning the existence of a single ‘system’ from the start. These other social wholes that can also be said to have undergone deterritorialisations and decodings are urban centres. Cities and towns can be classified in different ways, but a relevant distinction for present purposes is between landlocked cities acting as regional capitals – and later on, national capitals – and maritime ports acting as gateways to the outside, through their participation in international trade. Until the advent of the locomotive, sea transport was much more rapid than its terrestrial counterpart, and this, together with day-to-day contact with the open spaces formed by the seas and oceans, made maritime ports less territorialised than landlocked cities. As early as 1300 these deterritorialised cities entered into shipping networks through which everything – goods, money, people, news, contagious diseases – moved at a faster speed. These two types of urban centres also varied in the kind of material and symbolic cultural artifacts that accumulated within them. Regional capitals like Paris, Vienna, or Madrid attracted migrants from throughout the region they dominated, and over a few centuries they slowly distilled a unique regional culture, giving them a well-defined identity. In this sense, these cities were highly coded. Maritime gateways like Venice, Genoa, Lisbon, or Amsterdam, on the other hand, never acquired a sharp cultural identity since they mixed and matched elements from the variety of alien cultures with which they came into regular contact, making their identity less coded.44 Deleuze and Guattari are aware of this, but deny Braudel’s well-documented assertion that these cities were the birthplace of capitalism, as he defines it. These cities were not typically engaged in industrial production – although Venice’s arsenal was the largest military-industrial organisation of its time – but tended to concentrate on trade and finance. Since in the Marxist tradition exchange and credit do not produce value, the cities that engaged in those activities could not possibly have given rise to capitalism. As the authors put it:

There is therefore an adventure specific to the towns in the zones where the most intense decoding occurs, for example, the ancient Aegean world or the Western world of the Middle Ages and the Renaissance. Could it not be said that capitalism is the fruit of the towns, and arises when an urban recoding tends to replace State overcoding? This, however, was not the case. The towns did not create capitalism. The banking and commercial towns being unproductive and indifferent to the backcountry, did not perform a recoding without also inhibiting the general conjunction of decoded flows . . . [Therefore it] was through the State-form not the town-form that capitalism triumphed.45

Why are Deleuze and Guattari so deeply committed to this idea? Because they remained Marxists till the end. Thus, they write: ‘If Marx demonstrated the functioning of capitalism as an axiomatic, it was above all in the famous chapter of the tendency of the rate of profit to fall.’46 The problem with this statement is that this ‘tendency’ is entirely fictitious, its only basis being the labour theory of value. The argument in defence of this tendency proceeds in stages. First, in order to be able to classify wage labour as a mode of surplus extraction, and hence to make it directly comparable with slavery and serfdom, Marx made the assumption that industrial workers were the only source of economic value. If we accept this explanation then we must agree that there is no such a thing as fair wages, since every bit of profit is ultimately a product of labour and wages are only a portion of the profits. Second, Marx made the related assumption that machines are merely the coagulated labour of the workers who assembled them, so any economic value they generate is really created by labour. Finally, he added an extra premise which does have a basis in reality: given that workers can present factory owners with a variety of problems (demanding higher wages or better working conditions) there will be a tendency to replace humans with more docile machines. But as this happens, the ratio of humans to machines, of productive to unproductive assets, will decrease, and with it the rate of profit. The argument is impeccable, but has anyone ever produced evidence that its premises are true? Or to put this in a simple way, has anyone seen a factory run entirely by robots that does not produce any profits? Of course not. Machines are not just a product of labour but, much more importantly, of engineering design and science. And design, both functional and aesthetic, also produces value, as does the organisation of production. Even oppressive forms of industrial organisation (like Taylorism) can be a source of value, though less than is normally assumed because of the hidden costs they carry, such as the deskilling of a worker population and the resulting loss of human capital.

The labour theory of value explains why Deleuze and Guattari would assert that cities like Venice in the fourteenth century, Genoa in the sixteenth, or Amsterdam in the seventeenth were unproductive: if only the labour of workers is productive then trade and finance must be unproductive. Braudel has gathered and analysed extensive evidence that this statement is simply false. But in addition to sticking to dogma from the love of a tradition, there is another reason why Deleuze and Guattari get their economic history wrong: a social ontology consisting of individuals, groups, and social fields. The two kinds of historical agents needed in order to understand Braudel’s analysis, economic organisations and cities, do not have a place in that ontology, and are only brought into the discussion as unanalysed extra factors, to be quickly dismissed. And similarly for ‘the State’. In the sentence following the previous quotation, Deleuze and Guattari mention Braudel’s argument that territorial states eventually won the race against city-states as evidence that only the former supported the emergence of an axiomatic. But here we have another ontological mismatch: the term ‘territorial states’ refers to kingdoms, empires, and eventually, nation-states, as geopolitical entities, not as governments, that is, as assemblages of organisations. The latter are part of the former, and the jurisdiction of their authority coincides with the former’s frontiers, but they are not the same thing, unless, of course, we view both as aspects of the ‘axiomatized social field’. The reason why territorial states eventually engulfed and subjugated previously autonomous cities is easy to understand: they had a larger population and hence a much bigger reservoir of recruits for their armies, and a more extensive potential tax base with which to fund the increasingly expensive arms races. When the French army crossed the Alps in 1494 and invaded the Italian city-states, it brought with it powerful artillery, with greater destructive capacity than the bombards used by the Turks in their siege of Constantinople, and more importantly, mounted on carriages to make it mobile. The old style of defensive fortification, with its tall walls made out of rigid but brittle stone, collapsed when confronted with the new weapons, an event that sent Italian military architects rushing to the drawing board to invent a brand-new style of fortification based on defence-in-depth. But the new citadels with their ditches, ramparts, and bastions were so expensive that city-states found it impossible to keep pace with their larger rivals.47

This is not to say that governments did not have an economic role to play. Let’s return to the argument given above to replace the notion of ‘the Market’ with a series of progressively larger trading areas viewed as assemblages. The assembly of urban bazaars into regional markets occurred largely spontaneously. It was not commanded from above by urban governments but resulted from the decentralised activities of hundreds of merchants, and the reaching of thresholds in the volume and regularity of trade. Provincial markets were also to a large extent the result of impersonal forces. But the creation of the first national market in England did involve an active role for government organisations based in London, like the planned construction of new roads and canals, as well the destruction of provincial tariffs, tolls, and taxes. In a very real sense the pre-existing province-wide trading areas had to be forcibly stitched together into a larger whole. Had governmental organisations in London not spent all the financial and political capital needed for such forceful union, national markets would not have come together, depriving large-scale factories in Manchester and elsewhere of the demand needed to profit from their mass-produced textile goods. Because this massive governmental intervention extended wholesale price synchronisation so as to coincide with national frontiers, we can acknowledge that certain organisations (not ‘the State’) did have decoding effects. But other government organisations had the opposite effect. It is now well established that the discipline and routinisation of labour needed to realise economies of scale are of military origin, born in armouries and arsenals in France and the United States.48 In this case, optimised routines imposed by commands replaced skills and craft, resulting in an overcoding of labour. Deleuze and Guattari acknowledge this other effect in passing when they write that ‘it can be said not only that there is no longer a need for skilled or qualified labor, but also that there is a need for unskilled or unqualified labor, for a disqualification of labor’.49

We will return later to the military origin of industrial discipline, but at this point it should be clear that eliminating a society-wide system and replacing it with a population of organisations that includes military arsenals is necessary to explain economic history. This population may indeed be coextensive with an entire country, in the sense that the reach of national corporations coincides with the jurisdiction of its central government organisations, but this does not imply that they are all part of an overall system, much less that the nature of this system can be understood using reified generalities like the Market and the State. Much of the academic left today has become prey to the double danger of politically targeting reified generalities (Power, Resistance, Capital, Labour) while at the same time abandoning realism. A new left may yet emerge from these ashes but only if it recovers its footing on a mind-independent reality and if it focuses its efforts at the right social scale, that is, if it leaves behind the dream of a Revolution that changes the entire system. This is where assemblage theory may one day make a real difference.

Notes