Scholarly article on topic 'Social Integration and System Integration: Reconsidering the Classical Distinction'

Social Integration and System Integration: Reconsidering the Classical Distinction Academic research paper on "Sociology"

0
0
Share paper
Academic journal
Sociology
OECD Field of science
Keywords
{""}

Academic research paper on topic "Social Integration and System Integration: Reconsidering the Classical Distinction"

SOCIOLOGY Vol. 32 No. 3 491-507

August 1998

SOCIAL INTEGRATION AND SYSTEM INTEGRATION: RECONSIDERING THE CLASSICAL DISTINCTION

Markus Perkmann

Abstract In a recent article in Sociology, Mouzelis argues that Lockwood's distinction between system integration and social integration is, with some modifications, still to be retained because of its logical coherence and its methodological virtues. While its basic value is recognised, this article reconsiders the distinction in the light of some recent achievements in social theory. It is argued that the distinction does not merely offer two different perspectives on society but that, in the actual social world, both dimensions are intertwined in a number of ways. In particular, reflexive actors can draw the distinction themselves which raises questions about the objective status of 'incompatibilities'. The implications are developed through some reflections on the nature of social processes as well as the role of institutions, strategic action and conflicts. On the one hand, it is shown that system integration is an important objective for social agency, since many actions are aimed at controlling the performance of processes. On the other, social integration has a decisive impact on those processes because conflict and cooperation transform the institutional preconditions those processes are based upon. TTie social-system integration distinction offers therefore interesting insights into how social actors themselves attempt to cope with their worlds. Apart from being a methodological tool, the distinction expresses a substantial characteristic of society.

Key words: agency, institution, process, social structure, system.

In a recent article in Sociology, Nicos Mouzelis (1997) presents a brief genealogy of the classical distinction between system integration and social integration.1 Following Lockwood's seminal article in 1964, other prominent sociologists such as Habermas and Giddens attempted to offer a tool for conceptualising the eternal dichotomy of structure and agency in sociological theory. For Mouzelis, Lockwood's understanding is, with some modification, still the most coherent approach. It offers 'logical congruence' because 'social integration' unequivocally refers to the 'co-operative/conflictual relationships between actors' whereas 'system integration' describes the 'compatibilities/ incompatibilities between "parts" that are always viewed as institutionalised complexes portraying different degrees of durability/malleability' (Mouzelis 1997:113-14). This allows for a Lockwoodian 'perspectivism', i.e. the possibility of perceiving 'the same social phenomena from two different perspectives' (1997:114). Habermas's (1987) distinction, for Mouzelis, is logically incongruent with his mechanism of co-ordination because system integration (externalist perspective) is associated with economic and political 'systems' whereas social integration (internalist perspective) is exclusively concerned with the 'lifeworld' (Mouzelis 1997:114-16). Consequently, economies and

from the SAGE Social Science Collections. All Rights Reserved.

politics cannot be conceptualised from a viewpoint of social integration whereas other spheres of society cannot be dealt with from a system integration perspective. For Giddens (1986:139-44), in his attempt to overcome the structure/agency dualism, social integration refers to circumstances of co-presence, and system integration refers to interaction stretched over time-space. Mouzelis objects that this disregards the fact that certain micro-interactions may well entail consequences extended over time and space (1997:116-17). Giddens, according to Mouzelis, is also forced to re-introduce the subjectivist-objectivist divide by the backdoor in the guise of his distinction between 'institutional analysis' and 'strategic-conduct analysis' which corresponds, more or less, to Lockwood's original distinction (Mouzelis 1997:117).

Whereas I agree with Mouzelis's criticisms directed against Habermas and Giddens, I would argue that even a reformulated version of Lockwood's notion is still not satisfactory. Therefore, as a footnote, as it were, to this 'latent' debate, I would like to suggest an alternative view. Epistemologically, an antifoundationalist perspective is adopted, i.e. no substantial assumptions are made about how society is structured. Instead, attention is paid to how 'systems' are actually constructed and reproduced by actors themselves without ever denying the existence of such systems. Thus, whereas the distinction between the 'systemic' and the 'interactive' aspects of society should definitely be retained, theorising the link between them needs to be more developed.

In a first step - after a brief theoretical contextualisation of the debate - I investigate the nature of systemic incompatibilities as implied in the system/ social integration distinction. The notion of 'processes' is introduced for conceptualising the 'systemic' dimension of society because it offers a better understanding of how the latter is linked to the 'social' dimension. In a second step, it is argued that social processes (and, therefore, incompatibilities) are not 'objective' but that they are linked to specific 'sites of agency'. However, actors themselves can conceptualise social reality as a set of objective processes (i.e. in terms of system integration) and employ the latter within their activities. Thirdly, this is specified by looking at the role of institutions and the 'production of effects' through institutional modification. Finally, I show how system and social integration are inter-linked and entangled in the social world. It is argued that system integration is an important objective for social agency, and, in turn, social integration has a decisive impact on the coherence of social processes.

A Brief Theoretical Contextualisation

A few comments on the theoretical context of this debate precede the detailed work. The discussion on the 'artificial' dichotomy (Lockwood) between social

and system integration has to be located within the context of more general problems in social theory. Two traditional issues are of particular relevance here, the problem of social order and disorder, on the one hand, and the problem of structure and agency, on the other. The former is not just concerned with the question of how social order is constituted but, perhaps more importantly, of how social order is understood. For instance, the Durkheimian tradition puts the emphasis on a common framework of shared norms and values as the basis for social order (Lockwood 1992). To the contrary, in classical Marxist theorizing, the integration of capitalist societies is thought to be based on a 'matching' articulation between productive forces and the relations of production (Lockwood 1992). In the former case, 'social integration' is stressed whereas in the latter, 'system integration' is held responsible for social order. The precise relationship between social and system integration is already implicit in the specific meta-perspectives on society adopted. Lockwood's early call for an analytical integration of social and system integration was therefore nothing less than the defence of a certain model of society where both the interactive and systemic aspects of social order would be taken into consideration:2 'The problem of social integration . . . cannot be fully understood without taking account of the ways in which changes in both the normative and realistic conditions of action are usually the unintended consequences, or system effects, of the interrelations of a society's economic, political, and religious subsystems' (Lockwood 1992:377-8; emphasis added). Although one can agree with Lockwood's general line, a certain ambiguity remains in two aspects. First, how real are those 'realistic' conditions of action, and, secondly, what is the social ontology behind those presupposed 'subsystems'? In other words, how are functional processes in societies to be understood? Before dealing in more detail with these issues, a few remarks on a second 'grand theme' of social theory must be made.

The structure-agency problem is concerned with the paradoxical relationship between the structural determination and the transformative potential of agency. Although the structure-agency issue has certainly to do with the social-system integration distinction, they are not identical. Whereas the paradoxical relationship between structure and agency is at the root of a theoretical problem, the distinction between social and system integration is rather a theoretical tool. Consider, for instance, that a 'structuralist' solution to the structure-agency issue can be combined either with a view focusing mainly on social integration (normative functionalism), or, alternatively, on system integration (structuralist Marxism). Thus, a rash identification of social integration with agency, and system integration with structure is inappropriate. As can be inferred from Lockwood's original contribution, agency can affect system integration,3 and structural patterns can be conducive or obstructive to social integration.

Whereas Lockwood emphasises the need to take into consideration both social and system integrative mechanisms for the analysis of social order, a

number of recent contributions to the structure-agency debate have pointed to both the recursive nature of social structure, and the reflexive capacities of agency. From such a perspective, a 're-entry' of structure into agency can occur, i.e. the situation where conditions of systemic reproduction are reflexively and cognitively appropriated by actors and employed in their actions.4 I focus in particular on this latter aspect during the following discussion of the social-system integration distinction.

Giddens (1986, 1994), in his theory of structuration, holds that structural moments are both moments and the result of social practices (duality of structure). In particular, actors are assumed to be knowledgeable subjects and capable of reflexively shaping the conditions of system reproduction. As Giddens points out, in organisations particularly the reflexive guidance of the conditions of systemic reproduction is essential.

Jessop (1996), from a strategic-relational perspective, 'relativises' the absolute opposition of 'external constraints' and 'free-willed actions' and turns it into the more dialectical opposition of 'strategically inscribed strategic selectivity' and 'strategically calculated oriented action'. In this view, structural constraints 'do not exist outside of specific spatial and temporal horizons of action pursued by specific actors' (1996:126). In turn, agents, as learning 'strategically calculating subjects', are capable of modifying structural contexts. In the words of Jessop, such an approach 'does not posit abstract, atemporal and unlocated structures or wholly routinised activities performed by "cultural dupes'" (1996:126).

Inspired by Mouzelis's and Lockwood's arguments on systemic incompatibilities, in the following I aim to reconsider 'structural constraints' by taking into account ideas of recursive, reflexive and cognitively grounded action. The more (Jessop) or less (Giddens) radical relativisation of 'structure' obviously has consequences for the very status of incompatibilities or contradictions. The question is where the 'objectivity' of structure has ended up. Could it be that the constitution of structural constants and the subsequent instantiation of constraints can be linked even closer to the transformative and strategically calculating capacities of (in particular: collective) action? Bearing in mind specifically the strategic-relational reflections, this hypothesis is examined in the next section.

System Integration Reconsidered

For Mouzelis, system integration refers to compatibilities/incompatibilities between institutionalised complexes with different degrees of malleability (1997:113-14). What do incompatibilities exactly consist of? And also, from what perspective are those complexes incompatible? In Mouzelis's account it seems that systemic contradictions occur merely because the more stable institutions affect the more malleable ones. This might be correct but it is too

abstract to understand exactly how this works. The notion of system integration here seems to invoke the functionalist argument of, in some way, 'objective' contradictions. This is why it is crucial to elaborate on the link between 'structure' and 'agency'.

In this regard, and this is in line with Mouzelis's position, two requirements have to be fulfilled. On the one hand, voluntarism must be avoided, i.e. it cannot be assumed that any actors have full control over any 'system'. This does, however, not mean that strategic analysis should not remain possible. On the other, the a priori ontological assumption that society consists of a series of 'systems', 'structures' or 'infrastructures' has to be rejected without, however, dismissing functional analysis (Mouzelis 1995). Allowing for both strategic and functional analysis implies that 'perspectivism' (the option to look at the social world from, at least, two perspectives) is to be maintained. This is crucial because perspectivism mirrors the fundamental dialectic condition of the social world: on the one hand, society can be regarded as an object (i.e. as 'system') but, on the other, this apparent object is, at the same time, 'subject' to transformative practices. Society is necessary and contingent at the same time, depending on the standpoint one adopts. Those two perspectives can be called 'externalist' and 'internalist' (Mouzelis 1995).5 However, the two perspectives do not merely exist for the external observer (the social scientist) but for social actors as well. The methodological distinction therefore re-appears as a substantial feature within the object of investigation and is an important cognitive resource actors draw upon. This statement will be made more explicit below.

With regard to the nature of systemic incompatibilities, two observations are of particular relevance: first, almost trivially, systemic incompatibilities can only occur if they are perceived as such at least by some actors. Secondly, incompatibilities can only be specified when related to specific social processes occurring over time. Starting with the latter, in the following I will attempt to specify the implications of these two points for a reworked distinction of system integration and social integration.

A 'systemic incompatibility' can be defined as a dysfunctional interaction between parts of'systems' or between 'systems'. Thus, incompatibilities are a 'failure' of a system to perform 'properly'.6 As such, they can only occur over time. There are no 'synchronic' incompatibilities, but they become virulent only when followed over time. Thus, incompatibilities are a characteristic of processes through time. A process can be defined as an ordered sequence of events over time.7 Rather than being viewed as 'systems', processes are better understood as a variety of 'programmes of action' if one wants to adopt the heuristic offered by actor network theory (Law 1991). In this view, society can be seen as consisting of a multiplicity of overlapping and entangled processes on all levels and within different realms. Processes comprise such different examples as the 7 per cent growth per annum of the Hong-Kong economy, the performance of the market for refrigerators, an environmental boycott

against a multinational or even 'global warming'.8 Their crucial characteristic is that they are 'mastered' (though not dominated) by some specific actors (government, government agencies, management), requiring, at the same time, the 'enlisting' of a variety of less concerned actors (consumers, workers, environmentalists).

To keep processes in motion, their permanent maintenance through a myriad of social actions is required. Those latter are 'ordered' according to a variety of institutional patterns: social structure.9 'Social order' would prove to be rather 'psychotic' if there were no institutions. Thus, the existence of processes relies on many dispersed institutional regularities occurring at any moment in time. In other words, diachronic processes are based upon synchronic institutional regularities. Those latter constitute 'social structure' which is - due to its inertia - beyond the immediate reach of actions attempting their modification or abolition. Consequently, no process can be controlled by any actors. But, because of its incidence on the performance of processes over time, institutional modification is one crucial option for shaping a process. In this way, the future is changed by changing the present (since a direct shaping of the future proves logically impossible). Though social actors are precisely concerned with securing themselves a relatively coherent future, the only thing they can do is to intervene into 'present' social structure. This requires a certain anticipation of the potential effects of such interventions on the envisaged processes.

This leads over to the first point on agency mentioned above. If processes are sequences of events, they can be cognitively appropriated or even shaped by social actors. Processes, whether it is economic growth, profit maximisation, the performance of a non-governmental organisation or an individual career, do not exist outside of some site of agency. In other words, from an antifoundationalist standpoint, no 'systems' such as economics, politics or law exist. Processes are always under the particular attention of somebody, be it the majority of social actors or just one individual. Thus, different processes are of different relevance for different actors (whether they support them or contest them).

It follows from these considerations that systemic incompatibilities can be seen as malfunctions within processes as they are perceived by concerned actors.10 This presupposes that the 'proper' performance over time must be at least in part anticipated; otherwise unforeseen events would be no surprise and one could not speak of malfunctions. Therefore, one can find the criteria for incompatibilities within the cognitive conceptualisations of those processes retained by the actors themselves. Consequently, no 'objective' criteria for incompatibilities have to be established since they are delivered by the actors themselves! From the viewpoint of such a more radical perspectivism, or 'relationism' (Latour 1993), systemic contradictions are co-constituted with a site of agency. Though they are not objective in a strict sense but they are objectified by, and for, the relevant actors. Since the

existence of processes cannot be presupposed, it is crucial to look at how they come into existence and how they are sustained over time. This must, however, not imply an 'overly postmodernist' position, i.e. denying their existence at all. Here the metatheoretical programme of 'New French social science', such as the 'sociology of dispute' and 'convention theory' (Wagner 1994, Boltanski and Thevenot 1991, Orlean 1994) offers a useful heuristic for such thinking.

What are the consequences for the notion of 'objectivity'? Necessarily, reducing the complexity of the social world involves a selective cognitive appropriation of reality.11 This does not, however, imply that it is an arbitrary exercise. Actors face a 'resistance of the object' (Malpas and Wickham 1995: 46) when trying to realise their intentions, visions and projects based upon a certain 'image' of reality. Contradictions are 'objective' only in so far as they appear as an emergent property within single or interacting processes. But at the same time they are also 'subjective', since their perception depends on how the complexity of the social world was reduced to a relatively coherent social reality.

This is not to say that such an 'objectivity' cannot be highly generalised and commonly accepted. In such a 'hegemonic' situation a certain interpretation of social reality is shared by a majority of individual and collective actors. Consider a 'classical' big-scale process, a 'Fordist' economy where steady growth rates were achieved through an articulation of a multiplicity of institutional regularities. Centralised wage negotiations, central bank policies, macro-economic policies but also certain norms of consumption led to the emergence of a certain process through time: a relatively steady growth rate with low unemployment. In the late 1960s and 1970s, however, a series of 'contradictions' emerged, such as rising unemployment, high inflation and decreasing profit rates. Apparently, the existing institutional regularities did not sustain the Fordist accumulation process as they had done before. But are the contradictions really as objective as they appear? In the following, I would like to show that such incompatibilities have more to do with social agency than it appears in the first moment.

Incompatibilities for Whom?

Consider again the example of Fordism where a model of how economies 'should' work was generalised over a majority of important social actors, including governments, business and trade unions. For different reasons, a high unemployment rate would have been seen as a serious 'incompatibility' by most actors. Over time, however, the process evolved in a way that the trade-off between low unemployment and other objectives such as 'adequate' profit rates, inflation rates and public expenditure levels increased steadily. Thus, when unemployment began to rise this was, for certain actors, not

necessarily an 'incompatibility5 any more since - due to their interests and/or persuasion - they were more concerned with labour costs, inflation and state deficits. Thus, system integration turns out to be a question of perspective. It has to do with how a process is supposed to function from the viewpoint of specific social actors, and, thus, with a selective appropriation of reality. One could argue that there is a systemic contradiction between the patterns of income distribution in 'post-Fordist' societies and the prospect of long-term economic growth (cf. Lipietz 1996). Since the relative purchasing power of a widening part of even the working population is declining, internal macro-economic demand is weak and public sector budgets tend to rise. But one must add that this 'contradiction' is only problematic from the viewpoint of a certain (however defensible!) social project focusing on steady economic growth, equal income distribution and a just distribution of working hours. For the actors organised on the top of the 'hour glass' (Lipietz) this is not particularly an issue - given their increased gains from such a new state-society-economy formation. At least this holds as long as no other adverse effects arise, such as decreasing social cohesion (crime), which are seen as important enough to be countered.

Thus, sites of agency are intrinsically linked with their 'own' objectivity from which systemic incompatibilities are perceived. As Mouzelis points out, they are, in a sense, indeed 'normative'. However, systemic (in)compatibilities do not only owe their existence to the normative dimension underpinning institutional regularities, but their perception itself is a consequence of a particular 'normativity'. Thus, there are many objectivities (and normativities linked to them) across social reality.

To come back to the notion of processes and social structure, this means that specific processes exist only for certain actors. The 'economy' is very different for the coalition between the Chancellor and the Bank of England and an unemployed person suffering from a monetarist low inflation policy. Each of them also have their own strategies and methods of how to cope with the process they are dealing with: 'running' the economy and looking for a job. Here a useful readjustment of the internalist-externalist distinction can be adopted. The internalist perspective is not just a social scientist's viewpoint but exists for every social actor. From this standpoint, actors see themselves as autonomous social subjects equipped with a transformative capacity instantiated by their normative orientations. In order to realise their projects and visions, they have to form coalitions, to mobilise support, to shape identities and they might sometimes provoke protest. Thus, to be capable of dealing with these normative and interest-led issues, is an important matter for actors attempting to achieve social change. In this sense, the production of 'social integration' is not something which is automatically achieved (as part of the 'social order') but is dealt with actively and consciously by actors themselves. If the Bank of England wants to raise interest rates, this is not only a question of efficacy (for containing inflation) but also of a certain social acceptance.

The same applies to the externalist perspective. Not only social scientists, but also social actors are capable of perceiving systemic incompatibilities. To what extent and on which level this 're-entry' takes place, depends on the actor's positioning within the institutional structure of society. For government, malfunctions in the economy can be substantially different from malfunctions perceived from the viewpoint of the profit maximisation strategies of firms or individuals seeking (quality of) work. Another example is the illustration of Weber's work on 'patrimonialism' in Lockwood's original article (1964). According to Lockwood, in patrimonial bureaucratic societies, taxation is 'the strategic functional problem' for the state bureaucracy (1964: 253-4). For state authorities, taxation is at their discretion (internalist viewpoint), but too high a level of taxes will unavoidably lead to adverse effects on tax revenue and governability (externalist viewpoint) as other interest groups will contest the measures delivered. In turn, these other groups will be subject to analogous limitations.

In sum, the internalist-externalist distinction (or, in the words of Archer 1996, 'analytical dualism') is doubled over and over in society. Due to this 'stroboscope effect',12 any systemic (in)compatibility on one level is, at the same time, generated by agency on another level. Equally, any agency results in systemic effects on other levels. It is crucial that this does not just occur to social science but to any actor in society in the same way.

Institutions

These ideas can be adopted for a closer view on the nature of institutions. Two perspectives can be identified. On the one hand, institutions can be seen as sets of rules, forming a 'virtual order' (Giddens 1986:17) or the 'paradigmatic dimension' (Mouzelis 1995:76) of society. On the other, institutions are observable regularities, sets of practices extended in time and space (Giddens 1986:17), or specific 'syntagmatic' configurations (Mouzelis 1995: 77). The first notion is 'internalist', because the adherence to rules and norms rests with the normative orientation of the actors. The second is 'externalist' because the behavioural regularities constituted by institutions are on the basis of 'objective' social processes, i.e. they affect systemic compatibilities and incompatibilities.

From the internalist perspective, what an institution is depends on its signification and its embedding into the value system of actors. In this sense, they also serve as criteria for action and non-action. This implies that they differ from actor to actor. As such, they are an important factor in the constitution of identities. From the externalist standpoint, institutions are certain patterns of synchronic 'social structure' and as such they constitute processes through time. Independently from whether they are acknowledged or not, institutions always have effects because they constitute certain patterns

in the form of 'statistically relevant' movements (Demirovic 1992:151). The important point is that these regularities show a 'diffuse efficacy', i.e. they can be constitutive for a potentially infinite number of processes and they exist independently from them. Such an assumption is indispensable, because any attempt to make a causal link from those institutional effects back to specific processes would inevitably result in functionalism.

However, as mentioned earlier, the externalist standpoint can be adopted also by social actors. This means, that institutional regularities can be 'instrumentalised' by social agency since institutional modification is the most important means for the production of effects. Actors modify institutions for achieving effects on 'their' processes.13 Only in a completely 'unreflexive' society would such effects be completely ignored by the social actors themselves. In fact, one of the most important requirements for social action is knowing about institutional regularities and employing them within projects and strategies. Institutions make social life intelligible for social actors (including social scientists).

Because, from the internalist viewpoint, institutions are linked to the identity of actors, institutional modifications can often be disruptive for the affected individuals or groups. Since this might have negative effects on the process pursued by the modifying actors (because anticipated regularities might be changed) this may even be (cynically or philanthropically) taken into account. If this does not work, a conflict is likely. In the light of the arguments presented earlier, conflicts can be interpreted as a contestation of institutional regularities. From the internalist perspective, this might appear as a defence of identities whereas, from the externalist perspective, incurred institutional modifications are crucial for the performance of processes, and therefore, indirectly, for the future positioning of actors. Most often, social actors are certainly aware of this more 'functional' or instrumental aspect. In many cases, arguments used in social conflicts invoke a 'logic of efficiency' (vs. a 'logic of equality') which is exactly adopting an externalist viewpoint on the part of social actors.14 Co-operation, by contrast, means readjusting institutional regularities through negotiation, concerted action and co-ordination to achieve certain effects on processes thought to be beneficial for the participants.

The capacity for institutional modification is unevenly distributed among actors and also varies according to the specific objectives pursued, due to the strategic selectivity of institutions (cf. Jessop 1990). Specific sets of institutions, actors and projects are configured within 'centres of power' (Scherrer 1995:478) providing privileged positions for reconfiguring synchronic structure. Hence, these centres of power allow for the production of effects through large-scale institutional re-articulations. Marginal variations can have major effects 'at a distance' in time and space and across organisational boundaries, because their normalising function lies at the root of the constitution of processes. A modification of labour legislation can have major consequences

for the 'competitiveness' of a country. Essentially, institutions allow for the amplification of social agency. Due to these effects on processes, institutional modifications always involve a repositioning of actors not just momentarily but over time. Changing large-scale regularities of action requires, in most (non-coercive) cases, a discursive mobilisation and is likely to involve the transformation of individual and collective identities. Vice versa, the proliferation of narratives possibly affects the efficacy of institutions for specific processes and can therefore result in material consequences for the complex variety of processes monitored by different actors. However, no institutional modification can be fully 'controlled' by any actor. Monetary policy authorities can change the interest rate but cannot directly influence the employment policies of firms; the unemployed person cannot influence the interest rate but can accept a lower wage or different work. Although there are different degrees to which actors can instrumentalise institutions for their own strategies and projects, they rely in part on the outcome of path-dependent evolution. This is where knowledge and social technology come in, since actors may take certain 'evolutionary' patterns into account and trust in 'meta-governance' Qessop 1995). The selective attention of actors leads to a duality of strategy and evolution.

Entangling System and Social Integration

The dichotomy between conflict/co-operation and compatibilities/incompatibilities is less clear-cut if one assumes reflexive and knowledgeable actors. For instance, conflicts between actors arise, in part, from their perception of incompatibilities. This happens when business contests high wages because of the supposedly negative effects on individual firm's competitiveness. Often, conflicts are fought in the name of such 'objective' functional requirements. Similarly, coalitions among actors can be formed because each of them regards them as beneficial for the processes they are engaged with: a 'peaceful' corporate climate reduces the hours lost due to strikes and can therefore compensate for higher wages. Thus, co-operation is both beneficial for the processes of profit-maximising and negotiating higher wages or minimum wages for workers. This presupposes that actors are knowledgeable and reflexive about these different processes and are able to shift between internalist and externalist viewpoints. The distinction, therefore, repeatedly re-appears within society (stroboscope effect). Through cognitive modelling, reflexive monitoring and learning, actors are able to draw the process-project distinction themselves. They themselves select which process they regard as hard to influence, and which they consider to be open for strategic shaping. In this sense, actors engage in the 'art of complexity' Qessop 1997), consisting both, on the one hand, in the art of drawing up viable projects and appropriate implementation techniques ('social technology') and, on the

other, in the art of obtaining support and gaining legitimacy ('polities' in a wider sense). As a consequence, there is an epistemological isomorphism between the social scientist's account and the actor's intelligence since they rely on knowledge and techniques for their transforming practices. Mouzelis acknowledges the importance of the cognitive dimension when he distinguishes between first- and second-order discourses. What he locates in the second-order discourse, i.e. the 'attempts by specialists to understand, criticise, legitimate, defend . . . institutions' (1997:n.3), is exactly how all actors try to understand and operationalise the social processes they are concerned with. Mouzelis is right in pointing out that some of the 'discourse practices' are more malleable than others but it must be added that the distinction between first and second order discourses is not necessarily a hierarchical one. The 'lay' first-order discourses may in fact contain 're-conceptualisations' of the 'expert' second-order discourses instantiated in the 'management' of processes, such as the monetary policy of the central bank. Consequently, discourses of much more than two orders exist together with the entangled and recursive co-existence of processes.

An implication of the last point is that a 're-entry' of system integration into social integration can occur. In other words, perceived patterns of systemic compatibilities can be employed by social agency, i.e. for forming coalitions or contesting existing co-operative arrangements. Actors can take such patterns for granted, and conceive them as 'black boxes'.15 Such attempts to 'objectify' complex patterns relying on a multiplicity of dispersed actions can be seen, in the language of actor-network theory, as 'punctualisations' (Gallon 1992:95); they are 'quasi-objects' (Latour 1993). In this sense, compatibilities can be employed by actors to pursue certain goals ('agency'). Furthermore, in big organisations not only the external environment is 'black-boxed', but also their 'internal environments' in the sense that their successful operation relies on numerous routines and standardised procedures.16

However, from the internalist perspective, system integration is always the result of certain co-operative-conflictive relationships since each actor concerned is, in principle, 'free' to act as she/he wants. Management might assume that higher wages or bonuses bring about higher efficiency because of the higher commitment of workers. Thus, they 'black-box' the behaviour of their employees into a simple correlation ('efficiency wages'), whereas for workers themselves their real behaviour depends on a complex articulation of many different factors and on their decision on how to act. Or, as Mouzelis points out, hierarchies can be employed as 'technologies' in the 'construction, reproduction and transformation' of 'macro spaces' by macro actors (1995: 146-7). Thus, system integration and social integration are continuously turned into each other, depending on the standpoint and the intention of different actors.

Another consequence is that attempts at strategic guidance on one level of agency can appear as an evolutionary trajectory on other levels. In other words,

what appears to be a structural constraint for one actor, appears as an opportunity for transformation to another actor (Jessop 1996:125). For instance, the strategic re-configuring of synchronic structures (higher wages) appear as (and are in fact) exogenous effects to these 'black-box' actors whether they are aware of them or not. In terms of Mouzelis's sophisticated framework, the latter find themselves in a position of 'paradigmatic duality5 and 'syntagmatic dualism' whereas the 're-configuring' actors operate in a position of'paradigmatic dualism' and 'syntagmatic duality' (Mouzelis 1995: 120-1). Put more simply, what for some actors appears to be a matter of fate and appears as objective 'evolution', is strategically shaped by others.

Yet, not even for the actors pursuing the strategic intervention can any object be fully under their control: reality 'resists'. Thus, the success of a project will always also depend on evolutionary dynamics which have not been foreseen. Even if an actor had the social side 'under control' (consensus and commitment), the 'systemic' side would be too complex to be completely mastered. However, actors can acquire some knowledge on some evolutionary dynamics and develop monitoring techniques. Thus, for actors the system-social integration distinction is never unambiguous since it is a matter of judgement what they should regard as 'systemic' and what as liable to transformation depending on what they intend to do.

Concluding Remarks

Mouzelis's reading of the system-social integration distinction tends towards an objectivist understanding of systemic (in)compatibilities. I suggested that there is no 'absolute' externalist outside from which society can exclusively be perceived as a 'system', because this perspective can always be turned into an internalist 'inside', i.e. a contingent field of action potentially liable to transformation. The concept of 'process' was proposed as an alternative to the 'system' perspective. If a process is an ordered sequence of events over time, any disturbance of that 'order' can be seen as a systemic incompatibility. Since such a 'malfunction' has to be observed and conceptualised from some point of observation, processes are linked to the specific 'objectivities' of certain social actors. Thus, incompatibilities are never objective but linked to specific sites of agency. In part, actors themselves regard their social worlds as a set of entangled processes and attempt to cope with occurring incompatibilities. However, this does not imply that processes (or social integration) are at the discretion of actors. Reality tends to resist, given that society consists of many entangled processes which are co-constituted with 'their' respective actors and constitute 'real' constraints for others. But since actors are reflexive and knowledgeable about the social processes going on around them, they can attempt to 'black-box' the actions pursued by other actors and consider them as 'systemic'. For this purpose, institutional modification is an important

resource at their disposal. Because institutions are decisive for the performance of processes, a change of these regularities can be employed for an adjustment of those processes, which means removing incompatibilities and malfunctions from a specific standpoint. This involves either conflicts and/or co-operation. Thus, system and social integration do not merely co-exist as two independent viewpoints but they are entangled and intertwined: on the one hand, compatibilities/incompatibilities are of crucial importance for actors in coping with processes, and, on the other, conflict/co-operation can result in significant effects on processes. Actors can engage in conflict/co-operation for the sake of establishing compatibilities, but compatibilities can be perceived as incompatibilities by others and lead to further conflict/co-operation. Finally, conflicts and co-operation on one level can appear as exogenous compatibilities/incompatibilities on another level of society. This shows, that system and social integration must be perceived as interlinked and intertwined in a number of ways while the distinction is certainly to be maintained not least because it represents a crucial cognitive resource of actors themselves.

Acknowledgements

I am grateful to Bob Jessop and Will Medd for their comments and linguistic advice. Obviously, the usual disclaimer applies.

1. The recent contribution 'follows up' an article published twenty-three years earlier which offers a more detailed discussion of the distinction in the context of the theoretical debates in the 1960s and 1970s (Mouzelis 1974).

2. In his seminal article, Lockwood (1964) criticised both the overemphasis of 'normative functionalism' on the harmonious integration of societies, and the 'overreaction' of 'conflict theories' which emphasised the coercive and forced nature of social order. In the view of Lockwood, both theoretical traditions tended to neglect the system integrative side of the coin.

3. For instance, according to Lockwood (1992), 'contradictions' can lead to social change only through the mediating power of agency.

4. A 're-entry' occurs when a distinction enters one of the two sides of the distinction itself (cf. Luhmann 1994). In the current case, this works as follows: The first distinction is between structure and agency, from the perspective of an 'external' observer (the social scientist?). But in a second moment, the same distinction can also be drawn by an actor, i.e. within 'agency', so to speak. Actors can face insurmountable barriers and structural constraints, but normally they also perceive themselves as 'autonomous' and 'self-determined'. Thus, the structure-agency distinction has 're-entered' itself through its 'privileged' pole (agency).

5. An implication is the recursivity of knowledge about society: 'objective' social science accounts can result in social transformation. See Demirovic (1992:146f.) on the 'self-reference' of knowledge and, in particular, social theory and Psychopedis's (1992) remarks on dialectics; on the political power of knowledge, see Richardson (1996).

6. On the 'sociology of failure* cf. Malpas and Wickham (1995).

7. Compared with Mouzelis's (1974:397) notion of 'social process', this is a more restricted and more abstract concept.

8. Even a seemingly 'natural' process such as global warming is highly socially determined, i.e. constituted by a number of norms, rules and conventions responsible for the use of fossile energy sources. An example of the attempt to reconfigure a set of sedimented institutions (producing effects for the global climate) with the objective of shaping the gigantic process of global warming, are the very controversial UN 'climate summits'.

9. Note that Luhmann's notion of social structure - defined as a more or less stable set of generalised expectations - also produces a link between the present and the future (cf. Luhmann 1995).

10. The argument that incompatibilities have to be perceived by actors does not necessarily imply that they cannot take the actors by surprise. Consider, for instance, that institutional-technical systems (e.g. stock exchange markets, or nuclear power generation) cannot be designed in anticipation of all possible events. It is therefore very likely that they contain some type of built-in structural incoherence, possibly leading to mismatches or 'accidents' over time« However, those latent incompatibilities do not become virulent until the unanticipated events happen. Such 'normal accidents' (Perrow 1984) have their root in the efforts of certain actors to 'design' a process over time and deviate negatively from the expected course of events. In this sense, those incompatibilities are linked to the perception of the concerned actors (I am grateful to one of the anonymous referees for raising this important point).

11. On the 'reduction of complexity' cf. Luhmann (1995:25-7).

12. This involves a slight modification of Lipietz's (1985:111) usage of the notion. Lipietz uses the term 'stroboscopic' to express the fact that the preconditions for the reproduction of social practices are identical to their results. In other words, if a set of institutional regularities is to be reproduced it must mobilise the continual support of actors. Thus, the existence of these institutions must be recognised and acknowledged by the actors concerned. This is why the reproduction of such regularities is always threatened and precarious. See also the related concept of the 'doubly tendential tendency' developed by Jessop (1990).

13. Institutional modifications can be seen as politics per se, for instance, from a discourse analysis perspective where politics is defined as the subversion and reconstitution of social structure. For a theoretical conceptualisation of the primacy of politics, see Laclau (1990), Bertramsen et al (1991), Torfing (1991).

14. The analysis of conventions, routines and rules of interaction, as presented by convention theory {Véconomie des conventions), offers some fascinating insights into the trade-offs and/or mutual enforcement of efficiency and equality. See, in particular, the seminal article by Favereau (1994) showing how different theories (among others, Rawls's 'theory of justice') deal with this dichotomy between efficiency and equality (mirroring the system-social integration distinction).

15. A 'black box' can be defined as a set of regularities over time (a process) actors rely upon without knowing its precise 'internal' functioning.

16. Cf. for instance the contributions of the 'new institutionalism' in organisational studies (Powells and DiMaggio 1991). The emphasis on 'myths' as a sort of truth proliferated within big organisations (and, indeed, 'societies of organisations', Perrow) marks an interesting congruence with the type of arguments proposed here.

References

Archer, M. 1996. 'Social Integration and System Integration: Developing the Distinction'. Sociology 30:679-99.

Bertramsen, R. B., Frolund Thomsen, J. P. and Torfing, J. 1991. State, Economy and Society. London: Unwyn Hyman.

Boltanski, L. and Thévenot, L. 1991. De la Justification. Paris: Gallimard. Callon, M. 1992. 'The Dynamics of Techno-Economic Networks', in R. Coombs, P. Saviotti and V. Walsh (eds.), Technological Change and Company Strategies: Economic and Sociological Perspectives. London: Harcourt Brace Jovanovich.

Demirovic, A. 1992. 'Regulation und Hegemonie. Intellektuelle, Wissenspraktiken und Akkumulation', in A. Demirovic, H-P. Krebs and T. Sablowski (eds.), Hegemonie und Staat. Kapitalistische Regulation als Projekt und Prozess. Münster: Westfälisches Dampfboot, pp. 128-57.

Favereau, O. 1994. 'Règles, organisation et apprentisage collectif: un paradigme non standard pour trois théories hétérodoxes', in A. Orléan (ed.), Analyse économique des conventions. Paris: PUF, pp. 113-38.

Giddens, A. 1986 [1984]. The Constitution of Society: Outline of a Theory of Structuration. Cambridge: Polity.

Giddens, A. 1994 [1985]. 'Time, Space and Regionalisation', in D. Gregory and J. Urry (eds.). Social Relations and Spatial Structures. London: Macmillan, pp. 265-95. Habermas, J. 1987. The Theory of Communicative Action, Vol. 2. London: Polity. Jessop, B. 1990. State Theory: Putting the Capitalist State in Its Place. Cambridge: Polity. Jessop, B. 1995. 'The Regulation Approach, Governance and Post-Fordism: Alternative Perspectives on Economic and Political Change?' Economy and Society 24: 307-33. Jessop, B. 1996. 'Interpretive Sociology and the Dialectic of Structure and Agency'. Theory, Culture and Society 13: 119-28.

Jessop, B. 1997. 'The Governance of Complexity and the Complexity of Governance', in A. Amin and J. Hausner (eds.), Beyond Markets and Hierarchy: Third Way Approaches to Transformation. Aldershot: Edward Elgar, pp. 111-47.

Laclau, E. 1990. New Reflections on the Revolution of Our Time. London/New York: Verso.

Latour, B. 1993. We Have Never Been Modern. Cambridge, Mass.: Harvard University Press.

Law, J. (ed.) 1991 .A Sociology of Monsters: Essays on Power, Technology and Domination. London, New York: Routledge.

Lipietz, A. 1985. 'Akkumulation, Krisen und Auswege aus der Krise: Einige methodische Überlegungen zum Begriff "Regulation"'. Prokla 58: 109-37. Lipietz, A. 1996. La société en sablier: Le partage du travail contre la déchirure sociale. Paris: La Découverte.

Lockwood, D. 1964. 'Social Integration and System Integration', in G. K. Zollschan and W. Hirsch (eds.), Explorations in Social Change. London: Routledge & Kegan Paul. Lockwood, D. 1992. Solidarity and Schism: (The Problem of Disorder3 in Durkheimian and Marxist Sociology. Oxford: Oxford University Press.

Luhmann, N. 1994. 'European Rationality', in G. Robinson and J. Rundell (eds.), Rethinking Imagination: Culture and Creativity. London: Routledge. Luhmann, N. 1995. Social Systems. Stanford, Calif.: Stanford University Press. Malpas, J. and Wickham, G. 1995. 'Governance and Failure: On the Limits of Sociology'. Australian and New Zealand Journal of Sociology 31:37-51. Mouzelis, N. 1974. 'Social and System Integration: Some Reflections on a Fundamental Distinction'. Sociology 25: 395-409.

Mouzelis, N. 1995. Sociological Theory: What Went Wrong? Diagnosis and Remedies. London: Routledge.

Mouzelis, N. 1997. 'Social and System Integration: Lockwood, Habermas, Giddens'. Sociology 31:111-19.

Orléan, A. (ed.) 1994. Analyse économique des conventions. Paris: PUF.

Perrow, C. 1984. Normal Accidents: Living with High Risk Technologies. New York:

Basic Books.

Powells, W. W. and DiMaggio, P. J. (eds.) 1991. The New Institutionalism in Organisational Analysis. Chicago: Chicago University Press.

Psychopedis, K. 1992. 'Dialectical Theory: Problems of Reconstruction', in W. Bonefeld, R. Gunn and K. Psychopedis (eds.), Open Marxism. London: Pluto. Richardson, T. 1996. 'Power and Truth in Urban and Regional Policy Making'. European Planning Studies 4(3).

Scherrer, C. 1995. 'Eine diskursanalytische Kritik der Regulationstheorie'. Prokla 100: 457-82.

Torfing, J. 1991. 'A Hegemony Approach to Capitalist Regulation', in Bertramsen et al. (eds.), State, Economy and Society. London: Unwyn Hyman, pp. 35-93. Wagner, P. 1994. 'Dispute, Uncertainty and Institution in Recent French Debates'. Journal of Political Philosophy 2:270-89.

Biographical note: MARKUS PERKMANN is a Marie Curie Fellow in the Department of Sociology at Lancaster University and is currently working on a doctoral dissertation politics.

Address: Department of Sociology, Lancaster University, Lancaster LAI 4YL.